Release 0.6.8

* Add `--build-env` for build environment information.

---------

Co-authored-by: Christoffer Lerno <christoffer@aegik.com>

.editorconfig

@@ -0,0 +1,25 @@
+# EditorConfig is awesome: https://editorconfig.org
+
+root = true
+
+[*]
+charset = utf-8
+end_of_line = lf
+
+[CMakeLists.txt]
+indent_style = space
+indent_size = 4
+
+[*.{c,cc,h}]
+indent_style = tab
+
+[*.{c3}]
+indent_style = tab
+
+[*.{json,toml,yml,gyp}]
+indent_style = space
+indent_size = 2
+
+[*.{py,pyi}]
+indent_style = tab
+

.gitattributes

@@ -1,4 +1,2 @@
 $ cat .gitattributes
 * text=auto
-*.c3 linguist-language=C
-*.c3t linguist-language=C

.github/FUNDING.yml

@@ -0,0 +1,14 @@
+# These are supported funding model platforms
+
+github: [c3lang]
+patreon: # Replace with a single Patreon username
+open_collective: # Replace with a single Open Collective username
+ko_fi: c3lang
+tidelift: # Replace with a single Tidelift platform-name/package-name e.g., npm/babel
+community_bridge: # Replace with a single Community Bridge project-name e.g., cloud-foundry
+liberapay: # Replace with a single Liberapay username
+issuehunt: # Replace with a single IssueHunt username
+otechie: # Replace with a single Otechie username
+lfx_crowdfunding: # Replace with a single LFX Crowdfunding project-name e.g., cloud-foundry
+polar: # Replace with a single Polar username
+custom: # Replace with up to 4 custom sponsorship URLs e.g., ['link1', 'link2']

.github/workflows/main.yml

@@ -2,13 +2,117 @@ name: CI
 
 on:
   push:
-    branches: [ master, dev, ci_testing ]
+    branches: [ master, dev, ci_testing, experiments ]
   pull_request:
-    branches: [ master ]
+    branches: [ master, dev ]
 
+env:
+  LLVM_RELEASE_VERSION_WINDOWS: 18
+  LLVM_RELEASE_VERSION_MAC: 17
+  LLVM_RELEASE_VERSION_LINUX: 17
+  LLVM_RELEASE_VERSION_UBUNTU20: 17
+  LLVM_DEV_VERSION: 21
 jobs:
+
+  build-msvc:
+    runs-on: windows-latest
+    strategy:
+      # Don't abort runners if a single one fails
+      fail-fast: false
+      matrix:
+        build_type: [ Release, Debug ]
+
+    defaults:
+      run:
+        shell: cmd
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: CMake
+        run: |
+          cmake -B build -G "Visual Studio 17 2022" -A x64 -DCMAKE_BUILD_TYPE=${{ matrix.build_type }}
+          cmake --build build --config ${{ matrix.build_type }}
+
+      - name: Compile and run some examples
+        run: |
+          cd resources
+          ..\build\${{ matrix.build_type }}\c3c.exe compile-run -L C:\ --print-linking examples\hello_world_many.c3
+          ..\build\${{ matrix.build_type }}\c3c.exe compile-run --print-linking examples\time.c3
+          ..\build\${{ matrix.build_type }}\c3c.exe compile-run --print-linking examples\fannkuch-redux.c3
+          ..\build\${{ matrix.build_type }}\c3c.exe compile-run examples\contextfree\boolerr.c3
+          ..\build\${{ matrix.build_type }}\c3c.exe compile-run examples\ls.c3
+          ..\build\${{ matrix.build_type }}\c3c.exe compile-run examples\load_world.c3
+          ..\build\${{ matrix.build_type }}\c3c.exe compile-run examples\process.c3
+          ..\build\${{ matrix.build_type }}\c3c.exe compile-run examples\args.c3 -- foo -bar "baz baz"
+          ..\build\${{ matrix.build_type }}\c3c.exe compile --no-entry --test -g -O0 --threads 1 --target macos-x64 examples\constants.c3
+          ..\build\${{ matrix.build_type }}\c3c.exe compile-run msvc_stack.c3
+
+      - name: Build testproject
+        run: |
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvarsall.bat" x64
+          cd resources/testproject
+          ..\..\build\${{ matrix.build_type }}\c3c.exe -vvv --emit-llvm run hello_world_win32 --trust=full
+          dir build\llvm\windows-x64
+          ..\..\build\${{ matrix.build_type }}\c3c.exe clean
+          dir build\llvm\windows-x64
+
+
+      - name: Build testproject lib
+        run: |
+          cd resources/testproject
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvarsall.bat" x64
+          ..\..\build\${{ matrix.build_type }}\c3c.exe -vvv build hello_world_win32_lib --trust=full
+
+      - name: Compile and run dynlib-test
+        run: |
+          cd resources/examples/dynlib-test
+          ..\..\..\build\${{ matrix.build_type }}\c3c.exe -vv dynamic-lib add.c3
+          ..\..\..\build\${{ matrix.build_type }}\c3c.exe -vv compile-run test.c3 -l ./add.lib          
+
+      - name: Compile and run staticlib-test
+        run: |
+          cd resources/examples/staticlib-test
+          ..\..\..\build\${{ matrix.build_type }}\c3c.exe -vv static-lib add.c3
+          ..\..\..\build\${{ matrix.build_type }}\c3c.exe -vv compile-run test.c3 -l ./add.lib          
+
+      - name: Vendor-fetch
+        run: |
+          build\${{ matrix.build_type }}\c3c.exe vendor-fetch raylib5
+
+      - name: Try raylib5
+        run: |
+          cd resources
+          ..\build\${{ matrix.build_type }}\c3c.exe vendor-fetch raylib5
+          ..\build\${{ matrix.build_type }}\c3c.exe compile --lib raylib5 --print-linking examples\raylib\raylib_arkanoid.c3
+          ..\build\${{ matrix.build_type }}\c3c.exe compile --lib raylib5 --print-linking examples\raylib\raylib_snake.c3
+          ..\build\${{ matrix.build_type }}\c3c.exe compile --lib raylib5 --print-linking examples\raylib\raylib_tetris.c3
+
+      - name: Compile run unit tests
+        run: |
+          cd test
+          ..\build\${{ matrix.build_type }}\c3c.exe compile-test unit -O1
+
+      - name: run compiler tests
+        run: |
+          cd test
+          ..\build\${{ matrix.build_type }}\c3c.exe compile-run -O1 src/test_suite_runner.c3 --stdlib ..\lib7 --enable-new-generics -- ..\build\${{ matrix.build_type }}\c3c.exe test_suite7/ --stdlib ../lib7 --no-terminal
+
+      - name: Test python script
+        run: |
+          py msvc_build_libraries.py --accept-license
+          dir msvc_sdk
+
+      - name: upload artifacts
+        uses: actions/upload-artifact@v4
+        with:
+          name: c3-windows-${{ matrix.build_type }}
+          path: |
+            build\${{ matrix.build_type }}\c3c.exe
+            build\${{ matrix.build_type }}\c3c_rt
+
   build-msys2-mingw:
     runs-on: windows-latest
+    # if: ${{ false }}
     strategy:
       # Don't abort runners if a single one fails
       fail-fast: false
@@ -19,29 +123,53 @@ jobs:
       run:
         shell: msys2 {0}
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
 
       - uses: msys2/setup-msys2@v2
         with:
           msystem: MINGW64
           update: true
-          install: git binutils mingw-w64-x86_64-mlir mingw-w64-x86_64-cmake mingw-w64-x86_64-toolchain mingw-w64-x86_64-llvm mingw-w64-x86_64-polly mingw-w64-x86_64-python mingw-w64-x86_64-lld
-
+          install: git binutils mingw-w64-x86_64-clang mingw-w64-x86_64-ninja mingw-w64-x86_64-cmake mingw-w64-x86_64-toolchain mingw-w64-x86_64-python
+      - shell: msys2 {0}
+        run: |
+            pacman --noconfirm -U https://mirror.msys2.org/mingw/mingw64/mingw-w64-x86_64-llvm-19.1.7-1-any.pkg.tar.zst
+            pacman --noconfirm -U https://mirror.msys2.org/mingw/mingw64/mingw-w64-x86_64-lld-19.1.7-1-any.pkg.tar.zst
       - name: CMake
         run: |
-          mkdir build && cd build
-          cmake .. -G "MinGW Makefiles" -DCMAKE_BUILD_TYPE=${{ matrix.build_type }}
-          cmake --build .
+          cmake -B build -G Ninja -DCMAKE_C_COMPILER=clang -DCMAKE_BUILD_TYPE=${{ matrix.build_type }}
+          cmake --build build
+
+      - name: Compile and run some examples
+        run: |
+          cd resources
+          ../build/c3c compile-run --print-linking examples/hello_world_many.c3
+          ../build/c3c compile-run --print-linking examples/process.c3
+          ../build/c3c compile-run --print-linking examples/time.c3
+          ../build/c3c compile-run --print-linking examples/fannkuch-redux.c3
+          ../build/c3c compile-run --print-linking examples/contextfree/boolerr.c3
+          ../build/c3c compile-run --print-linking examples/load_world.c3
+          ../build/c3c compile-run --print-linking examples/args.c3 -- foo -bar "baz baz"
+          ../build/c3c compile --no-entry --test -g -O0 --threads 1 --target macos-x64 examples/constants.c3
 
       - name: Build testproject
         run: |
           cd resources/testproject
-          ../../build/c3c build --debug-log
+          ../../build/c3c run -vvv --trust=full
+
+      - name: Vendor-fetch
+        run: |
+          ./build/c3c vendor-fetch raylib5
+
+      - name: Build testproject lib
+        run: |
+          cd resources/testproject
+          ../../build/c3c build hello_world_lib --cc cc -vvv --trust=full
 
       - name: run compiler tests
         run: |
           cd test
-          python3 src/tester.py ../build/c3c.exe test_suite/
+          ../build/c3c.exe compile --target windows-x64 -O1 src/test_suite_runner.c3 --stdlib ../lib7 --enable-new-generics
+          ./test_suite_runner.exe ../build/c3c.exe test_suite7/ --stdlib ../lib7 --no-terminal
 
   build-msys2-clang:
     runs-on: windows-latest
@@ -56,7 +184,7 @@ jobs:
       run:
         shell: msys2 {0}
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
 
       - uses: msys2/setup-msys2@v2
         with:
@@ -66,64 +194,419 @@ jobs:
 
       - name: CMake
         run: |
-          mkdir build && cd build
-          cmake .. -G "MinGW Makefiles" -DCMAKE_BUILD_TYPE=${{ matrix.build_type }}
-          cmake --build .
+          cmake -B build -G "MinGW Makefiles" -DCMAKE_BUILD_TYPE=${{ matrix.build_type }}
+          cmake --build build
+
+      - name: Compile and run some examples
+        run: |
+          cd resources
+          ../build/c3c compile-run examples/hello_world_many.c3
+          ../build/c3c compile-run examples/time.c3
+          ../build/c3c compile-run examples/fannkuch-redux.c3
+          ../build/c3c compile-run examples/contextfree/boolerr.c3
+          ../build/c3c compile-run examples/load_world.c3
+          ../build/c3c compile-run examples/args.c3 -- foo -bar "baz baz"
+          ../build/c3c compile --no-entry --test -g -O0 --threads 1 --target macos-x64 examples/constants.c3
       - name: Build testproject
         run: |
           cd resources/testproject
-          ../../build/c3c build --debug-log
+          ../../build/c3c run -vvv --trust=full
+
+      - name: Build testproject lib
+        run: |
+          cd resources/testproject
+          ../../build/c3c build hello_world_lib -vvv --trust=full
 
       - name: run compiler tests
         run: |
           cd test
-          python3 src/tester.py ../build/c3c.exe test_suite/
+          ../build/c3c.exe compile-run -O1 --stdlib ../lib7 src/test_suite_runner.c3 --enable-new-generics -- ../build/c3c.exe test_suite7/ --stdlib ../lib7 --no-terminal
 
   build-linux:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-22.04
     strategy:
       # Don't abort runners if a single one fails
       fail-fast: false
       matrix:
         build_type: [Release, Debug]
-        llvm_version: [12, 13, 14, 15]
+        llvm_version: [17, 18, 19, 20, 21]
 
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - name: Install common deps
         run: |
-          sudo apt-get install zlib1g zlib1g-dev python3
+          sudo apt-get install zlib1g zlib1g-dev python3 ninja-build curl
 
-      - name: Install Clang ${{ matrix.llvm_version }}
+      - name: Install Clang ${{matrix.llvm_version}}
         run: |
           wget -O - https://apt.llvm.org/llvm-snapshot.gpg.key | sudo apt-key add -
-          if [[ "${{matrix.llvm_version}}" < 15 ]]; then
+          if [[ "${{matrix.llvm_version}}" < 18 ]]; then
+            sudo add-apt-repository "deb http://apt.llvm.org/focal/ llvm-toolchain-focal-${{matrix.llvm_version}} main"
+            sudo apt-get update
+            sudo apt-get install -y -t llvm-toolchain-focal-${{matrix.llvm_version}} libpolly-${{matrix.llvm_version}}-dev \
+              clang-${{matrix.llvm_version}} llvm-${{matrix.llvm_version}} llvm-${{matrix.llvm_version}}-dev \
+              lld-${{matrix.llvm_version}} liblld-${{matrix.llvm_version}}-dev libmlir-${{matrix.llvm_version}} \
+              libmlir-${{matrix.llvm_version}}-dev mlir-${{matrix.llvm_version}}-tools
+          else
+            if [[ "${{matrix.llvm_version}}" < "${{env.LLVM_DEV_VERSION}}" ]]; then
+              sudo add-apt-repository "deb http://apt.llvm.org/focal/ llvm-toolchain-focal-${{matrix.llvm_version}} main"
+              sudo apt-get update
+              sudo apt-get install -y -t llvm-toolchain-focal-${{matrix.llvm_version}} libpolly-${{matrix.llvm_version}}-dev \
+                clang-${{matrix.llvm_version}} llvm-${{matrix.llvm_version}} llvm-${{matrix.llvm_version}}-dev \
+                lld-${{matrix.llvm_version}} liblld-${{matrix.llvm_version}}-dev
+            else
+              sudo add-apt-repository "deb http://apt.llvm.org/focal/ llvm-toolchain-focal main"
+              sudo apt-get install -y -t llvm-toolchain-focal libpolly-${{matrix.llvm_version}}-dev \
+                clang-${{matrix.llvm_version}} llvm-${{matrix.llvm_version}} llvm-${{matrix.llvm_version}}-dev \
+                lld-${{matrix.llvm_version}} liblld-${{matrix.llvm_version}}-dev
+            fi
+          fi
+      - name: CMake
+        if: matrix.llvm_version < 18 || matrix.llvm_version == env.LLVM_DEV_VERSION
+        run: |
+          cmake -B build \
+                -G Ninja \
+                -DCMAKE_BUILD_TYPE=${{matrix.build_type}} \
+                -DCMAKE_C_COMPILER=clang-${{matrix.llvm_version}} \
+                -DCMAKE_CXX_COMPILER=clang++-${{matrix.llvm_version}} \
+                -DCMAKE_LINKER=lld-link-${{matrix.llvm_version}} \
+                -DCMAKE_OBJCOPY=llvm-objcopy-${{matrix.llvm_version}} \
+                -DCMAKE_STRIP=llvm-strip-${{matrix.llvm_version}} \
+                -DCMAKE_DLLTOOL=llvm-dlltool-${{matrix.llvm_version}} \
+                -DC3_LLVM_VERSION=${{matrix.llvm_version}}
+          cmake --build build
+      - name: CMake18
+        if: matrix.llvm_version >= 18 && matrix.llvm_version != env.LLVM_DEV_VERSION
+        run: |
+          cmake -B build \
+                -G Ninja \
+                -DCMAKE_BUILD_TYPE=${{matrix.build_type}} \
+                -DCMAKE_C_COMPILER=clang-${{matrix.llvm_version}} \
+                -DCMAKE_CXX_COMPILER=clang++-${{matrix.llvm_version}} \
+                -DCMAKE_LINKER=lld-link-${{matrix.llvm_version}} \
+                -DCMAKE_OBJCOPY=llvm-objcopy-${{matrix.llvm_version}} \
+                -DCMAKE_STRIP=llvm-strip-${{matrix.llvm_version}} \
+                -DCMAKE_DLLTOOL=llvm-dlltool-${{matrix.llvm_version}} \
+                -DC3_LLVM_VERSION=${{matrix.llvm_version}}.1
+          cmake --build build
+
+      - name: Compile and run some examples
+        run: |
+          cd resources
+          ../build/c3c compile examples/base64.c3
+          ../build/c3c compile examples/binarydigits.c3
+          ../build/c3c compile examples/brainfk.c3
+          ../build/c3c compile examples/factorial_macro.c3
+          ../build/c3c compile examples/fasta.c3
+          ../build/c3c compile examples/gameoflife.c3
+          ../build/c3c compile examples/hash.c3
+          ../build/c3c compile-only examples/levenshtein.c3
+          ../build/c3c compile examples/load_world.c3
+          ../build/c3c compile-only examples/map.c3
+          ../build/c3c compile examples/mandelbrot.c3
+          ../build/c3c compile examples/plus_minus.c3
+          ../build/c3c compile examples/nbodies.c3
+          ../build/c3c compile examples/spectralnorm.c3
+          ../build/c3c compile examples/swap.c3
+          ../build/c3c compile examples/contextfree/boolerr.c3
+          ../build/c3c compile examples/contextfree/dynscope.c3
+          ../build/c3c compile examples/contextfree/guess_number.c3
+          ../build/c3c compile examples/contextfree/multi.c3
+          ../build/c3c compile examples/contextfree/cleanup.c3
+          ../build/c3c compile-run examples/hello_world_many.c3
+          ../build/c3c compile-run examples/time.c3
+          ../build/c3c compile-run examples/fannkuch-redux.c3
+          ../build/c3c compile-run examples/contextfree/boolerr.c3
+          ../build/c3c compile-run examples/load_world.c3
+          ../build/c3c compile-run examples/process.c3
+          ../build/c3c compile-run examples/ls.c3
+          ../build/c3c compile-run --linker=builtin linux_stack.c3
+          ../build/c3c compile-run linux_stack.c3
+          ../build/c3c compile-run examples/args.c3 -- foo -bar "baz baz"
+
+      - name: Compile and run dynlib-test
+        run: |
+          cd resources/examples/dynlib-test
+          ../../../build/c3c -vv dynamic-lib add.c3
+          mv add.so libadd.so
+          cc test.c -L. -ladd -Wl,-rpath=.         
+          ./a.out
+          ../../../build/c3c compile-run test.c3 -L . -l add -z -Wl,-rpath=.           
+
+      - name: Compile and run staticlib-test
+        run: |
+          cd resources/examples/staticlib-test
+          ../../../build/c3c -vv static-lib add.c3
+          mv add.a libadd.a
+          cc test.c -L. -ladd         
+          ./a.out
+          ../../../build/c3c compile-run test.c3 -L . -l add           
+
+      - name: Compile run unit tests
+        run: |
+          cd test
+          ../build/c3c compile-test unit
+
+      - name: Build testproject
+        run: |
+          cd resources/testproject
+          ../../build/c3c run -vvv --trust=full
+
+      - name: Test WASM
+        run: |
+          cd resources/testfragments
+          ../../build/c3c compile --target wasm32 -g0 --no-entry -Os wasm4.c3
+
+      - name: Install QEMU and Risc-V toolchain
+        run: |
+          sudo apt-get install opensbi qemu-system-misc u-boot-qemu gcc-riscv64-unknown-elf
+
+      - name: Compile and run Baremetal Risc-V Example
+        run: |
+          cd resources/examples/embedded/riscv-qemu
+          make C3C_PATH=../../../../build/ run
+
+      - name: Build testproject direct linker
+        run: |
+          cd resources/testproject
+          ../../build/c3c run -vvv --linker=builtin --trust=full
+
+      - name: Init a library & a project
+        run: |
+          ./build/c3c init-lib mylib
+          ls mylib.c3l
+          ./build/c3c init myproject
+          ls myproject
+
+      - name: run compiler tests
+        run: |
+          cd test
+          ../build/c3c compile-run -O1 src/test_suite_runner.c3 --stdlib ../lib7 --enable-new-generics -- ../build/c3c  test_suite7/ --stdlib ../lib7
+
+      - name: bundle_output
+        if: matrix.llvm_version == env.LLVM_RELEASE_VERSION_LINUX
+        run: |
+          mkdir c3
+          cp -r lib c3
+          cp msvc_build_libraries.py c3
+          cp build/c3c c3
+          tar czf c3-linux-${{matrix.build_type}}.tar.gz c3
+
+      - name: upload artifacts
+        if: matrix.llvm_version == env.LLVM_RELEASE_VERSION_LINUX
+        uses: actions/upload-artifact@v4
+        with:
+          name: c3-linux-${{matrix.build_type}}
+          path: c3-linux-${{matrix.build_type}}.tar.gz
+
+  build-linux-ubuntu20:
+    runs-on: ubuntu-20.04
+    strategy:
+      # Don't abort runners if a single one fails
+      fail-fast: false
+      matrix:
+        build_type: [Release, Debug]
+        llvm_version: [17, 18, 19, 20, 21]
+    steps:
+      - uses: actions/checkout@v4
+      - name: Install common deps
+        run: |
+          sudo apt-get install zlib1g zlib1g-dev python3 ninja-build curl
+
+      - name: Install Clang ${{matrix.llvm_version}}
+        run: |
+          wget -O - https://apt.llvm.org/llvm-snapshot.gpg.key | sudo apt-key add -
+          if [[ "${{matrix.llvm_version}}" < "${{env.LLVM_DEV_VERSION}}" ]]; then
             sudo add-apt-repository "deb http://apt.llvm.org/focal/ llvm-toolchain-focal-${{matrix.llvm_version}} main"
           else
             sudo add-apt-repository "deb http://apt.llvm.org/focal/ llvm-toolchain-focal main"
           fi
           sudo apt-get update
           sudo apt-get install -y clang-${{matrix.llvm_version}} llvm-${{matrix.llvm_version}} llvm-${{matrix.llvm_version}}-dev lld-${{matrix.llvm_version}} liblld-${{matrix.llvm_version}}-dev
-
-          if [[ "${{matrix.llvm_version}}" > 12 ]]; then
-            sudo apt-get install -y libmlir-${{matrix.llvm_version}} libmlir-${{matrix.llvm_version}}-dev mlir-${{matrix.llvm_version}}-tools
-          fi
-
-      - name: CMake
+          sudo apt-get install -y libmlir-${{matrix.llvm_version}} libmlir-${{matrix.llvm_version}}-dev mlir-${{matrix.llvm_version}}-tools
+          sudo apt-get install -y libpolly-${{matrix.llvm_version}}-dev
+      - name: CMake Old
+        if: matrix.llvm_version < 18 || matrix.llvm_version == env.LLVM_DEV_VERSION
         run: |
-          mkdir build && cd build
-          cmake .. -DCMAKE_C_COMPILER=clang-${{matrix.llvm_version}} -DCMAKE_CXX_COMPILER=clang++-${{matrix.llvm_version}} -DC3_LLVM_VERSION=${{matrix.llvm_version}} -DCMAKE_BUILD_TYPE=${{ matrix.build_type }}
-          cmake --build .
+          cmake -B build \
+                -G Ninja \
+                -DCMAKE_BUILD_TYPE=${{matrix.build_type}} \
+                -DCMAKE_C_COMPILER=clang-${{matrix.llvm_version}} \
+                -DCMAKE_CXX_COMPILER=clang++-${{matrix.llvm_version}} \
+                -DCMAKE_LINKER=lld-link-${{matrix.llvm_version}} \
+                -DCMAKE_OBJCOPY=llvm-objcopy-${{matrix.llvm_version}} \
+                -DCMAKE_STRIP=llvm-strip-${{matrix.llvm_version}} \
+                -DCMAKE_DLLTOOL=llvm-dlltool-${{matrix.llvm_version}} \
+                -DC3_LLVM_VERSION=${{matrix.llvm_version}}
+          cmake --build build
+      - name: CMake
+        if: matrix.llvm_version >= 18 && matrix.llvm_version != env.LLVM_DEV_VERSION
+        run: |
+          cmake -B build \
+                -G Ninja \
+                -DCMAKE_BUILD_TYPE=${{matrix.build_type}} \
+                -DCMAKE_C_COMPILER=clang-${{matrix.llvm_version}} \
+                -DCMAKE_CXX_COMPILER=clang++-${{matrix.llvm_version}} \
+                -DCMAKE_LINKER=lld-link-${{matrix.llvm_version}} \
+                -DCMAKE_OBJCOPY=llvm-objcopy-${{matrix.llvm_version}} \
+                -DCMAKE_STRIP=llvm-strip-${{matrix.llvm_version}} \
+                -DCMAKE_DLLTOOL=llvm-dlltool-${{matrix.llvm_version}} \
+                -DC3_LLVM_VERSION=${{matrix.llvm_version}}.1
+          cmake --build build
+      - name: Compile and run some examples
+        run: |
+          cd resources
+          ../build/c3c compile examples/gameoflife.c3
+          ../build/c3c compile-only examples/levenshtein.c3
+          ../build/c3c compile-only examples/map.c3
+          ../build/c3c compile examples/mandelbrot.c3
+          ../build/c3c compile examples/plus_minus.c3
+          ../build/c3c compile examples/spectralnorm.c3
+          ../build/c3c compile examples/swap.c3
+          ../build/c3c compile examples/contextfree/guess_number.c3
+          ../build/c3c compile-run examples/hash.c3
+          ../build/c3c compile-run examples/nbodies.c3
+          ../build/c3c compile-run examples/contextfree/boolerr.c3
+          ../build/c3c compile-run examples/contextfree/dynscope.c3
+          ../build/c3c compile-run examples/contextfree/multi.c3
+          ../build/c3c compile-run examples/contextfree/cleanup.c3
+          ../build/c3c compile-run examples/hello_world_many.c3
+          ../build/c3c compile-run examples/time.c3
+          ../build/c3c compile-run examples/fannkuch-redux.c3
+          ../build/c3c compile-run examples/load_world.c3
+          ../build/c3c compile-run examples/base64.c3
+          ../build/c3c compile-run examples/binarydigits.c3
+          ../build/c3c compile-run examples/brainfk.c3
+          ../build/c3c compile-run examples/factorial_macro.c3
+          ../build/c3c compile-run examples/fasta.c3
+          ../build/c3c compile-run examples/process.c3
+          ../build/c3c compile-run --linker=builtin linux_stack.c3
+          ../build/c3c compile-run linux_stack.c3
+          ../build/c3c compile-run examples/args.c3 -- foo -bar "baz baz"
+
+      - name: Compile run unit tests
+        run: |
+          cd test
+          ../build/c3c compile-test unit --sanitize=address
 
       - name: Build testproject
         run: |
           cd resources/testproject
-          ../../build/c3c build --debug-log
+          ../../build/c3c run -vvv --trust=full
+
+      - name: Build testproject direct linker
+        run: |
+          cd resources/testproject
+          ../../build/c3c run -vvv --linker=builtin --trust=full
 
       - name: run compiler tests
         run: |
           cd test
-          python3 src/tester.py ../build/c3c test_suite/
+          ../build/c3c compile-run -O1 src/test_suite_runner.c3 --stdlib ../lib7 --enable-new-generics -- ../build/c3c  test_suite7/ --stdlib ../lib7
+
+      - name: bundle_output
+        if: matrix.llvm_version == env.LLVM_RELEASE_VERSION_UBUNTU20
+        run: |
+          mkdir c3
+          cp -r lib c3
+          cp msvc_build_libraries.py c3
+          cp build/c3c c3
+          tar czf c3-ubuntu-20-${{matrix.build_type}}.tar.gz c3
+
+      - name: upload artifacts
+        if: matrix.llvm_version == env.LLVM_RELEASE_VERSION_UBUNTU20
+        uses: actions/upload-artifact@v4
+        with:
+          name: c3-ubuntu-20-${{matrix.build_type}}
+          path: c3-ubuntu-20-${{matrix.build_type}}.tar.gz
+  
+  build-with-docker:
+    runs-on: ubuntu-22.04
+    strategy:
+      fail-fast: false
+      matrix:
+        ubuntu_version: [20.04, 22.04]
+        build_type: [Release, Debug]
+        llvm_version: [17, 18, 19]
+
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Setup Docker Buildx
+        uses: docker/setup-buildx-action@v3
+
+      - name: Make script executable
+        run: chmod +x ./build-with-docker.sh
+
+      - name: Run build
+        run: |
+          LLVM_VERSION=${{ matrix.llvm_version }} UBUNTU_VERSION=${{ matrix.ubuntu_version }} CMAKE_BUILD_TYPE=${{ matrix.build_type }} ./build-with-docker.sh
+      
+      - name: Compile and run some examples
+        run: |
+          cd resources
+          ../build/c3c compile examples/base64.c3
+          ../build/c3c compile examples/binarydigits.c3
+          ../build/c3c compile examples/brainfk.c3
+          ../build/c3c compile examples/factorial_macro.c3
+          ../build/c3c compile examples/fasta.c3
+          ../build/c3c compile examples/gameoflife.c3
+          ../build/c3c compile examples/hash.c3
+          ../build/c3c compile-only examples/levenshtein.c3
+          ../build/c3c compile examples/load_world.c3
+          ../build/c3c compile-only examples/map.c3
+          ../build/c3c compile examples/mandelbrot.c3
+          ../build/c3c compile examples/plus_minus.c3
+          ../build/c3c compile examples/nbodies.c3
+          ../build/c3c compile examples/spectralnorm.c3
+          ../build/c3c compile examples/swap.c3
+          ../build/c3c compile examples/contextfree/boolerr.c3
+          ../build/c3c compile examples/contextfree/dynscope.c3
+          ../build/c3c compile examples/contextfree/guess_number.c3
+          ../build/c3c compile examples/contextfree/multi.c3
+          ../build/c3c compile examples/contextfree/cleanup.c3
+          ../build/c3c compile-run examples/hello_world_many.c3
+          ../build/c3c compile-run examples/time.c3
+          ../build/c3c compile-run examples/fannkuch-redux.c3
+          ../build/c3c compile-run examples/contextfree/boolerr.c3
+          ../build/c3c compile-run examples/load_world.c3
+          ../build/c3c compile-run examples/process.c3
+          ../build/c3c compile-run examples/ls.c3
+          ../build/c3c compile-run --linker=builtin linux_stack.c3
+          ../build/c3c compile-run linux_stack.c3
+          ../build/c3c compile-run examples/args.c3 -- foo -bar "baz baz"
+
+      - name: Compile run unit tests
+        run: |
+          cd test
+          ../build/c3c compile-test unit
+
+      - name: Build testproject
+        run: |
+          cd resources/testproject
+          ../../build/c3c run -vvv --trust=full
+
+      - name: Test WASM
+        run: |
+          cd resources/testfragments
+          ../../build/c3c compile --reloc=none --target wasm32 -g0 --no-entry -Os wasm4.c3
+
+      - name: Build testproject direct linker
+        run: |
+          cd resources/testproject
+          ../../build/c3c run -vvv --linker=builtin --trust=full
+
+      - name: Init a library & a project
+        run: |
+          ./build/c3c init-lib mylib
+          ls mylib.c3l
+          ./build/c3c init myproject
+          ls myproject
+
+      - name: run compiler tests
+        run: |
+          cd test
+          ../build/c3c compile-run -O1 src/test_suite_runner.c3 --stdlib ../lib7 --enable-new-generics -- ../build/c3c  test_suite7/ --stdlib ../lib7
 
   build-mac:
     runs-on: macos-latest
@@ -132,28 +615,190 @@ jobs:
       fail-fast: false
       matrix:
         build_type: [Release, Debug]
-        llvm_version: [12]
+        llvm_version: [17, 18]
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - name: Download LLVM
         run: |
-          brew install llvm@${{ matrix.llvm_version }} botan
-          echo "/usr/local/opt/llvm@${{ matrix.llvm_version }}/bin" >> $GITHUB_PATH
+          brew install llvm@${{ matrix.llvm_version }} ninja curl
+          echo "/opt/homebrew/opt/llvm@${{ matrix.llvm_version }}/bin" >> $GITHUB_PATH
           TMP_PATH=$(xcrun --show-sdk-path)/user/include
           echo "CPATH=$TMP_PATH" >> $GITHUB_ENV
 
       - name: CMake
+        if: matrix.llvm_version < 18
         run: |
-          mkdir build && cd build
-          cmake .. -DC3_LLVM_VERSION=${{matrix.llvm_version}} -DCMAKE_BUILD_TYPE=${{ matrix.build_type }}
-          cmake --build .
+          cmake -B build -G Ninja -DC3_LLVM_VERSION=${{matrix.llvm_version}} -DCMAKE_BUILD_TYPE=${{ matrix.build_type }}
+          cmake --build build
+      - name: CMake18
+        if: matrix.llvm_version >= 18
+        run: |
+          cmake -B build -G Ninja -DC3_LLVM_VERSION=${{matrix.llvm_version}}.1 -DCMAKE_BUILD_TYPE=${{ matrix.build_type }}
+          cmake --build build
+
+      - name: Vendor-fetch
+        run: |
+          ./build/c3c vendor-fetch raylib5
+
+      - name: Compile and run some examples
+        run: |
+          cd resources
+          ../build/c3c compile-run examples/hello_world_many.c3
+          ../build/c3c compile-run examples/time.c3
+          ../build/c3c compile-run examples/fannkuch-redux.c3
+          ../build/c3c compile-run examples/contextfree/boolerr.c3
+          ../build/c3c compile-run examples/process.c3
+          ../build/c3c compile-run examples/load_world.c3
+          ../build/c3c compile-run -O5 examples/load_world.c3
+          ../build/c3c compile-run examples/args.c3 -- foo -bar "baz baz"
+
+      - name: Compile and run dynlib-test
+        run: |
+          cd resources/examples/dynlib-test
+          ../../../build/c3c -vv dynamic-lib add.c3
+          ../../../build/c3c compile-run test.c3 -l ./add.dylib          
+
+      - name: Compile run unit tests
+        run: |
+          cd test
+          ../build/c3c compile-test unit -O1
+
+      - name: Test WASM
+        run: |
+          cd resources/testfragments
+          ../../build/c3c compile --target wasm32 -g0 --no-entry -Os wasm4.c3
 
       - name: Build testproject
         run: |
           cd resources/testproject
-          ../../build/c3c build --debug-log
+          ../../build/c3c run -vvv --trust=full
+
+      - name: Build testproject direct linker
+        run: |
+          cd resources/testproject
+          ../../build/c3c run -vvv --linker=builtin --trust=full
+
+      - name: Build testproject lib
+        run: |
+          cd resources/testproject
+          ../../build/c3c build hello_world_lib -vvv --trust=full
 
       - name: run compiler tests
         run: |
           cd test
-          python3 src/tester.py ../build/c3c test_suite/
+          ../build/c3c compile -O1 src/test_suite_runner.c3 --stdlib ../lib7 --enable-new-generics
+          ./test_suite_runner ../build/c3c test_suite7/ --stdlib ../lib7
+
+      - name: run build test suite runner
+        run: |
+          cd test
+          ../build/c3c compile -O1 src/test_suite_runner.c3 --stdlib ../lib7 --enable-new-generics
+
+      - name: bundle_output
+        if: matrix.llvm_version == env.LLVM_RELEASE_VERSION_MAC
+        run: |
+          mkdir macos
+          cp -r lib macos
+          cp msvc_build_libraries.py macos
+          cp build/c3c macos
+          zip -r c3-macos-${{matrix.build_type}}.zip macos
+
+      - name: upload artifacts
+        if: matrix.llvm_version == env.LLVM_RELEASE_VERSION_MAC
+        uses: actions/upload-artifact@v4
+        with:
+          name: c3-macos-${{matrix.build_type}}
+          path: c3-macos-${{matrix.build_type}}.zip
+
+  build-nix:
+    runs-on: ubuntu-22.04
+    strategy:
+      # Don't abort runners if a single one fails
+      fail-fast: false
+      matrix:
+        build_type: [ Release, Debug ]
+        nixpkgs: [ Lock, Latest ]
+    steps:
+    - uses: actions/checkout@v4
+
+    - name: Install Nix
+      uses: cachix/install-nix-action@v30
+      with: 
+        github_access_token: ${{ secrets.GITHUB_TOKEN }}
+
+    - name: Update flake (if necessary)
+      run: |
+        if [[ matrix.nixpkgs == "Latest" ]]; then 
+          nix flake update
+        fi
+        nix flake info
+
+    - name: Build and check 
+      run: |
+        if [[ ${{ matrix.build_type }} = "Debug" ]]; then 
+          nix build -L ".#c3c-debug-checks" 
+        else 
+          nix build -L ".#c3c-checks"
+        fi
+
+  release:
+    runs-on: ubuntu-22.04
+    needs: [build-msvc, build-linux, build-mac, build-linux-ubuntu20]
+    if: github.ref == 'refs/heads/master'
+
+    steps:
+      - uses: actions/checkout@v4
+      - name: delete tag
+        continue-on-error: true
+        uses: actions/github-script@v6
+        with:
+          script: |
+            github.rest.git.deleteRef({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              ref: 'tags/latest',
+              sha: context.sha
+            })
+      - name: create tag
+        uses: actions/github-script@v6
+        with:
+          script: |
+            github.rest.git.createRef({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              ref: 'refs/tags/latest',
+              sha: context.sha
+            })
+
+      - uses: actions/download-artifact@v4
+      - run: cp -r lib c3-windows-Release
+      - run: cp -r lib c3-windows-Debug
+      - run: cp msvc_build_libraries.py c3-windows-Release
+      - run: cp msvc_build_libraries.py c3-windows-Debug
+      - run: zip -r c3-windows.zip c3-windows-Release
+      - run: zip -r c3-windows-debug.zip c3-windows-Debug
+      - run: mv c3-linux-Release/c3-linux-Release.tar.gz c3-linux-Release/c3-linux.tar.gz
+      - run: mv c3-linux-Debug/c3-linux-Debug.tar.gz c3-linux-Debug/c3-linux-debug.tar.gz
+      - run: mv c3-ubuntu-20-Release/c3-ubuntu-20-Release.tar.gz c3-ubuntu-20-Release/c3-ubuntu-20.tar.gz
+      - run: mv c3-ubuntu-20-Debug/c3-ubuntu-20-Debug.tar.gz c3-ubuntu-20-Debug/c3-ubuntu-20-debug.tar.gz
+      - run: mv c3-macos-Release/c3-macos-Release.zip c3-macos-Release/c3-macos.zip
+      - run: mv c3-macos-Debug/c3-macos-Debug.zip c3-macos-Debug/c3-macos-debug.zip
+
+      - id: create_release
+        uses: softprops/action-gh-release@v2
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        with:
+          tag_name: latest
+          release_name: latest
+          draft: false
+          prerelease: true
+          files: |
+            c3-windows.zip
+            c3-windows-debug.zip
+            c3-linux-Release/c3-linux.tar.gz
+            c3-linux-Debug/c3-linux-debug.tar.gz
+            c3-ubuntu-20-Release/c3-ubuntu-20.tar.gz
+            c3-ubuntu-20-Debug/c3-ubuntu-20-debug.tar.gz
+            c3-macos-Release/c3-macos.zip
+            c3-macos-Debug/c3-macos-debug.zip

.gitignore

@@ -19,6 +19,7 @@
 
 # Libraries
 *.lib
+*.tlb
 *.a
 *.la
 *.lo
@@ -67,3 +68,20 @@ out/
 
 /cmake-build-debug/
 /cmake-build-release/
+
+# Emacs files
+TAGS
+
+# Clangd LSP files
+/.cache/
+/compile_commands.json
+
+# 'nix build' resulting symlink
+result
+
+# macOS
+.DS_Store
+
+# tests
+/test/tmp/*
+/test/testrun

.gitmodules

@@ -0,0 +1,3 @@
+[submodule "tilde-backend"]
+	path = tilde-backend
+	url = https://github.com/c3lang/tilde-backend

CMakeLists.txt

@@ -1,13 +1,71 @@
-cmake_minimum_required(VERSION 3.10)
-project(c3c)
+cmake_minimum_required(VERSION 3.20)
+
+# Grab the version
+file(READ "src/version.h" ver)
+if (NOT ${ver} MATCHES "COMPILER_VERSION \"([0-9]+.[0-9]+.[0-9]+)\"")
+    message(FATAL_ERROR "Compiler version could not be parsed from version.h")
+endif()
+
+# Set the project and version
+project(c3c VERSION ${CMAKE_MATCH_1})
+message("C3C version: ${CMAKE_PROJECT_VERSION}")
+
+# Avoid warning for FetchContent
+if (CMAKE_VERSION VERSION_GREATER_EQUAL "3.24.0")
+    cmake_policy(SET CMP0135 NEW)
+endif()
+
+if (NOT DEFINED CMAKE_INSTALL_LIBDIR)
+    if (MSVC)
+        set(CMAKE_INSTALL_LIBDIR "c:\\c3c\\lib")
+        set(CMAKE_INSTALL_BINDIR "c:\\c3c")
+    else ()
+        set(CMAKE_INSTALL_LIBDIR "/usr/local/lib/c3")
+        set(CMAKE_INSTALL_BINDIR "/usr/local/bin/c3c")
+    endif()
+endif ()
+
+# Enable fetching (for Windows)
 include(FetchContent)
 include(FeatureSummary)
 
 set(CMAKE_FIND_PACKAGE_SORT_ORDER NATURAL)
 set(CMAKE_FIND_PACKAGE_SORT_DIRECTION DEC)
+
+# We use C11 and C++17
+set(CMAKE_C_STANDARD 11)
+set(CMAKE_CXX_STANDARD 17)
+
+if(MSVC)
+    message(STATUS "MSVC version ${MSVC_VERSION}")
+    set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} /O2 /EHsc /utf-8")
+    set(CMAKE_C_FLAGS_RELEASE "${CMAKE_C_FLAGS_RELEASE} /O2 /EHsc /utf-8")
+    set(CMAKE_CXX_FLAGS_DEBUG "${CMAKE_CXX_FLAGS_DEBUG} /Od /Zi /EHa /utf-8")
+    set(CMAKE_C_FLAGS_DEBUG "${CMAKE_C_FLAGS_DEBUG} /Od /Zi /EHa /utf-8")
+else()
+    if (true)
+        set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} -O0 -fno-exceptions")
+        set(CMAKE_CXX_FLAGS_DEBUG "${CMAKE_CXX_FLAGS_DEBUG} -gdwarf-3 -O0 -fno-exceptions")
+        set(CMAKE_C_FLAGS_RELEASE "${CMAKE_C_FLAGS_RELEASE} -gdwarf-3 -O3 -fno-exceptions")
+        set(CMAKE_C_FLAGS_DEBUG "${CMAKE_C_FLAGS_DEBUG} -gdwarf-3 -fno-exceptions")
+    else()
+        set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} -gdwarf-3 -O3 -fsanitize=undefined,address -fno-exceptions")
+        set(CMAKE_CXX_FLAGS_DEBUG "${CMAKE_CXX_FLAGS_DEBUG} -gdwarf-3 -O1 -fsanitize=undefined,address -fno-exceptions")
+        set(CMAKE_C_FLAGS_RELEASE "${CMAKE_C_FLAGS_RELEASE} -gdwarf-3 -O3 -fsanitize=undefined,address -fno-exceptions")
+        set(CMAKE_C_FLAGS_DEBUG "${CMAKE_C_FLAGS_DEBUG} -gdwarf-3 -O1 -fsanitize=undefined,address -fno-exceptions")
+        set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -fsanitize=undefined,address -fno-exceptions")
+    endif()
+endif()
+
+option(C3_LINK_DYNAMIC "link dynamically with LLVM/LLD libs")
+
 set(C3_LLVM_VERSION "auto" CACHE STRING "Use LLVM version [default: auto]")
 option(C3_USE_MIMALLOC "Use built-in mimalloc" OFF)
+option(C3_USE_TB "Use TB" OFF)
 set(C3_MIMALLOC_TAG "v1.7.3" CACHE STRING "Used version of mimalloc")
+option(C3_WITH_LLVM "Build with LLVM" ON)
+option(C3_LLD_DIR "Use custom LLD directory" "")
+option(LLVM_CRT_LIBRARY_DIR "Use custom llvm's compiler-rt directory" "")
 
 set(C3_USE_MIMALLOC OFF)
 if(C3_USE_MIMALLOC)
@@ -16,145 +74,241 @@ if(C3_USE_MIMALLOC)
     option(MI_PADDING OFF)
     option(MI_DEBUG_FULL OFF)
     FetchContent_Declare(
-    mimalloc
-    GIT_REPOSITORY https://github.com/microsoft/mimalloc.git
-    GIT_TAG ${C3_MIMALLOC_TAG}
+            mimalloc
+            GIT_REPOSITORY https://github.com/microsoft/mimalloc.git
+            GIT_TAG ${C3_MIMALLOC_TAG}
     )
     FetchContent_MakeAvailable(mimalloc)
 endif()
-
-if(NOT C3_LLVM_VERSION STREQUAL "auto")
-    if(${C3_LLVM_VERSION} VERSION_LESS 12 OR ${C3_LLVM_VERSION} VERSION_GREATER 15)
-        message(FATAL_ERROR "LLVM ${C3_LLVM_VERSION} is not supported!")
+if (NOT WIN32)
+    find_package(CURL)
+endif()
+if(C3_WITH_LLVM)
+    if (NOT C3_LLVM_VERSION STREQUAL "auto")
+        if (${C3_LLVM_VERSION} VERSION_LESS 17 OR ${C3_LLVM_VERSION} VERSION_GREATER 21)
+            message(FATAL_ERROR "LLVM ${C3_LLVM_VERSION} is not supported!")
+        endif()
     endif()
-    find_package(LLVM ${C3_LLVM_VERSION} REQUIRED CONFIG)
-else()
-   find_package(LLVM REQUIRED CONFIG)
 endif()
 
-message(STATUS "Found LLVM ${LLVM_PACKAGE_VERSION}")
-message(STATUS "Using LLVMConfig.cmake in: ${LLVM_DIR}")
-message(STATUS "Libraries located in: ${LLVM_LIBRARY_DIRS}")
+find_package(Git QUIET)
+if(C3_USE_TB AND GIT_FOUND AND EXISTS "${CMAKE_SOURCE_DIR}/.git")
+    # Update submodules as needed
+    option(GIT_SUBMODULE "Check submodules during build" ON)
+    if(GIT_SUBMODULE)
+        message(STATUS "Submodule update")
+        execute_process(COMMAND ${GIT_EXECUTABLE} submodule update --init --recursive
+                WORKING_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR}
+                RESULT_VARIABLE GIT_SUBMOD_RESULT)
+        if(NOT GIT_SUBMOD_RESULT EQUAL "0")
+            message(FATAL_ERROR "git submodule update --init --recursive failed with ${GIT_SUBMOD_RESULT}, please checkout submodules")
+        endif()
+    endif()
+endif()
 
-include_directories(${LLVM_INCLUDE_DIRS})
-add_definitions(${LLVM_DEFINITIONS})
+# Clangd LSP support
+option(C3_ENABLE_CLANGD_LSP "Enable/Disable output of compile commands during generation." OFF)
+if(C3_ENABLE_CLANGD_LSP)
+    set(CMAKE_EXPORT_COMPILE_COMMANDS ON)
+    execute_process(
+            COMMAND ${CMAKE_COMMAND} -E create_symlink
+            ${CMAKE_BINARY_DIR}/compile_commands.json
+            ${CMAKE_SOURCE_DIR}/compile_commands.json
+    )
+endif(C3_ENABLE_CLANGD_LSP)
 
-set(CMAKE_C_STANDARD 11)
-set(CMAKE_CXX_STANDARD 17)
+if(C3_WITH_LLVM)
+    if(CMAKE_C_COMPILER_ID STREQUAL "MSVC")
+        if (C3_LLVM_VERSION STREQUAL "auto")
+            set(C3_LLVM_VERSION "19")
+        endif()
+        FetchContent_Declare(
+                LLVM_Windows
+                URL https://github.com/c3lang/win-llvm/releases/download/llvm_19_1_5/llvm-19.1.5-windows-amd64-msvc17-libcmt.7z
+        )
+        FetchContent_Declare(
+                LLVM_Windows_debug
+                URL https://github.com/c3lang/win-llvm/releases/download/llvm_19_1_5/llvm-19.1.5-windows-amd64-msvc17-libcmt-dbg.7z
+        )
+        if(CMAKE_BUILD_TYPE STREQUAL "Debug")
+            message("Loading Windows LLVM debug libraries, this may take a while...")
+            FetchContent_MakeAvailable(LLVM_Windows_debug)
+            set(llvm_dir ${llvm_windows_debug_SOURCE_DIR})
+        else()
+            message("Loading Windows LLVM libraries, this may take a while...")
+            FetchContent_MakeAvailable(LLVM_Windows)
+            set(llvm_dir ${llvm_windows_SOURCE_DIR})
+        endif()
+        set(CMAKE_SYSTEM_PREFIX_PATH ${llvm_dir} ${CMAKE_SYSTEM_PREFIX_PATH})
 
-set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} -O1")
-set(CMAKE_C_FLAGS_RELEASE "${CMAKE_C_FLAGS_RELEASE} -gdwarf-3 -O3")
-set(CMAKE_C_FLAGS_DEBUG "${CMAKE_C_FLAGS_DEBUG} -gdwarf-3")
+        find_package(LLVM REQUIRED CONFIG)
+        find_package(LLD REQUIRED CONFIG)
+    else()
+        if (NOT C3_LLVM_VERSION STREQUAL "auto")
+            find_package(LLVM ${C3_LLVM_VERSION} REQUIRED CONFIG)
+        else()
+            find_package(LLVM REQUIRED CONFIG)
+        endif()
+    endif()
 
-set(LLVM_LINK_COMPONENTS
-        AllTargetsAsmParsers
-        AllTargetsCodeGens
-        AllTargetsDescs
-        AllTargetsDisassemblers
-        AllTargetsInfos
-        Analysis
-        AsmPrinter
-        BitReader
-        Core
-        DebugInfoPDB
-        InstCombine
-        IrReader
-        LibDriver
-        Linker
-        LTO
-        MC
-        MCDisassembler
-        native
-        nativecodegen
-        Object
-        Option
-        ScalarOpts
-        Support
-        Target
-        TransformUtils
-        WindowsManifest
+    if (EXISTS /usr/lib)
+        # Some systems (such as Alpine Linux) seem to put some of the relevant
+        # LLVM files in /usr/lib, but this doesn't seem to be included in the
+        # value of LLVM_LIBRARY_DIRS.
+        list(APPEND LLVM_LIBRARY_DIRS /usr/lib)
+    endif()
+
+    message(STATUS "Found LLVM ${LLVM_PACKAGE_VERSION}")
+    message(STATUS "Using LLVMConfig.cmake in: ${LLVM_DIR}")
+    message(STATUS "Libraries located in: ${LLVM_LIBRARY_DIRS}")
+
+    if (NOT LLVM_PACKAGE_VERSION VERSION_GREATER_EQUAL 15.0)
+        message(FATAL_ERROR "LLVM version 15.0 or later is required.")
+    endif()
+
+    if(LLVM_ENABLE_RTTI)
+        message(STATUS "LLVM was built with RTTI")
+    else()
+        message(STATUS "LLVM was not built with RTTI")
+    endif()
+
+    string(REPLACE "." ";" VERSION_LIST ${LLVM_PACKAGE_VERSION})
+    list(GET VERSION_LIST 0 LLVM_MAJOR_VERSION)
+
+    include_directories(${LLVM_INCLUDE_DIRS})
+    link_directories(${LLVM_LIBRARY_DIRS})
+    add_definitions(${LLVM_DEFINITIONS})
+
+    if(NOT C3_LINK_DYNAMIC)
+        set(LLVM_LINK_COMPONENTS
+                AllTargetsAsmParsers
+                AllTargetsCodeGens
+                AllTargetsDescs
+                AllTargetsDisassemblers
+                AllTargetsInfos
+                Analysis
+                AsmPrinter
+                BitReader
+                Core
+                DebugInfoPDB
+                InstCombine
+                IrReader
+                LibDriver
+                Linker
+                LTO
+                MC
+                MCDisassembler
+                native
+                nativecodegen
+                Object
+                Option
+                ScalarOpts
+                Support
+                Target
+                TransformUtils
+                WindowsManifest
+                WindowsDriver
         )
 
-if (${LLVM_PACKAGE_VERSION} VERSION_GREATER 14)
-    set(LLVM_LINK_COMPONENTS ${LLVM_LINK_COMPONENTS} WindowsDriver)
+        llvm_map_components_to_libnames(llvm_libs ${LLVM_LINK_COMPONENTS})
+
+        if(NOT ${C3_LLD_DIR} EQUAL "" AND EXISTS ${C3_LLD_DIR})
+            message("C3_LLD_DIR: " ${C3_LLD_DIR})
+            set(LLVM_LIBRARY_DIRS
+                    "${LLVM_LIBRARY_DIRS}"
+                    "${C3_LLD_DIR}"
+            )
+            list(REMOVE_DUPLICATES LLVM_LIBRARY_DIRS)
+        endif()
+
+        # These don't seem to be reliable on windows.
+        message(STATUS "using find_library")
+        find_library(LLD_COFF NAMES liblldCOFF.dylib lldCOFF.lib lldCOFF.a liblldCOFF.dll.a liblldCOFF.a PATHS ${LLVM_LIBRARY_DIRS} NO_DEFAULT_PATH)
+        find_library(LLD_COMMON NAMES liblldCommon.dylib lldCommon.lib lldCommon.a liblldCommon.dll.a liblldCommon.a PATHS ${LLVM_LIBRARY_DIRS} NO_DEFAULT_PATH)
+        find_library(LLD_ELF NAMES liblldELF.dylib lldELF.lib lldELF.a liblldELF.dll.a liblldELF.a PATHS ${LLVM_LIBRARY_DIRS} NO_DEFAULT_PATH)
+        find_library(LLD_MACHO NAMES liblldMachO.dylib lldMachO.lib lldMachO.a liblldMachO.dll.a liblldMachO.a PATHS ${LLVM_LIBRARY_DIRS} NO_DEFAULT_PATH)
+        find_library(LLD_MINGW NAMES liblldMinGW.dylib lldMinGW.lib lldMinGW.a liblldMinGW.dll.a liblldMinGW.a PATHS ${LLVM_LIBRARY_DIRS} NO_DEFAULT_PATH)
+        find_library(LLD_WASM NAMES liblldWasm.dylib lldWasm.lib lldWasm.a liblldWasm.dll.a liblldWasm.a PATHS ${LLVM_LIBRARY_DIRS} NO_DEFAULT_PATH)
+    else()
+        find_library(LLVM NAMES libLLVM.so PATHS ${LLVM_LIBRARY_DIRS} NO_DEFAULT_PATH)
+        set(llvm_libs ${LLVM})
+
+        # These don't seem to be reliable on windows.
+        message(STATUS "using find_library")
+        find_library(LLD_COFF NAMES liblldCOFF.so PATHS ${LLVM_LIBRARY_DIRS} NO_DEFAULT_PATH)
+        find_library(LLD_COMMON NAMES liblldCommon.so PATHS ${LLVM_LIBRARY_DIRS} NO_DEFAULT_PATH)
+        find_library(LLD_ELF NAMES liblldELF.so PATHS ${LLVM_LIBRARY_DIRS} NO_DEFAULT_PATH)
+        find_library(LLD_MACHO NAMES liblldMachO.so PATHS ${LLVM_LIBRARY_DIRS} NO_DEFAULT_PATH)
+        find_library(LLD_MINGW NAMES liblldMinGW.so PATHS ${LLVM_LIBRARY_DIRS} NO_DEFAULT_PATH)
+        find_library(LLD_WASM NAMES liblldWasm.so PATHS ${LLVM_LIBRARY_DIRS} NO_DEFAULT_PATH)
+    endif()
 endif()
 
-llvm_map_components_to_libnames(llvm_libs ${LLVM_LINK_COMPONENTS})
-
-file(COPY ${CMAKE_SOURCE_DIR}/lib DESTINATION ${CMAKE_BINARY_DIR})
-
-# These don't seem to be reliable on windows.
-if(UNIX)
-    message(STATUS "using find_library")
-#    find_library(TB_LIB NAMES tinybackend.a PATHS ${CMAKE_SOURCE_DIR}/resources/tblib)
-    find_library(LLD_COFF NAMES lldCOFF.a liblldCOFF.a PATHS ${LLVM_LIBRARY_DIRS})
-    find_library(LLD_COMMON NAMES lldCommon.a liblldCommon.a PATHS ${LLVM_LIBRARY_DIRS})
-    find_library(LLD_ELF NAMES lldELF.a liblldELF.a PATHS ${LLVM_LIBRARY_DIRS})
-    find_library(LLD_MACHO NAMES lldMachO.a liblldMachO.a PATHS ${LLVM_LIBRARY_DIRS})
-    find_library(LLD_MINGW NAMES lldMinGW.a liblldMinGW.a PATHS ${LLVM_LIBRARY_DIRS})
-    find_library(LLD_WASM NAMES lldWasm.a liblldWasm.a PATHS ${LLVM_LIBRARY_DIRS})
-
-    if (${LLVM_PACKAGE_VERSION} VERSION_LESS 14)
-        find_library(LLD_CORE NAMES lldCore.a liblldCore.a PATHS ${LLVM_LIBRARY_DIRS})
-        find_library(LLD_DRIVER NAMES lldDriver.a liblldDriver.a PATHS ${LLVM_LIBRARY_DIRS})
-        find_library(LLD_READER_WRITER NAMES lldReaderWriter.a liblldReaderWriter.a PATHS ${LLVM_LIBRARY_DIRS})
-        find_library(LLD_YAML NAMES lldYAML.a liblldYAML.a PATHS ${LLVM_LIBRARY_DIRS})
-    endif()
+if (NOT(${CMAKE_BINARY_DIR} EQUAL ${CMAKE_SOURCE_DIR}))
+    file(REMOVE_RECURSE ${CMAKE_BINARY_DIR}/lib)
+    file(COPY ${CMAKE_SOURCE_DIR}/lib DESTINATION ${CMAKE_BINARY_DIR})
+endif()
 
+if(C3_WITH_LLVM)
+    find_library(LLD_LOONG NAMES libLLVMLoongArchCodeGen.lib libLLVMLoongArchAsmParser.lib libLLVMLoongArchCodeGen.a libLLVMLoongArchAsmParser.a PATHS ${LLVM_LIBRARY_DIRS} NO_DEFAULT_PATH)
     set(lld_libs
-        ${LLD_COFF}
-        ${LLD_COMMON}
-        ${LLD_WASM}
-        ${LLD_MINGW}
-        ${LLD_ELF}
-        ${LLD_DRIVER}
-        ${LLD_READER_WRITER}
-        ${LLD_MACHO}
-        ${LLD_YAML}
-        ${LLD_CORE}
+            ${LLD_COFF}
+            ${LLD_WASM}
+            ${LLD_MINGW}
+            ${LLD_ELF}
+            ${LLD_MACHO}
+            ${LLD_COMMON}
+    )
+
+    if (APPLE)
+        set(lld_libs ${lld_libs} xar)
+        find_file(RT_ASAN_DYNAMIC NAMES libclang_rt.asan_osx_dynamic.dylib PATHS "${LLVM_LIBRARY_DIR}/clang/${LLVM_MAJOR_VERSION}/lib/darwin" ${LLVM_CRT_LIBRARY_DIR})
+        find_file(RT_TSAN_DYNAMIC NAMES libclang_rt.tsan_osx_dynamic.dylib PATHS "${LLVM_LIBRARY_DIR}/clang/${LLVM_MAJOR_VERSION}/lib/darwin" ${LLVM_CRT_LIBRARY_DIR})
+        find_file(RT_UBSAN_DYNAMIC NAMES libclang_rt.ubsan_osx_dynamic.dylib PATHS "${LLVM_LIBRARY_DIR}/clang/${LLVM_MAJOR_VERSION}/lib/darwin" ${LLVM_CRT_LIBRARY_DIR})
+        find_file(RT_LSAN_DYNAMIC NAMES libclang_rt.lsan_osx_dynamic.dylib PATHS "${LLVM_LIBRARY_DIR}/clang/${LLVM_MAJOR_VERSION}/lib/darwin" ${LLVM_CRT_LIBRARY_DIR})
+        set(sanitizer_runtime_libraries
+                ${RT_ASAN_DYNAMIC}
+                ${RT_TSAN_DYNAMIC}
+                # Unused
+                # ${RT_UBSAN_DYNAMIC}
+                # ${RT_LSAN_DYNAMIC}
         )
-    message(STATUS "linking to llvm libs ${llvm_libs} ${lld_libs}")
+    endif()
+
+    message(STATUS "linking to llvm libs ${lld_libs}")
+    message(STATUS "Found lld libs ${lld_libs}")
 endif()
 
-message(STATUS "Found LLD ${lld_libs}")
-
-add_library(c3c_wrappers STATIC wrapper/src/wrapper.cpp)
+add_library(miniz STATIC dependencies/miniz/miniz.c)
 
 add_executable(c3c
         src/build/builder.c
         src/build/build_options.c
         src/build/project_creation.c
+        src/build/project_manipulation.c
+        src/build/libraries.c
         src/compiler/ast.c
         src/compiler/bigint.c
-        src/compiler/c_abi_internal.h
         src/compiler/codegen_general.c
         src/compiler/compiler.c
         src/compiler/compiler.h
+        src/compiler/subprocess.c
+        src/compiler/subprocess.h
         src/compiler/context.c
         src/compiler/copying.c
         src/compiler/diagnostics.c
-        src/compiler/dwarf.h
-        src/compiler/enums.h
         src/compiler/float.c
         src/compiler/headers.c
+        src/compiler/json_output.c
         src/compiler/lexer.c
         src/compiler/linker.c
-        src/compiler/llvm_codegen.c
-        src/compiler/llvm_codegen_c_abi_aarch64.c
-        src/compiler/llvm_codegen_c_abi.c
-        src/compiler/llvm_codegen_c_abi_riscv.c
-        src/compiler/llvm_codegen_c_abi_wasm.c
-        src/compiler/llvm_codegen_c_abi_win64.c
-        src/compiler/llvm_codegen_c_abi_x64.c
-        src/compiler/llvm_codegen_c_abi_x86.c
-        src/compiler/llvm_codegen_debug_info.c
-        src/compiler/llvm_codegen_expr.c
-        src/compiler/llvm_codegen_function.c
-        src/compiler/llvm_codegen_instr.c
-        src/compiler/llvm_codegen_module.c
-        src/compiler/llvm_codegen_stmt.c
-        src/compiler/llvm_codegen_type.c
-        src/compiler/llvm_codegen_value.c
+        src/compiler/abi/c_abi_aarch64.c
+        src/compiler/abi/c_abi.c
+        src/compiler/abi/c_abi_riscv.c
+        src/compiler/abi/c_abi_wasm.c
+        src/compiler/abi/c_abi_win64.c
+        src/compiler/abi/c_abi_x64.c
+        src/compiler/abi/c_abi_x86.c
         src/compiler/module.c
         src/compiler/number.c
         src/compiler/parse_expr.c
@@ -167,6 +321,9 @@ add_executable(c3c
         src/compiler/sema_expr.c
         src/compiler/sema_internal.h
         src/compiler/sema_name_resolution.c
+        src/compiler/sema_errors.c
+        src/compiler/sema_builtins.c
+        src/compiler/sema_initializers.c
         src/compiler/semantic_analyser.c
         src/compiler/sema_passes.c
         src/compiler/sema_stmts.c
@@ -174,11 +331,7 @@ add_executable(c3c
         src/compiler/source_file.c
         src/compiler/symtab.c
         src/compiler/target.c
-        src/compiler/tb_codegen.c
-        src/compiler/tilde_codegen.c
-        src/compiler/tilde_codegen_instr.c
-        src/compiler/tilde_codegen_value.c
-        src/compiler/tilde_codegen_storeload.c
+        src/compiler/sema_asm.c
         src/compiler_tests/benchmark.c
         src/compiler_tests/tests.c
         src/compiler/tokens.c
@@ -195,54 +348,213 @@ add_executable(c3c
         src/utils/vmem.c
         src/utils/vmem.h
         src/utils/whereami.c
+        src/utils/cpus.c
+        src/utils/unzipper.c
+        src/compiler/c_codegen.c
         src/compiler/decltable.c
-        src/compiler/tilde_codegen_storeload.c
-        src/compiler/llvm_codegen_storeload.c
-        src/compiler/tilde_codegen_expr.c
-        src/compiler/tilde_codegen_stmt.c
-        src/compiler/tilde_codegen_type.c)
+        src/compiler/mac_support.c
+        src/compiler/windows_support.c
+        src/compiler/codegen_asm.c
+        src/compiler/asm_target.c
+        src/compiler/expr.c
+        src/utils/time.c
+        src/utils/http.c
+        src/compiler/sema_liveness.c
+        src/build/common_build.c
+        src/compiler/sema_const.c
+        ${CMAKE_BINARY_DIR}/git_hash.h
+)
 
-if(NOT CMAKE_C_COMPILER_ID STREQUAL "MSVC")
-    message(STATUS "using gcc/clang warning switches")
-    target_compile_options(c3c PRIVATE -Wall -Werror -Wno-unknown-pragmas -Wno-unused-result
-             -Wno-unused-function -Wno-unused-variable -Wno-unused-parameter)
+if(GIT_FOUND AND EXISTS "${CMAKE_SOURCE_DIR}/.git")
+    # We are inside of a git repository so rebuilding the hash every time something changes.
+    add_custom_command(
+            OUTPUT ${CMAKE_BINARY_DIR}/git_hash.h
+            COMMAND ${CMAKE_COMMAND} -P "${CMAKE_CURRENT_LIST_DIR}/git_hash.cmake"
+            DEPENDS "${CMAKE_CURRENT_LIST_DIR}/.git")
+else()
+    # We are NOT inside of a git repository. Building the has only once.
+    add_custom_command(
+            OUTPUT ${CMAKE_BINARY_DIR}/git_hash.h
+            COMMAND ${CMAKE_COMMAND} -P "${CMAKE_CURRENT_LIST_DIR}/git_hash.cmake")
+endif()
+
+if(C3_WITH_LLVM)
+    target_sources(c3c PRIVATE
+            src/compiler/llvm_codegen.c
+            src/compiler/llvm_codegen_debug_info.c
+            src/compiler/llvm_codegen_expr.c
+            src/compiler/llvm_codegen_function.c
+            src/compiler/llvm_codegen_instr.c
+            src/compiler/llvm_codegen_module.c
+            src/compiler/llvm_codegen_stmt.c
+            src/compiler/llvm_codegen_type.c
+            src/compiler/llvm_codegen_value.c
+            src/compiler/llvm_codegen_storeload.c
+            src/compiler/llvm_codegen_builtins.c)
+
+    target_compile_definitions(c3c PUBLIC LLVM_AVAILABLE=1)
+    add_library(c3c_wrappers STATIC wrapper/src/wrapper.cpp)
+else()
+    target_sources(c3c PRIVATE src/utils/hostinfo.c)
+    target_compile_definitions(c3c PUBLIC LLVM_AVAILABLE=0)
 endif()
 
 target_include_directories(c3c PRIVATE
-        "${CMAKE_SOURCE_DIR}/src/")
+        "${CMAKE_SOURCE_DIR}/src/"
+        "${CMAKE_BINARY_DIR}")
 
-target_include_directories(c3c PRIVATE
-        "${CMAKE_SOURCE_DIR}/tb/")
+target_include_directories(miniz PUBLIC
+        "${CMAKE_SOURCE_DIR}/dependencies/miniz/")
 
-target_include_directories(c3c_wrappers PRIVATE
-        "${CMAKE_SOURCE_DIR}/wrapper/src/")
+if (C3_USE_TB)
+    file(GLOB tilde-sources
+            tilde-backend/src/tb/*.c
+            tilde-backend/src/tb/codegen/*.c
+            tilde-backend/src/tb/bigint/*.c
+            tilde-backend/src/tb/objects/*.c
+            tilde-backend/src/tb/system/*.c
+            tilde-backend/src/tb/debug/cv/*.c
+            tilde-backend/src/tb/opt/*.c
+            tilde-backend/src/tb/x64/*.c
+            tilde-backend/src/tb/wasm/*.c
+            tilde-backend/src/tb/aarch64/*.c
+    )
+    target_sources(c3c PRIVATE
+            src/compiler/tilde_codegen.c
+            src/compiler/tilde_codegen_instr.c
+            src/compiler/tilde_codegen_value.c
+            src/compiler/tilde_codegen_storeload.c
+            src/compiler/tilde_codegen_expr.c
+            src/compiler/tilde_codegen_stmt.c
+            src/compiler/tilde_codegen_type.c
+            src/compiler/tilde_codegen_abi.c
+            src/compiler/tilde_codegen_storeload.c)
 
+    target_compile_definitions(c3c PUBLIC TB_AVAILABLE=1)
+    target_link_libraries(c3c tilde-backend)
+    add_library(tilde-backend STATIC ${tilde-sources})
+    target_include_directories(tilde-backend PRIVATE
+            "${CMAKE_SOURCE_DIR}/tilde-backend/src/" "${CMAKE_SOURCE_DIR}/tilde-backend/include")
 
+    target_include_directories(c3c PRIVATE
+            "${CMAKE_SOURCE_DIR}/tilde-backend/include/")
 
-if(UNIX)
-    target_link_libraries(c3c_wrappers ${llvm_libs} ${lld_libs})
-    target_link_libraries(c3c m ${llvm_libs} c3c_wrappers ${lld_libs})
-#    target_link_libraries(c3c m ${llvm_libs} c3c_wrappers ${TB_LIB} ${lld_libs})
-    if(C3_USE_MIMALLOC)
-        target_link_libraries(c3c m mimalloc-static)
-    endif()
 else()
-    # todo: maybe get this from llvm-config somehow? it should be in LLVM_DIR\..\..\..\bin I think.
-    set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -LIBPATH:C:\\llvm\\llvm\\build\\Release\\lib") # needed for lldCommon.lib
-    #set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -LIBPATH:${LLVM_LIBPATH}") # This doesn't seem to work for some reason
-    message(STATUS "${LLVM_LIBPATH}")
-    target_link_libraries(c3c debug ${llvm_libs} c3c_wrappers lldCommon lldCore lldCOFF lldWASM lldMinGW lldELF lldDriver lldReaderWriter lldMachO lldYAML Advapi32)
-    target_link_libraries(c3c optimized ${llvm_libs} c3c_wrappers lldCommon lldCore lldCOFF lldWASM lldMinGW lldELF lldDriver lldReaderWriter lldMachO lldYAML Advapi32)
+
+    target_compile_definitions(c3c PUBLIC TB_AVAILABLE=0)
+
+endif()
+
+if(C3_WITH_LLVM)
+    target_link_libraries(c3c ${llvm_libs} miniz c3c_wrappers ${lld_libs})
+
+    target_include_directories(c3c PRIVATE
+            "${CMAKE_SOURCE_DIR}/wrapper/include/")
+
+    target_include_directories(c3c_wrappers PRIVATE
+            "${CMAKE_SOURCE_DIR}/wrapper/include/")
+
+    target_link_libraries(c3c_wrappers ${llvm_libs} ${lld_libs})
+
+else()
+
+    target_link_libraries(c3c ${llvm_libs} miniz ${lld_libs})
+
+endif()
+
+if(C3_USE_MIMALLOC)
+    target_link_libraries(c3c mimalloc-static)
 endif()
 
 if (WIN32)
-    if (CMAKE_C_COMPILER_ID STREQUAL "Clang" OR CMAKE_C_COMPILE_ID STREQUAL "GNU")
-        target_link_options(c3c PRIVATE -pthread)
-        target_compile_definitions(c3c PRIVATE USE_PTHREAD=1)
-        target_compile_options(c3c PRIVATE -mlong-double-64)
+    target_link_libraries(c3c Winhttp.lib)
+endif()
+
+if(MINGW)
+    message("Increase stack for msys")
+    set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -Wl,--stack,8388608")
+endif ()
+
+if (CURL_FOUND)
+    target_link_libraries(c3c ${CURL_LIBRARIES})
+    target_include_directories(c3c PRIVATE ${CURL_INCLUDES})
+    target_compile_definitions(c3c PUBLIC CURL_FOUND=1)
+else()
+    target_compile_definitions(c3c PUBLIC CURL_FOUND=0)
+endif()
+
+
+if(MSVC)
+    message("Adding MSVC options")
+    target_compile_options(c3c PRIVATE /wd4068 /wd4090 /WX /Wv:18)
+    if(C3_WITH_LLVM)
+        target_compile_options(c3c_wrappers PUBLIC /wd4624 /wd4267 /wd4244 /WX /Wv:18)
+        if(NOT LLVM_ENABLE_RTTI)
+            target_compile_options(c3c_wrappers PUBLIC /GR-)
+        endif()
+        target_link_options(c3c_wrappers PUBLIC /ignore:4099)
     endif()
+    if(CMAKE_BUILD_TYPE STREQUAL "Debug")
+        target_compile_options(c3c PUBLIC /MTd)
+        if (C3_WITH_LLVM)
+            target_compile_options(c3c_wrappers PUBLIC /MTd)
+        endif()
+        target_compile_options(miniz PUBLIC /MTd)
+        if (C3_USE_TB)
+            target_compile_options(tilde-backend PUBLIC /MTd)
+        endif()
+    else()
+        target_compile_options(c3c PUBLIC /MT)
+        if (C3_WITH_LLVM)
+            target_compile_options(c3c_wrappers PUBLIC /MT)
+        endif()
+        target_compile_options(miniz PUBLIC /MT)
+        if (C3_USE_TB)
+            target_compile_options(tilde-backend PUBLIC /MT)
+        endif()
+    endif()
+    if(C3_WITH_LLVM)
+        set(clang_lib_dir ${llvm_dir}/lib/clang/${C3_LLVM_VERSION}/lib/windows)
+        set(sanitizer_runtime_libraries
+                ${clang_lib_dir}/clang_rt.asan-x86_64.lib
+                ${clang_lib_dir}/clang_rt.asan_dynamic-x86_64.lib
+                ${clang_lib_dir}/clang_rt.asan_dynamic-x86_64.dll
+                ${clang_lib_dir}/clang_rt.asan_dynamic_runtime_thunk-x86_64.lib)
+    endif()
+else()
+    message(STATUS "using gcc/clang warning switches")
+    target_link_options(c3c PRIVATE -pthread)
+    if (C3_WITH_LLVM AND NOT LLVM_ENABLE_RTTI)
+        target_compile_options(c3c_wrappers PRIVATE -fno-rtti)
+    endif()
+    target_compile_options(c3c PRIVATE -pthread -Wall -Werror -Wno-unknown-pragmas -Wno-unused-result
+            -Wno-unused-function -Wno-unused-variable -Wno-unused-parameter)
 endif()
 
 install(TARGETS c3c DESTINATION bin)
+install(DIRECTORY lib/ DESTINATION lib/c3)
+
+# Man page install (OSX/Linux only)
+if (NOT WIN32)
+    install(FILES c3c.1 DESTINATION "share/man/man1")
+endif()
+
+if (C3_WITH_LLVM AND DEFINED sanitizer_runtime_libraries)
+    add_custom_command(TARGET c3c POST_BUILD
+            COMMAND "${CMAKE_COMMAND}" -E rm -rf -- $<TARGET_FILE_DIR:c3c>/c3c_rt
+            COMMAND "${CMAKE_COMMAND}" -E make_directory $<TARGET_FILE_DIR:c3c>/c3c_rt
+            COMMAND "${CMAKE_COMMAND}" -E copy ${sanitizer_runtime_libraries} $<TARGET_FILE_DIR:c3c>/c3c_rt
+            VERBATIM
+            COMMENT "Copying sanitizer runtime libraries to output directory")
+
+    if (APPLE)
+        # Change LC_ID_DYLIB to be rpath-based instead of having an absolute path
+        add_custom_command(TARGET c3c POST_BUILD
+                COMMAND find $<TARGET_FILE_DIR:c3c>/c3c_rt -type f -name "*.dylib" -execdir ${LLVM_TOOLS_BINARY_DIR}/llvm-install-name-tool -id @rpath/{} {} $<SEMICOLON>
+                VERBATIM)
+    endif()
+
+    install(DIRECTORY $<TARGET_FILE_DIR:c3c>/c3c_rt/ DESTINATION bin/c3c_rt)
+endif()
 
 feature_summary(WHAT ALL)

CODESTYLE.md

@@ -116,7 +116,8 @@ more tests as you go along.
 ### Don't bring in dependencies
 
 External libraries has maintainability issues. Try to depend on as few libraries
-as possible. Currently c3c only depends on LLVM and libc.
+as possible. Currently, c3c only depends on LLVM and libc with an optional 
+dependency on libcurl.
 
 Do use rewrites of subsets of other libraries to bring in functionality, but avoid
 copying in libraries that needs to be updated separately.

README.md

@@ -1,8 +1,17 @@
 # C3 Language
 
-C3 is a C-like language striving to be an evolution of C, rather than a 
-completely new language. As an alternative in the C/C++ niche it
-aims to be fast and close to the metal.
+C3 is a programming language that builds on the syntax and semantics of the C language,
+with the goal of evolving it while still retaining familiarity for C programmers. 
+
+It's an evolution, not a revolution: the C-like 
+for programmers who like C.
+
+Precompiled binaries for the following operating systems are available:
+ 
+- Windows x64 [download](https://github.com/c3lang/c3c/releases/download/latest/c3-windows.zip), [install instructions](#installing-on-windows-with-precompiled-binaries).
+- Debian x64 [download](https://github.com/c3lang/c3c/releases/download/latest/c3-linux.tar.gz), [install instructions](#installing-on-debian-with-precompiled-binaries).
+- Ubuntu x86 [download](https://github.com/c3lang/c3c/releases/download/latest/c3-ubuntu-20.tar.gz), [install instructions](#installing-on-ubuntu-with-precompiled-binaries).
+- MacOS Arm64 [download](https://github.com/c3lang/c3c/releases/download/latest/c3-macos.zip), [install instructions](#installing-on-mac-with-precompiled-binaries).
 
 The manual for C3 can be found at [www.c3-lang.org](http://www.c3-lang.org).
 
@@ -24,17 +33,16 @@ whole new language.
 
 ### Example code
 
-The following code shows [generic modules](http://www.c3-lang.org/generics/) (more examples can be found at http://www.c3-lang.org/examples/). 
+The following code shows [generic modules](https://c3-lang.org/references/docs/generics/) (more examples can be found at https://c3-lang.org/references/docs/examples/). 
 
-```c++
-module stack <Type>;
+```cpp
+module stack (<Type>);
 // Above: the parameterized type is applied to the entire module.
-import std::mem;
 
 struct Stack
 {
-    usize capacity;
-    usize size;
+    usz capacity;
+    usz size;
     Type* elems;
 }
 
@@ -47,7 +55,8 @@ fn void Stack.push(Stack* this, Type element)
     if (this.capacity == this.size)
     {
         this.capacity *= 2;
-        this.elems = mem::realloc(this.elems, $sizeof(Type) * this.capacity);
+        if (this.capacity < 16) this.capacity = 16;
+        this.elems = realloc(this.elems, Type.sizeof * this.capacity);
     }
     this.elems[this.size++] = element;
 }
@@ -71,20 +80,20 @@ import stack;
 
 // Define our new types, the first will implicitly create 
 // a complete copy of the entire Stack module with "Type" set to "int"
-define IntStack = Stack<int>;
+def IntStack = Stack(<int>);
 // The second creates another copy with "Type" set to "double"
-define DoubleStack = Stack<double>;
+def DoubleStack = Stack(<double>);
 
-// If we had added "define IntStack2 = Stack<int>"
+// If we had added "define IntStack2 = Stack(<int>)"
 // no additional copy would have been made (since we already
-// have an parameterization of Stack<int>) so it would
+// have an parameterization of Stack(<int>)) so it would
 // be same as declaring IntStack2 an alias of IntStack
 
 // Importing an external C function is straightforward
 // here is an example of importing libc's printf:
 extern fn int printf(char* format, ...);
 
-fn void test()
+fn void main()
 {
     IntStack stack;
     // Note that C3 uses zero initialization by default
@@ -104,17 +113,17 @@ fn void test()
     dstack.push(2.3);
     dstack.push(3.141);
     dstack.push(1.1235);
-    // Prints pop: 1.1235
+    // Prints pop: 1.123500
     printf("pop: %f\n", dstack.pop());
 }
 ```
 
-### In what ways do C3 differ from C?
+### In what ways does C3 differ from C?
 
 - No mandatory header files
 - New semantic macro system
 - Module based name spacing
-- Subarrays (slices)
+- Slices
 - Compile time reflection
 - Enhanced compile time execution
 - Generics based on generic modules
@@ -123,36 +132,60 @@ fn void test()
 - Value methods
 - Associated enum data
 - No preprocessor
-- Less undefined behaviour and runtime checks in "safe" mode
+- Less undefined behaviour and added runtime checks in "safe" mode
 - Limited operator overloading to enable userland dynamic arrays
 - Optional pre and post conditions
 
 ### Current status
 
-The current version of the compiler is alpha release 0.1.0.
+The current stable version of the compiler is **version 0.6.7**.
 
-It's possible to try out the current C3 compiler in the browser: https://ide.judge0.com/ – this is courtesy of the
-developer of Judge0. 
-
-Design work on C3 complete aside from fleshing out details, such as
-inline asm. As the standard library work progresses, changes and improvements
-to the language will happen continuously.
+The upcoming 0.6.x releases will focus on expanding the standard library.
 Follow the issues [here](https://github.com/c3lang/c3c/issues).
 
 If you have suggestions on how to improve the language, either [file an issue](https://github.com/c3lang/c3c/issues) 
 or discuss C3 on its dedicated Discord: [https://discord.gg/qN76R87](https://discord.gg/qN76R87).
 
-The compiler should compile on Linux, Windows (under MSVC, Mingw or MSYS2) and MacOS, 
-but needs some install documentation for Windows. 
+The compiler is currently verified to compile on Linux, Windows and MacOS.
 
+**Support matrix**
 
+| Platform                 | Native C3 compiler available? | Target supported        | Stack trace | Threads  | Sockets  | Inline asm |
+|--------------------------|-------------------------------|-------------------------|-------------|----------|----------|------------|
+| Win32 x64                | Yes                           | Yes + cross compilation | Yes         | Yes      | Yes      | Yes*       |
+| Win32 Aarch64            | Untested                      | Untested                | Untested    | Untested | Untested | Yes*       |
+| MacOS x64                | Yes                           | Yes + cross compilation | Yes         | Yes      | Yes      | Yes*       |
+| MacOS Aarch64            | Yes                           | Yes + cross compilation | Yes         | Yes      | Yes      | Yes*       |
+| iOS Aarch64              | No                            | Untested                | Untested    | Yes      | Yes      | Yes*       |
+| Linux x86                | Yes                           | Yes                     | Yes         | Yes      | Yes      | Yes*       |
+| Linux x64                | Yes                           | Yes                     | Yes         | Yes      | Yes      | Yes*       |
+| Linux Aarch64            | Yes                           | Yes                     | Yes         | Yes      | Yes      | Yes*       |
+| Linux Riscv32            | Yes                           | Yes                     | Yes         | Yes      | Yes      | Untested   |
+| Linux Riscv64            | Yes                           | Yes                     | Yes         | Yes      | Yes      | Untested   |
+| ELF freestanding x86     | No                            | Untested                | No          | No       | No       | Yes*       |
+| ELF freestanding x64     | No                            | Untested                | No          | No       | No       | Yes*       |
+| ELF freestanding Aarch64 | No                            | Untested                | No          | No       | No       | Yes*       |
+| ELF freestanding Riscv64 | No                            | Untested                | No          | No       | No       | Untested   |
+| ELF freestanding Riscv32 | No                            | Untested                | No          | No       | No       | Untested   |
+| FreeBSD x86              | Untested                      | Untested                | No          | Yes      | Untested | Yes*       |
+| FreeBSD x64              | Untested                      | Untested                | No          | Yes      | Untested | Yes*       |
+| NetBSD x86               | Untested                      | Untested                | No          | Yes      | Untested | Yes*       |
+| NetBSD x64               | Untested                      | Untested                | No          | Yes      | Untested | Yes*       |
+| OpenBSD x86              | Untested                      | Untested                | No          | Yes      | Untested | Yes*       |
+| OpenBSD x64              | Untested                      | Untested                | No          | Yes      | Untested | Yes*       |
+| MCU x86                  | No                            | Untested                | No          | No       | No       | Yes*       |
+| Wasm32                   | No                            | Yes                     | No          | No       | No       | No         |
+| Wasm64                   | No                            | Untested                | No          | No       | No       | No         |
+
+*\* Inline asm is still a work in progress*
+
+More platforms will be supported in the future.
 
 #### What can you help with?
 
 - If you wish to contribute with ideas, please file issues or discuss on Discord.
 - Interested in contributing to the stdlib? Please get in touch on Discord.
-- Are you a Windows dev and know your way around Github CI? Please help us get MSVC CI working!
-- Install instructions for other Linux and Unix variants are appreciated.
+- Compilation instructions for other Linux and Unix variants are appreciated.
 - Would you like to contribute bindings to some library? It would be nice to have support for SDL, Raylib and more.
 - Build something with C3 and show it off and give feedback. The language is still open for significant tweaks.
 - Start work on the C -> C3 converter which takes C code and does a "best effort" to translate it to C3. The first version only needs to work on C headers.
@@ -160,20 +193,59 @@ but needs some install documentation for Windows.
 
 ### Installing
 
-#### Installing on Ubuntu 20.10
+#### Installing on Windows with precompiled binaries
+1. Download the zip file: [https://github.com/c3lang/c3c/releases/download/latest/c3-windows.zip](https://github.com/c3lang/c3c/releases/download/latest/c3-windows.zip) 
+(debug version [here](https://github.com/c3lang/c3c/releases/download/latest/c3-windows-debug.zip))
+2. Unzip exe and standard lib.
+3. If you don't have Visual Studio 17 installed you can either do so, or run the `msvc_build_libraries.py` Python script which will download the necessary files to compile on Windows. 
+4. Run `c3c.exe`.
 
-1. Make sure you have a C compiler that handles C11 and a C++ compiler, such as GCC or Clang. Git also needs to be installed.
-2. Install CMake: `sudo apt install cmake`
-3. Install LLVM 12 (or greater: C3C supports LLVM 12-15): `sudo apt-get install clang-12 zlib1g zlib1g-dev libllvm12 llvm-12 llvm-12-dev llvm-12-runtime liblld-12-dev liblld-12`
-4. Clone the C3C github repository: `git clone https://github.com/c3lang/c3c.git`
-5. Enter the C3C directory `cd c3c`.
-6. Create a build directory `mkdir build`
-7. Change directory to the build directory `cd build`
-8. Build: `cmake --build .`
+#### Installing on Debian with precompiled binaries
+1. Download tar file: [https://github.com/c3lang/c3c/releases/download/latest/c3-linux.tar.gz](https://github.com/c3lang/c3c/releases/download/latest/c3-linux.tar.gz)
+   (debug version [here](https://github.com/c3lang/c3c/releases/download/latest/c3-linux-debug.tar.gz))
+2. Unpack executable and standard lib.
+3. Run `./c3c`.
 
-You should now have a `c3c` executable.
+#### Installing on Ubuntu with precompiled binaries
+1. Download tar file: [https://github.com/c3lang/c3c/releases/download/latest/c3-ubuntu-20.tar.gz](https://github.com/c3lang/c3c/releases/download/latest/c3-ubuntu-20.tar.gz)
+   (debug version [here](https://github.com/c3lang/c3c/releases/download/latest/c3-ubuntu-20-debug.tar.gz))
+2. Unpack executable and standard lib.
+3. Run `./c3c`.
 
-You can try it out by running some sample code: `./c3c compile ../resources/examples/hash.c3`
+#### Installing on MacOS with precompiled binaries
+1. Make sure you have XCode with command line tools installed.
+2. Download the zip file: [https://github.com/c3lang/c3c/releases/download/latest/c3-macos.zip](https://github.com/c3lang/c3c/releases/download/latest/c3-macos.zip)
+   (debug version [here](https://github.com/c3lang/c3c/releases/download/latest/c3-macos-debug.zip))
+3. Unzip executable and standard lib.
+4. Run `./c3c`.
+
+(*Note that there is a known issue with debug symbol generation on MacOS 13, see [issue #1086](https://github.com/c3lang/c3c/issues/1086))
+
+#### Installing on Arch Linux
+Arch includes c3c in the official 'extra' repo. It can be easily installed the usual way:
+
+```sh
+sudo pacman -S c3c
+# or paru -S c3c
+# or yay -S c3c
+# or aura -A c3c
+```
+
+There is also an AUR package for the c3c compiler : [c3c-git](https://aur.archlinux.org/packages/c3c-git).
+
+You can use your AUR package manager:
+```sh
+paru -S c3c-git
+# or yay -S c3c-git
+# or aura -A c3c-git
+```
+
+Or clone it manually:
+```sh
+git clone https://aur.archlinux.org/c3c-git.git
+cd c3c-git
+makepkg -si
+```
 
 #### Building via Docker
 
@@ -195,7 +267,7 @@ A `c3c` executable will be found under `bin/`.
 #### Installing on OS X using Homebrew
 
 2. Install CMake: `brew install cmake`
-3. Install LLVM 13: `brew install llvm`
+3. Install LLVM 17+: `brew install llvm`
 4. Clone the C3C github repository: `git clone https://github.com/c3lang/c3c.git`
 5. Enter the C3C directory `cd c3c`.
 6. Create a build directory `mkdir build`
@@ -203,6 +275,14 @@ A `c3c` executable will be found under `bin/`.
 8. Set up CMake build for debug: `cmake ..`
 9. Build: `cmake --build .`
 
+#### Installing on Windows using Scoop
+
+c3c is included in 'Main' bucket.
+
+```sh
+scoop install c3
+```
+
 #### Getting started with a "hello world"
 
 Create a `main.c3` file with:
@@ -212,7 +292,7 @@ import std::io;
 
 fn void main()
 {
-   io::println("Hello, world!");
+   io::printn("Hello, world!");
 }
 ```
 
@@ -223,7 +303,89 @@ Then run
 c3c compile main.c3
 ```
 
-The generated binary will be called `a.out`.
+The generated binary will by default be named after the module that contains the main
+function. In our case that is `hello_world`, so the resulting binary will be
+called `hello_world` or `hello_world.exe`depending on platform.
+
+### Compiling
+
+#### Compiling on Windows
+
+1. Make sure you have Visual Studio 17 2022 installed or alternatively install the "Buildtools for Visual Studio" (https://aka.ms/vs/17/release/vs_BuildTools.exe) and then select "Desktop development with C++"
+2. Install CMake
+3. Clone the C3C github repository: `git clone https://github.com/c3lang/c3c.git`
+4. Enter the C3C directory `cd c3c`.
+5. Set up the CMake build `cmake -B build -G "Visual Studio 17 2022" -A x64 -DCMAKE_BUILD_TYPE=Release`
+6. Build: `cmake --build build --config Release`
+7. You should now have the c3c.exe
+
+You should now have a `c3c` executable.
+
+You can try it out by running some sample code: `c3c.exe compile ../resources/examples/hash.c3`
+
+*Note that if you run into linking issues when building, make sure that you are using the latest version of VS17.*
+
+
+#### Compiling on Ubuntu 24.04 LTS
+
+1. Make sure you have a C compiler that handles C11 and a C++ compiler, such as GCC or Clang. Git also needs to be installed.
+2. Install LLVM 18 `sudo apt-get install cmake git clang zlib1g zlib1g-dev libllvm18 llvm llvm-dev llvm-runtime liblld-dev liblld-18 libpolly-18-dev`
+3. Clone the C3C github repository: `git clone https://github.com/c3lang/c3c.git`
+4. Enter the C3C directory `cd c3c`.
+5. Create a build directory `mkdir build`
+6. Change directory to the build directory `cd build`
+7. Set up CMake build: `cmake ..`
+8. Build: `cmake --build .`
+
+You should now have a `c3c` executable.
+
+You can try it out by running some sample code: `./c3c compile ../resources/examples/hash.c3`
+
+
+#### Compiling on Void Linux
+
+1. As root, ensure that all project dependencies are installed: `xbps-install git cmake llvm17 llvm17-devel lld17-devel libcurl-devel ncurses-devel zlib-devel libzstd-devel libxml2-devel`
+2. Clone the C3C repository: `git clone https://github.com/c3lang/c3c.git`
+    - If you only need the latest commit, you may want to make a shallow clone instead: `git clone https://github.com/c3lang/c3c.git --depth=1`
+3. Enter the directory: `cd c3c`
+4. Create a build directory: `mkdir build`
+5. Enter the build directory: `cd build`
+6. Create the CMake build cache: `cmake ..`
+7. Build: `cmake --build .`
+
+Your c3c executable should have compiled properly. You may want to test it: `./c3c compile ../resources/examples/hash.c3`  
+For a sytem-wide installation, run the following as root: `cmake --install .`
+
+
+#### Compiling on Fedora
+
+1. Install required project dependencies: `dnf install cmake clang git llvm llvm-devel lld lld-devel ncurses-devel`
+2. Optionally, install additional dependencies: `dnf install libcurl-devel zlib-devel libzstd-devel libxml2-devel libffi-devel`
+3. Clone the C3C repository: `git clone https://github.com/c3lang/c3c.git`
+    - If you only need the latest commit, you may want to make a shallow clone: `git clone https://github.com/c3lang/c3c.git --depth=1`
+4. Enter the C3C directory: `cd c3c`
+5. Create a build directory and navigate into it: `mkdir build && cd build`
+6. Create the CMake build cache. The Fedora repositories provide `.so` libraries for lld, so you need to set the C3_LINK_DYNAMIC flag: `cmake .. -DC3_LINK_DYNAMIC=1`
+7. Build the project: `cmake --build .`
+
+The c3c binary should be created in the build directory. You can try it out by running some sample code: `./c3c compile ../resources/examples/hash.c3`
+
+
+#### Compiling on other Linux / Unix variants
+
+1. Install CMake.
+2. Install or compile LLVM and LLD *libraries* (version 17+ or higher)
+3. Clone the C3C github repository: `git clone https://github.com/c3lang/c3c.git`
+4. Enter the C3C directory `cd c3c`.
+5. Create a build directory `mkdir build`
+6. Change directory to the build directory `cd build`
+7. Set up CMake build for debug: `cmake ..`. At this point you may need to manually 
+provide the link path to the LLVM CMake directories, e.g. `cmake -DLLVM_DIR=/usr/local/opt/llvm/lib/cmake/llvm/ ..`
+8. Build: `cmake --build .`
+
+*A note on compiling for Linux/Unix/MacOS: to be able to fetch vendor libraries
+libcurl is needed. The CMake script should detect it if it is available. Note that
+this functionality is non-essential and it is perfectly fine to user the compiler without it.*
 
 #### Licensing
 
@@ -232,4 +394,19 @@ MIT licensed.
 
 #### Editor plugins
 
-Editor plugins can be found at https://github.com/c3lang/editor-plugins.
+Editor plugins can be found at https://github.com/c3lang/editor-plugins.
+
+#### Contributing unit tests
+
+1. Write the test, either adding to existing test files in `/test/unit/` or add
+   a new file. (If testing the standard library, put it in the `/test/unit/stdlib/` subdirectory).
+2. Make sure that the test functions have the `@test` attribute. 
+3. Run tests and see that they pass. (Recommended settings: `c3c compile-test -O0 test/unit`.
+   - in this example `test/unit/` is the relative path to the test directory, so adjust as required)
+4. Make a pull request for the new tests.
+
+## Thank yous
+
+A huge **THANK YOU** goes out to all contributors and sponsors.
+
+A special thank you to sponsors [Caleb-o](https://github.com/Caleb-o) and [devdad](https://github.com/devdad) for going the extra mile.

benchmarks/stdlib/sort/quicksort.c3

@@ -0,0 +1,69 @@
+module sort_bench;
+
+import std::sort;
+
+fn void init() @init
+{
+    set_benchmark_warmup_iterations(5);
+    set_benchmark_max_iterations(10_000);
+}
+
+fn void quicksort_bench() @benchmark
+{
+	// test set: 500 numbers between 0 and 99;
+	int[] data = {
+		71, 28, 2, 13, 62, 10, 54, 78, 63, 86,
+		33, 65, 89, 51, 58, 0, 51, 16, 87, 30,
+		89, 14, 52, 41, 88, 25, 83, 91, 56, 86,
+		14, 64, 76, 18, 39, 24, 79, 62, 34, 58,
+		90, 24, 56, 73, 85, 82, 79, 63, 47, 69,
+		78, 29, 49, 28, 43, 47, 56, 53, 79, 56,
+		19, 63, 29, 52, 71, 93, 61, 46, 30, 11,
+		21, 26, 37, 86, 93, 74, 62, 0, 41, 17,
+		26, 27, 34, 11, 54, 69, 72, 44, 74, 3,
+		61, 62, 80, 90, 3, 82, 16, 12, 28, 1,
+		2, 49, 4, 44, 57, 86, 63, 74, 33, 41,
+		76, 77, 56, 57, 56, 88, 74, 71, 6, 59,
+		40, 42, 94, 55, 21, 17, 17, 63, 21, 83,
+		73, 19, 39, 88, 93, 74, 21, 0, 63, 45,
+		69, 66, 22, 68, 86, 86, 85, 67, 8, 50,
+		23, 98, 64, 80, 64, 36, 40, 30, 73, 36,
+		23, 14, 1, 77, 82, 8, 18, 73, 37, 86,
+		29, 70, 27, 87, 64, 81, 13, 0, 4, 83,
+		90, 17, 71, 66, 38, 39, 54, 22, 86, 18,
+		84, 66, 77, 25, 64, 93, 80, 91, 2, 92,
+		47, 32, 90, 16, 46, 29, 56, 87, 70, 73,
+		89, 41, 5, 54, 93, 63, 16, 39, 71, 84,
+		74, 91, 69, 59, 49, 87, 74, 37, 75, 83,
+		77, 19, 51, 44, 79, 62, 94, 20, 24, 83,
+		37, 70, 57, 32, 93, 8, 29, 11, 7, 92,
+		8, 23, 20, 21, 7, 70, 28, 20, 96, 6,
+		50, 58, 30, 61, 66, 42, 50, 54, 64, 7,
+		10, 53, 63, 44, 16, 39, 83, 73, 3, 29,
+		97, 32, 36, 68, 84, 64, 73, 5, 29, 13,
+		48, 3, 84, 65, 75, 68, 66, 22, 39, 33,
+		39, 24, 27, 85, 18, 34, 3, 63, 32, 9,
+		29, 66, 24, 90, 75, 50, 11, 95, 47, 14,
+		92, 1, 76, 45, 76, 41, 55, 54, 38, 67,
+		43, 40, 5, 61, 97, 11, 61, 24, 92, 24,
+		76, 53, 60, 34, 78, 80, 70, 75, 30, 90,
+		65, 99, 80, 61, 94, 75, 63, 67, 10, 35,
+		23, 42, 31, 48, 14, 68, 84, 14, 79, 1,
+		25, 94, 23, 53, 49, 69, 44, 73, 63, 51,
+		44, 96, 88, 51, 94, 24, 64, 72, 59, 81,
+		73, 93, 14, 35, 9, 53, 25, 48, 50, 88,
+		46, 97, 67, 40, 27, 17, 2, 42, 11, 82,
+		0, 46, 44, 38, 31, 88, 63, 88, 10, 82,
+		77, 61, 24, 39, 27, 33, 10, 91, 69, 22,
+		42, 74, 71, 13, 32, 56, 12, 46, 81, 74,
+		17, 26, 45, 50, 76, 84, 76, 36, 43, 65,
+		81, 64, 0, 49, 70, 11, 76, 19, 60, 55,
+		15, 98, 31, 91, 56, 8, 97, 9, 3, 94,
+		3, 88, 7, 2, 3, 98, 10, 51, 21, 79,
+		99, 3, 8, 76, 52, 13, 40, 90, 85, 15,
+		70, 77, 43, 30, 4, 89, 18, 21, 59, 17,
+	};
+	sort::quicksort(data);
+}
+
+

build-with-docker.sh

@@ -1,56 +1,44 @@
 #!/bin/bash
-## build-with-docker.sh
-## @author gdm85
-##
-## Script to build c3c for either Ubuntu 20, 21 or 22.
-##
-#
 
-if [ $# -ne 1 -a $# -ne 2 ]; then
-	echo "Usage: build-with-docker.sh (20|21|22) [Debug|Release]" 1>&2
-	exit 1
+: ${DOCKER:=docker}
+: ${IMAGE:="c3c-builder"}
+: ${CMAKE_BUILD_TYPE:=Release}
+: ${LLVM_VERSION:=18}
+: ${UBUNTU_VERSION:="22.04"}
+: ${CMAKE_VERSION:="3.20.0"}
+
+cd docker || exit 1  # Exit if the 'docker' directory doesn't exist
+
+$DOCKER build \
+    --build-arg LLVM_VERSION=$LLVM_VERSION \
+    --build-arg CMAKE_VERSION=$CMAKE_VERSION \
+    --build-arg UBUNTU_VERSION=$UBUNTU_VERSION \
+    -t $IMAGE .
+
+if [ $? -ne 0 ]; then
+    echo "Docker image build failed. Exiting."
+    exit 1
 fi
 
-set -e
-
-DOCKER=docker
-DOCKER_RUN=""
-IMAGE="c3c-builder"
-if type podman 2>/dev/null >/dev/null; then
-	DOCKER=podman
-	DOCKER_RUN="--userns=keep-id"
-	IMAGE="localhost/$IMAGE"
-fi
-
-if [ -z "$2" ]; then
-	CMAKE_BUILD_TYPE=Debug
-else
-	CMAKE_BUILD_TYPE="$2"
-fi
-
-TAG="$1"
-if [ "$1" = 20 ]; then
-	UBUNTU_VERSION="20.04"
-	LLVM_VERSION="12"
-elif [ "$1" = 21 ]; then
-	UBUNTU_VERSION="21.10"
-	LLVM_VERSION="13"
-elif [ "$1" = 22 ]; then
-	UBUNTU_VERSION="22.04"
-	LLVM_VERSION="14"
-else
-	echo "ERROR: expected 20, 21 or 22 as Ubuntu version argument" 1>&2
-	exit 2
-fi
-IMAGE="$IMAGE:$TAG"
-
-cd docker && $DOCKER build -t $IMAGE --build-arg UID=$(id -u) --build-arg GID=$(id -g) \
-	--build-arg DEPS="llvm-$LLVM_VERSION-dev liblld-$LLVM_VERSION-dev clang-$LLVM_VERSION libllvm$LLVM_VERSION llvm-$LLVM_VERSION-runtime" \
-	--build-arg UBUNTU_VERSION="$UBUNTU_VERSION" .
 cd ..
 
 rm -rf build bin
 mkdir -p build bin
 
-exec $DOCKER run -ti --rm --tmpfs=/tmp $DOCKER_RUN -v "$PWD":/home/c3c/source -w /home/c3c/source $IMAGE bash -c \
-	"cd build && cmake -DCMAKE_BUILD_TYPE=$CMAKE_BUILD_TYPE -DC3_LLVM_VERSION=$LLVM_VERSION .. && cmake --build . && mv c3c lib ../bin/"
+chmod -R 777 build bin
+
+exec $DOCKER run -i --rm \
+    -v "$PWD":/home/c3c/source \
+    -w /home/c3c/source $IMAGE bash -c \
+    "cmake -S . -B build \
+            -G Ninja \
+            -DCMAKE_BUILD_TYPE=$CMAKE_BUILD_TYPE \
+            -DCMAKE_C_COMPILER=clang-$LLVM_VERSION \
+            -DCMAKE_CXX_COMPILER=clang++-$LLVM_VERSION \
+            -DCMAKE_LINKER=lld-$LLVM_VERSION \
+            -DCMAKE_OBJCOPY=llvm-objcopy-$LLVM_VERSION \
+            -DCMAKE_STRIP=llvm-strip-$LLVM_VERSION \
+            -DCMAKE_DLLTOOL=llvm-dlltool-$LLVM_VERSION \
+            -DC3_LLVM_VERSION=auto && \
+    cmake --build build && \
+    cp -r build/c3c build/lib bin"

c3c.1

@@ -0,0 +1,552 @@
+.TH "c3c" "1" "2024-10-27" "C3 Compiler" "User Commands"
+.SH NAME
+c3c \- Compiler for the C3 programming language
+
+.SH SYNOPSIS
+.B c3c
+[\fIoptions\fR ...] \fIcommand\fR [\fIargs\fR ...]
+
+.SH DESCRIPTION
+.B c3c
+is the compiler for the C3 language, providing commands for compilation, project
+management, testing, and distribution. The available commands allow users to 
+compile files, initialize projects, build targets, run benchmarks, clean build 
+files, and more.
+
+.SH COMMANDS
+.PP
+.B c3c compile 
+\fIfile1\fR [\fIfile2\fR ...]
+.RS
+Compile files without a project into an executable.
+.RE
+.PP
+.B c3c init 
+\fIproject name\fR
+.RS
+Initialize a new project structure.
+.RE
+.PP
+.B c3c init-lib 
+\fIlibrary name\fR
+.RS
+Initialize a new library structure.
+.RE
+.PP
+.B c3c build 
+[\fItarget\fR]
+.RS
+Build the target in the current project.
+.RE
+.PP
+.B c3c benchmark
+.RS
+Run the benchmarks in the current project.
+.RE
+.PP
+.B c3c test
+.RS
+Run the unit tests in the current project.
+.RE
+.PP
+.B c3c clean
+.RS
+Clean all build files.
+.RE
+.PP
+.B c3c run 
+[\fItarget\fR] [-- [\fIarg1\fR ...]]
+.RS
+Run (and build if needed) the target in the current project.
+.RE
+.PP
+.B c3c dist 
+[\fItarget\fR]
+.RS
+Clean and build a target for distribution.
+.RE
+.PP
+.B c3c directives 
+[\fItarget\fR]
+.RS
+Generate documentation for the target.
+.RE
+.PP
+.B c3c bench 
+[\fItarget\fR]
+.RS
+Benchmark a target.
+.RE
+.PP
+.B c3c clean-run 
+[\fItarget\fR] [-- [\fIarg1\fR ...]]
+.RS
+Clean, then run the target.
+.RE
+.PP
+.B c3c compile-run 
+\fIfile1\fR [\fIfile2\fR ...] [-- [\fIarg1\fR ...]]
+.RS
+Compile files then immediately run the result.
+.RE
+.PP
+.B c3c compile-only 
+\fIfile1\fR [\fIfile2\fR ...]
+.RS
+Compile files but do not perform linking.
+.RE
+.PP
+.B c3c compile-benchmark 
+\fIfile1\fR [\fIfile2\fR ...]
+.RS
+Compile files into an executable and run benchmarks.
+.RE
+.PP
+.B c3c compile-test 
+\fIfile1\fR [\fIfile2\fR ...]
+.RS
+Compile files into an executable and run unit tests.
+.RE
+.PP
+.B c3c static-lib 
+\fIfile1\fR [\fIfile2\fR ...]
+.RS
+Compile files without a project into a static library.
+.RE
+.PP
+.B c3c dynamic-lib 
+\fIfile1\fR [\fIfile2\fR ...]
+.RS
+Compile files without a project into a dynamic library.
+.RE
+.PP
+.B c3c headers 
+\fIfile1\fR [\fIfile2\fR ...]
+.RS
+Analyze files and generate C headers for public methods.
+.RE
+.PP
+.B c3c vendor-fetch 
+\fIlibrary\fR ...
+.RS
+Fetch one or more libraries from the vendor collection.
+.RE
+.PP
+.B c3c project 
+\fIsubcommand\fR ...
+.RS
+Manipulate or view project files.
+.RE
+
+.SH OPTIONS
+.PP
+.B --stdlib 
+\fIdir\fR
+.RS
+Use this directory as the C3 standard library path.
+.RE
+.PP
+.B --no-entry
+.RS
+Do not generate (or require) a main function.
+.RE
+.PP
+.B --libdir 
+\fIdir\fR
+.RS
+Add this directory to the C3 library search paths.
+.RE
+.PP
+.B --lib 
+\fIname\fR
+.RS
+Add this library to the compilation.
+.RE
+.PP
+.B --path 
+\fIdir\fR
+.RS
+Use this as the base directory for the current command.
+.RE
+.PP
+.B --template 
+\fItemplate\fR
+.RS
+Select template for 'init': "exe", "static-lib", "dynamic-lib" or a path.
+.RE
+.PP
+.B --about
+Prints a short description of C3.
+.PP
+.B --symtab 
+\fIvalue\fR
+.RS
+Sets the preferred symtab size.
+.RE
+.PP
+.B --max-mem 
+\fIvalue\fR
+.RS
+Sets the preferred max memory size.
+.RE
+.PP
+.B --run-once
+.RS
+After running the output file, delete it immediately.
+.RE
+.PP
+.B -V, --version
+Print version information.
+.PP
+.B -E
+Lex only.
+.PP
+.B -P
+Only parse and output the AST as JSON.
+.PP
+.B -C
+Only lex, parse and check.
+.PP
+.B -
+\fIcode\fR...
+.RS
+Read code from standard in.
+.RE
+.PP
+.B -o 
+\fIfile\fR
+.RS
+Write output to \fIfile\fR.
+.RE
+.PP
+.B -O0
+Safe, no optimizations, emit debug info.
+.PP
+.B -O1
+Safe, high optimization, emit debug info.
+.PP
+.B -O2
+Unsafe, high optimization, emit debug info.
+.PP
+.B -O3
+Unsafe, high optimization, single module, emit debug info.
+.PP
+.B -O4
+Unsafe, highest optimization, relaxed maths, single module, emit debug info, no panic messages.
+.PP
+.B -O5
+Unsafe, highest optimization, fast maths, single module, emit debug info, no panic messages, no backtrace.
+.PP
+.B -Os
+Unsafe, high optimization, small code, single module, no debug info, no panic messages.
+.PP
+.B -Oz
+Unsafe, high optimization, tiny code, single module, no debug info, no panic messages, no backtrace.
+.PP
+.B -D 
+\fIname\fR
+.RS
+Add feature flag \fIname\fR.
+.RE
+.PP
+.B -U 
+\fIname\fR
+.RS
+Remove feature flag \fIname\fR.
+.RE
+.PP
+.B --trust=
+\fIoption\fR
+.RS
+Trust level: none (default), include ($include allowed), full ($exec / exec allowed).
+.RE
+.PP
+.B --output-dir 
+\fIdir\fR
+.RS
+Override general output directory.
+.RE
+.PP
+.B --threads 
+\fInumber\fR
+.RS
+Set the number of threads to use for compilation.
+.RE
+.PP
+.B --show-backtrace=
+\fIyes|no\fR
+.RS
+Show detailed backtrace on segfaults.
+.RE
+
+.PP
+.B -g
+Emit debug info.
+.PP
+.B -g0
+Emit no debug info.
+
+
+.PP
+.B -l 
+\fIlibrary\fR
+.RS
+Link with the library provided.
+.RE
+.PP
+.B -L 
+\fIlibrary\fR \fIdir\fR
+.RS
+Append the directory to the linker search paths.
+.RE
+.PP
+.B -z 
+\fIargument\fR
+.RS
+Send the \fIargument\fR as a parameter to the linker.
+.RE
+.PP
+.B --cc 
+\fIpath\fR
+.RS
+Set C compiler (for C files in projects and use as system linker).
+.RE
+.PP
+.B --linker=
+\fIoption\fR [\fIpath\fR]
+.RS
+Specify the linker: builtin, cc, custom (default is 'cc'). 'Custom' requires a path.
+.RE
+
+.PP
+.B --use-stdlib=
+\fIyes|no\fR
+.RS
+Include the standard library (default: yes).
+.RE
+.PP
+.B --link-libc=
+\fIyes|no\fR
+.RS
+Link libc and other default libraries (default: yes).
+.RE
+.PP
+.B --emit-stdlib=
+\fIyes|no\fR
+.RS
+Output files for the standard library (default: yes).
+.RE
+.PP
+.B --panicfn 
+\fIname\fR
+.RS
+Override the panic function name.
+.RE
+.PP
+.B --testfn 
+\fIname\fR
+.RS
+Override the test runner function name.
+.RE
+.PP
+.B --benchfn 
+\fIname\fR
+.RS
+Override the benchmark runner function name.
+.RE
+
+.PP
+.B --reloc=
+\fIoption\fR
+.RS
+Specify the relocation model: none, pic, PIC, pie, PIE.
+.RE
+.PP
+.B --x86cpu=
+\fIoption\fR
+.RS
+Set general level of x64 CPU: baseline, ssse3, sse4, avx1, avx2-v1, avx2-v2 (Skylake/Zen1+), avx512 (Icelake/Zen4+), native.
+.RE
+.PP
+.B --x86vec=
+\fIoption\fR
+.RS
+Set maximum type of vector use: none, mmx, sse, avx, avx512, default.
+.RE
+.PP
+.B --riscvfloat=
+\fIoption\fR
+.RS
+Set type of RISC-V float support: none, float, double.
+.RE
+.PP
+.B --memory-env=
+\fIoption\fR
+.RS
+Set the memory environment: normal, small, tiny, none.
+.RE
+.PP
+.B --strip-unused=
+\fIyes|no\fR
+.RS
+Strip unused code and globals from the output (default: yes).
+.RE
+.PP
+.B --fp-math=
+\fIoption\fR
+.RS
+Specify floating-point math behavior: strict, relaxed, fast.
+.RE
+.PP
+.B --win64-simd=
+\fIoption\fR
+.RS
+Windows SIMD ABI: array, full.
+.RE
+
+.PP
+.B --debug-stats
+Print debug statistics.
+.PP
+.B --print-linking
+Print linker arguments.
+.PP
+.B --debug-log
+Print debug logging to stdout.
+
+
+.PP
+.B --benchmarking
+Run built-in benchmarks.
+.PP
+.B --testing
+Run built-in tests.
+
+
+.PP
+.B --list-attributes
+List all attributes.
+.PP
+.B --list-builtins
+List all builtins.
+.PP
+.B --list-keywords
+List all keywords.
+.PP
+.B --list-operators
+List all operators.
+.PP
+.B --list-precedence
+List operator precedence order.
+.PP
+.B --list-project-properties
+List all available keys used in project.json files.
+.PP
+.B --list-manifest-properties
+List all available keys used in manifest.json files.
+.PP
+.B --list-targets
+List all architectures the compiler supports.
+.PP
+.B --list-type-properties
+List all type properties.
+
+
+.PP
+.B --print-output
+Print the object files created to stdout.
+.PP
+.B --print-input
+Print inputted C3 files to stdout.
+
+
+.PP
+.B --winsdk 
+\fIdir\fR
+.RS
+Set the directory for Windows system library files for cross-compilation.
+.RE
+.PP
+.B --wincrt=
+\fIoption\fR
+.RS
+Windows CRT linking: none, static-debug, static, dynamic-debug (default if debug info enabled), dynamic (default).
+.RE
+.PP
+.B --windef 
+\fIfile\fR
+.RS
+Use Windows 'def' file for function exports instead of 'dllexport'.
+.RE
+
+.PP
+.B --macossdk 
+\fIdir\fR
+.RS
+Set the directory for the MacOS SDK for cross-compilation.
+.RE
+.PP
+.B --macos-min-version 
+\fIver\fR
+.RS
+Set the minimum MacOS version to compile for.
+.RE
+.PP
+.B --macos-sdk-version 
+\fIver\fR
+.RS
+Set the MacOS SDK version to compile for.
+.RE
+
+.PP
+.B --linux-crt 
+\fIdir\fR
+.RS
+Set the directory to use for finding crt1.o and related files.
+.RE
+.PP
+.B --linux-crtbegin 
+\fIdir\fR
+.RS
+Set the directory to use for finding crtbegin.o and related files.
+.RE
+
+.PP
+.B --vector-conv=
+\fIoption\fR
+.RS
+Set vector conversion behavior: default, old.
+.RE
+.PP
+.B --sanitize=
+\fIoption\fR
+.RS
+Enable a sanitizer: address, memory, thread.
+.RE
+
+.SH EXAMPLES
+.PP
+Create a project:
+.RS
+.B c3c init new_project
+.RE
+.PP
+Create a library project:
+.RS
+.B c3c init-lib new_library
+.RE
+.PP
+Compile a file:
+.RS
+.B c3c compile main.c3
+.RE
+.PP
+Build the current project:
+.RS
+.B c3c build
+.RE
+.PP
+Run tests for the current project:
+.RS
+.B c3c test
+.RE

docker/Dockerfile

@@ -1,16 +1,49 @@
+ARG UBUNTU_VERSION=22.04
+FROM ubuntu:${UBUNTU_VERSION}
 
-ARG UBUNTU_VERSION
-FROM ubuntu:$UBUNTU_VERSION
+ARG LLVM_VERSION=18
+ENV LLVM_DEV_VERSION=20
 
-ARG DEPS
+ARG CMAKE_VERSION=3.20
 
-RUN export DEBIAN_FRONTEND=noninteractive && export TERM=xterm && apt-get update && apt-get install -y build-essential cmake zlib1g zlib1g-dev  \
-        $DEPS && \
-	rm -rf /var/lib/apt/lists/*
+RUN apt-get update && apt-get install -y wget gnupg software-properties-common zlib1g zlib1g-dev python3 ninja-build curl g++ && \
+    wget https://github.com/Kitware/CMake/releases/download/v${CMAKE_VERSION}/cmake-$CMAKE_VERSION-linux-x86_64.sh && \
+    mkdir -p /opt/cmake && \
+    sh cmake-${CMAKE_VERSION}-linux-x86_64.sh --prefix=/opt/cmake --skip-license && \
+    rm cmake-${CMAKE_VERSION}-linux-x86_64.sh && \
+    ln -s /opt/cmake/bin/cmake /usr/local/bin/cmake
 
-ARG GID=1000
-ARG UID=1000
+RUN wget -O - https://apt.llvm.org/llvm-snapshot.gpg.key | apt-key add - && \
+    if [ "${LLVM_VERSION}" -lt 18 ]; then \
+        add-apt-repository "deb http://apt.llvm.org/focal/ llvm-toolchain-focal-${LLVM_VERSION} main" && \
+        apt-get update && \
+        apt-get install -y -t llvm-toolchain-focal-${LLVM_VERSION} \
+          libpolly-${LLVM_VERSION}-dev \
+          clang-${LLVM_VERSION} llvm-${LLVM_VERSION} llvm-${LLVM_VERSION}-dev \
+          lld-${LLVM_VERSION} liblld-${LLVM_VERSION}-dev libmlir-${LLVM_VERSION} \
+          libmlir-${LLVM_VERSION}-dev mlir-${LLVM_VERSION}-tools; \
+    elif [ "${LLVM_VERSION}" -lt "${LLVM_DEV_VERSION}" ]; then \
+        add-apt-repository "deb http://apt.llvm.org/focal/ llvm-toolchain-focal-${LLVM_VERSION} main" && \
+        apt-get update && \
+        apt-get install -y -t llvm-toolchain-focal-${LLVM_VERSION} \
+          libpolly-${LLVM_VERSION}-dev \
+          clang-${LLVM_VERSION} clang++-${LLVM_VERSION} llvm-${LLVM_VERSION} llvm-${LLVM_VERSION}-dev \
+          lld-${LLVM_VERSION} liblld-${LLVM_VERSION}-dev; \
+    else \
+        add-apt-repository "deb http://apt.llvm.org/focal/ llvm-toolchain-focal main" && \
+        apt-get update && \
+        apt-get install -y -t llvm-toolchain-focal \
+          libpolly-${LLVM_VERSION}-dev \
+          clang-${LLVM_VERSION} llvm-${LLVM_VERSION} llvm-${LLVM_VERSION}-dev \
+          lld-${LLVM_VERSION} liblld-${LLVM_VERSION}-dev; \
+    fi && \
+    rm -rf /var/lib/apt/lists/*
 
-RUN groupadd --gid=$GID c3c && useradd --gid=$GID --uid=$GID --create-home --shell /bin/bash c3c
+RUN groupadd -g 1337 c3c && \
+    useradd -m -u 1337 -g c3c c3c
 
+# Add cmake to PATH for user c3c
 USER c3c
+ENV PATH="/opt/cmake/bin:${PATH}"
+
+WORKDIR /home/c3c

flake.lock

@@ -0,0 +1,61 @@
+{
+  "nodes": {
+    "flake-utils": {
+      "inputs": {
+        "systems": "systems"
+      },
+      "locked": {
+        "lastModified": 1731533236,
+        "narHash": "sha256-l0KFg5HjrsfsO/JpG+r7fRrqm12kzFHyUHqHCVpMMbI=",
+        "owner": "numtide",
+        "repo": "flake-utils",
+        "rev": "11707dc2f618dd54ca8739b309ec4fc024de578b",
+        "type": "github"
+      },
+      "original": {
+        "owner": "numtide",
+        "repo": "flake-utils",
+        "type": "github"
+      }
+    },
+    "nixpkgs": {
+      "locked": {
+        "lastModified": 1738297584,
+        "narHash": "sha256-AYvaFBzt8dU0fcSK2jKD0Vg23K2eIRxfsVXIPCW9a0E=",
+        "owner": "nixos",
+        "repo": "nixpkgs",
+        "rev": "9189ac18287c599860e878e905da550aa6dec1cd",
+        "type": "github"
+      },
+      "original": {
+        "owner": "nixos",
+        "ref": "nixpkgs-unstable",
+        "repo": "nixpkgs",
+        "type": "github"
+      }
+    },
+    "root": {
+      "inputs": {
+        "flake-utils": "flake-utils",
+        "nixpkgs": "nixpkgs"
+      }
+    },
+    "systems": {
+      "locked": {
+        "lastModified": 1681028828,
+        "narHash": "sha256-Vy1rq5AaRuLzOxct8nz4T6wlgyUR7zLU309k9mBC768=",
+        "owner": "nix-systems",
+        "repo": "default",
+        "rev": "da67096a3b9bf56a91d16901293e51ba5b49a27e",
+        "type": "github"
+      },
+      "original": {
+        "owner": "nix-systems",
+        "repo": "default",
+        "type": "github"
+      }
+    }
+  },
+  "root": "root",
+  "version": 7
+}

flake.nix

@@ -0,0 +1,44 @@
+{
+  description = "C3 compiler flake";
+
+  inputs = {
+    nixpkgs.url = "github:nixos/nixpkgs?ref=nixpkgs-unstable";
+    flake-utils.url = "github:numtide/flake-utils";
+  };
+
+  outputs = { self, ... } @ inputs: inputs.flake-utils.lib.eachDefaultSystem 
+  (system: 
+    let pkgs = import inputs.nixpkgs { inherit system; }; 
+        call = set: pkgs.callPackage ./nix/default.nix ( 
+          set // { 
+            rev = self.rev or "unknown"; 
+          } 
+        );
+    in {
+      packages = {
+        default = self.packages.${system}.c3c;
+
+        c3c = call {};
+        
+        c3c-checks = pkgs.callPackage ./nix/default.nix { 
+          checks = true; 
+        };
+
+        c3c-debug = pkgs.callPackage ./nix/default.nix { 
+          debug = true; 
+        };
+
+        c3c-debug-checks = pkgs.callPackage ./nix/default.nix { 
+          debug = true; 
+          checks = true; 
+        };
+      };
+
+      devShells = {
+        default = pkgs.callPackage ./nix/shell.nix {
+          c3c = self.packages.${system}.c3c-debug; 
+        };
+      };
+    }
+  );
+}

git_hash.cmake

@@ -0,0 +1,15 @@
+find_package(Git QUIET)
+
+set(GIT_HASH "unknown")
+
+if(GIT_FOUND AND EXISTS "${CMAKE_CURRENT_LIST_DIR}/.git")
+    execute_process(COMMAND ${GIT_EXECUTABLE} rev-parse HEAD
+                    WORKING_DIRECTORY ${CMAKE_CURRENT_LIST_DIR}
+                    OUTPUT_VARIABLE GIT_HASH
+                    OUTPUT_STRIP_TRAILING_WHITESPACE
+                    COMMAND_ERROR_IS_FATAL ANY)
+endif()
+
+message("Git Hash: ${GIT_HASH}")
+
+file(WRITE ${CMAKE_BINARY_DIR}/git_hash.h "#pragma once\n#define GIT_HASH \"${GIT_HASH}\"\n")

lib/std/array.c3

@@ -1,19 +0,0 @@
-// Copyright (c) 2021 Christoffer Lerno. All rights reserved.
-// Use of this source code is governed by the MIT license
-// a copy of which can be found in the LICENSE_STDLIB file.
-module std::array;
-import std::mem;
-
-macro make($Type, usize elements)
-{
-    assert(elements > 0);
-    $Type* ptr = mem::alloc($sizeof($Type), elements);
-    return ptr[0..(elements - 1)];
-}
-
-macro make_zero($Type, usize elements)
-{
-    assert(elements > 0);
-    $Type* ptr = mem::calloc($sizeof($Type), elements);
-    return ptr[0..(elements - 1)];
-}

lib/std/ascii.c3

@@ -0,0 +1,77 @@
+module std::ascii;
+
+macro bool in_range_m(c, start, len) => (uint)(c - start) < len;
+macro bool is_lower_m(c) => in_range_m(c, 0x61, 26);
+macro bool is_upper_m(c) => in_range_m(c, 0x41, 26);
+macro bool is_digit_m(c) => in_range_m(c, 0x30, 10);
+macro bool is_bdigit_m(c) => in_range_m(c, 0x30, 2);
+macro bool is_odigit_m(c) => in_range_m(c, 0x30, 8);
+macro bool is_xdigit_m(c) => in_range_m(c | 32, 0x61, 6) || is_digit_m(c);
+macro bool is_alpha_m(c) => in_range_m(c | 32, 0x61, 26);
+macro bool is_print_m(c) => in_range_m(c, 0x20, 95);
+macro bool is_graph_m(c) => in_range_m(c, 0x21, 94);
+macro bool is_space_m(c) => in_range_m(c, 0x9, 5) || c == 0x20;
+macro bool is_alnum_m(c) => is_alpha_m(c) || is_digit_m(c);
+macro bool is_punct_m(c) => !is_alnum_m(c) && is_graph_m(c);
+macro bool is_blank_m(c) => c == 0x20 || c == 0x9;
+macro bool is_cntrl_m(c) => c < 0x20 || c == 0x7f;
+macro to_lower_m(c) => is_upper_m(c) ? c + 0x20 : c;
+macro to_upper_m(c) => is_lower_m(c) ? c - 0x20 : c;
+
+fn bool in_range(char c, char start, char len) => in_range_m(c, start, len);
+fn bool is_lower(char c) => is_lower_m(c);
+fn bool is_upper(char c) => is_upper_m(c);
+fn bool is_digit(char c) => is_digit_m(c);
+fn bool is_bdigit(char c) => is_bdigit_m(c);
+fn bool is_odigit(char c) => is_odigit_m(c);
+fn bool is_xdigit(char c) => is_xdigit_m(c);
+fn bool is_alpha(char c) => is_alpha_m(c);
+fn bool is_print(char c) => is_print_m(c);
+fn bool is_graph(char c) => is_graph_m(c);
+fn bool is_space(char c) => is_space_m(c);
+fn bool is_alnum(char c) => is_alnum_m(c);
+fn bool is_punct(char c) => is_punct_m(c);
+fn bool is_blank(char c) => is_blank_m(c);
+fn bool is_cntrl(char c) => is_cntrl_m(c);
+fn char to_lower(char c) => (char)to_lower_m(c);
+fn char to_upper(char c) => (char)to_upper_m(c);
+
+fn bool char.in_range(char c, char start, char len) => in_range_m(c, start, len);
+fn bool char.is_lower(char c) => is_lower_m(c);
+fn bool char.is_upper(char c) => is_upper_m(c);
+fn bool char.is_digit(char c) => is_digit_m(c);
+fn bool char.is_bdigit(char c) => is_bdigit_m(c);
+fn bool char.is_odigit(char c) => is_odigit_m(c);
+fn bool char.is_xdigit(char c) => is_xdigit_m(c);
+fn bool char.is_alpha(char c) => is_alpha_m(c);
+fn bool char.is_print(char c) => is_print_m(c);
+fn bool char.is_graph(char c) => is_graph_m(c);
+fn bool char.is_space(char c) => is_space_m(c);
+fn bool char.is_alnum(char c) => is_alnum_m(c);
+fn bool char.is_punct(char c) => is_punct_m(c);
+fn bool char.is_blank(char c) => is_blank_m(c);
+fn bool char.is_cntrl(char c) => is_cntrl_m(c);
+fn char char.to_lower(char c) => (char)to_lower_m(c);
+fn char char.to_upper(char c) => (char)to_upper_m(c);
+<*
+ @require c.is_xdigit()
+*>
+fn char char.from_hex(char c) => c.is_digit() ? c - '0' : 10 + (c | 0x20) - 'a';
+
+fn bool uint.in_range(uint c, uint start, uint len) => in_range_m(c, start, len);
+fn bool uint.is_lower(uint c) => is_lower_m(c);
+fn bool uint.is_upper(uint c) => is_upper_m(c);
+fn bool uint.is_digit(uint c) => is_digit_m(c);
+fn bool uint.is_bdigit(uint c) => is_bdigit_m(c);
+fn bool uint.is_odigit(uint c) => is_odigit_m(c);
+fn bool uint.is_xdigit(uint c) => is_xdigit_m(c);
+fn bool uint.is_alpha(uint c) => is_alpha_m(c);
+fn bool uint.is_print(uint c) => is_print_m(c);
+fn bool uint.is_graph(uint c) => is_graph_m(c);
+fn bool uint.is_space(uint c) => is_space_m(c);
+fn bool uint.is_alnum(uint c) => is_alnum_m(c);
+fn bool uint.is_punct(uint c) => is_punct_m(c);
+fn bool uint.is_blank(uint c) => is_blank_m(c);
+fn bool uint.is_cntrl(uint c) => is_cntrl_m(c);
+fn uint uint.to_lower(uint c) => (uint)to_lower_m(c);
+fn uint uint.to_upper(uint c) => (uint)to_upper_m(c);

lib/std/atomic.c3

@@ -1,5 +1,520 @@
-// Copyright (c) 2021 Christoffer Lerno. All rights reserved.
+// Copyright (c) 2023 Eduardo José Gómez Hernández. All rights reserved.
 // Use of this source code is governed by the MIT license
 // a copy of which can be found in the LICENSE_STDLIB file.
-module std::atomic;
+module std::atomic::types(<Type>);
 
+struct Atomic
+{
+	Type data;
+}
+
+<*
+ Loads data atomically, by default this uses SEQ_CONSISTENT ordering.
+
+ @param ordering "The ordering, cannot be release or acquire-release."
+ @require ordering != RELEASE && ordering != ACQUIRE_RELEASE : "Release and acquire-release are not valid for load"
+*>
+macro Type Atomic.load(&self, AtomicOrdering ordering = SEQ_CONSISTENT)
+{
+	Type* data = &self.data;
+	switch(ordering)
+	{
+		case NOT_ATOMIC: return $$atomic_load(data, false, AtomicOrdering.NOT_ATOMIC.ordinal);
+		case UNORDERED: return $$atomic_load(data, false, AtomicOrdering.UNORDERED.ordinal);
+		case RELAXED: return $$atomic_load(data, false, AtomicOrdering.RELAXED.ordinal);
+		case ACQUIRE: return $$atomic_load(data, false, AtomicOrdering.ACQUIRE.ordinal);
+		case SEQ_CONSISTENT: return $$atomic_load(data, false, AtomicOrdering.SEQ_CONSISTENT.ordinal);
+		case ACQUIRE_RELEASE:
+		case RELEASE: unreachable("Invalid ordering.");
+	}
+}
+<*
+ Stores data atomically, by default this uses SEQ_CONSISTENT ordering.
+
+ @param ordering "The ordering, cannot be acquire or acquire-release."
+ @require ordering != ACQUIRE && ordering != ACQUIRE_RELEASE : "Acquire and acquire-release are not valid for store"
+*>
+macro void Atomic.store(&self, Type value, AtomicOrdering ordering = SEQ_CONSISTENT)
+{
+	Type* data = &self.data;
+	switch(ordering) 
+	{
+		case NOT_ATOMIC: $$atomic_store(data, value, false, AtomicOrdering.NOT_ATOMIC.ordinal);
+		case UNORDERED: $$atomic_store(data, value, false, AtomicOrdering.UNORDERED.ordinal);
+		case RELAXED: $$atomic_store(data, value, false, AtomicOrdering.RELAXED.ordinal);
+		case RELEASE: $$atomic_store(data, value, false, AtomicOrdering.RELEASE.ordinal);
+		case SEQ_CONSISTENT: $$atomic_store(data, value, false, AtomicOrdering.SEQ_CONSISTENT.ordinal);
+		case ACQUIRE_RELEASE:
+		case ACQUIRE: unreachable("Invalid ordering.");
+	}
+}
+
+macro Type Atomic.add(&self, Type value, AtomicOrdering ordering = SEQ_CONSISTENT)
+{
+	Type* data = &self.data;
+	return @atomic_exec(atomic::fetch_add, data, value, ordering);
+}
+
+macro Type Atomic.sub(&self, Type value, AtomicOrdering ordering = SEQ_CONSISTENT)
+{
+	Type* data = &self.data;
+	return @atomic_exec(atomic::fetch_sub, data, value, ordering);
+}
+
+macro Type Atomic.mul(&self, Type value, AtomicOrdering ordering = SEQ_CONSISTENT)
+{
+	Type* data = &self.data;
+	return @atomic_exec(atomic::fetch_mul, data, value, ordering);
+}
+
+macro Type Atomic.div(&self, Type value, AtomicOrdering ordering = SEQ_CONSISTENT)
+{
+	Type* data = &self.data;
+	return @atomic_exec(atomic::fetch_div, data, value, ordering);
+}
+
+macro Type Atomic.max(&self, Type value, AtomicOrdering ordering = SEQ_CONSISTENT)
+{
+	Type* data = &self.data;
+	return @atomic_exec(atomic::fetch_div, data, value, ordering);
+}
+
+macro Type Atomic.min(&self, Type value, AtomicOrdering ordering = SEQ_CONSISTENT)
+{
+	Type* data = &self.data;
+	return @atomic_exec(atomic::fetch_min, data, value, ordering);
+}
+
+macro Type Atomic.or(&self, uint value, AtomicOrdering ordering = SEQ_CONSISTENT) @if(!types::is_float(Type))
+{
+	Type* data = &self.data;
+	return @atomic_exec(atomic::fetch_or, data, value, ordering);
+}
+
+fn Type Atomic.xor(&self, uint value, AtomicOrdering ordering = SEQ_CONSISTENT) @if(!types::is_float(Type))
+{
+	Type* data = &self.data;
+	return @atomic_exec(atomic::fetch_xor, data, value, ordering);
+}
+
+macro Type Atomic.and(&self, uint value, AtomicOrdering ordering = SEQ_CONSISTENT) @if(!types::is_float(Type))
+{
+	Type* data = &self.data;
+	return @atomic_exec(atomic::fetch_and, data, value, ordering);
+}
+
+macro Type Atomic.shift_right(&self, uint amount, AtomicOrdering ordering = SEQ_CONSISTENT) @if(!types::is_float(Type))
+{
+	Type* data = &self.data;
+	return @atomic_exec(atomic::fetch_shift_right, data, amount, ordering);
+}
+
+macro Type Atomic.shift_left(&self, uint amount, AtomicOrdering ordering = SEQ_CONSISTENT) @if(!types::is_float(Type))
+{
+	Type* data = &self.data;
+	return @atomic_exec(atomic::fetch_shift_left, data, amount, ordering);
+}
+
+macro @atomic_exec(#func, data, value, ordering) @local
+{
+	switch(ordering) 
+	{
+		case RELAXED: return #func(data, value, RELAXED);
+		case ACQUIRE: return #func(data, value, ACQUIRE);
+		case RELEASE: return #func(data, value, RELEASE);
+		case ACQUIRE_RELEASE: return #func(data, value, ACQUIRE_RELEASE);
+		case SEQ_CONSISTENT: return #func(data, value, SEQ_CONSISTENT);
+		default: unreachable("Ordering may not be non-atomic or unordered.");
+	}
+}
+
+module std::atomic;
+import std::math;
+
+<*
+ @param [&in] ptr "the variable or dereferenced pointer to the data."
+ @param [in] y "the value to be added to ptr."
+ @param $ordering "atomic ordering of the load, defaults to SEQ_CONSISTENT"
+ @return "returns the old value of ptr"
+
+ @require !$alignment || math::is_power_of_2($alignment) "Alignment must be a power of two."
+ @require types::is_int($typeof(*ptr)) || types::is_float($typeof(*ptr)) "Only integer/float pointers may be used."
+ @require $ordering != AtomicOrdering.NOT_ATOMIC && $ordering != AtomicOrdering.UNORDERED "Acquire ordering is not valid."
+*>
+macro fetch_add(ptr, y, AtomicOrdering $ordering = SEQ_CONSISTENT, bool $volatile = false, usz $alignment = 0)
+{
+	$if $alignment == 0:
+		$alignment = $typeof(*ptr).sizeof;
+	$endif
+	return $$atomic_fetch_add(ptr, y, $volatile, $ordering.ordinal, $alignment);
+}
+
+<*
+ @param [&in] ptr "the variable or dereferenced pointer to the data."
+ @param [in] y "the value to be added to ptr."
+ @param $ordering "atomic ordering of the load, defaults to SEQ_CONSISTENT"
+ @return "returns the old value of ptr"
+
+ @require !$alignment || math::is_power_of_2($alignment) "Alignment must be a power of two."
+ @require types::is_int($typeof(*ptr)) || types::is_float($typeof(*ptr)) "Only integer/float pointers may be used."
+ @require $ordering != AtomicOrdering.NOT_ATOMIC && $ordering != AtomicOrdering.UNORDERED "Acquire ordering is not valid."
+*>
+macro fetch_sub(ptr, y, AtomicOrdering $ordering = SEQ_CONSISTENT, bool $volatile = false, usz $alignment = 0)
+{
+	$if $alignment == 0:
+		$alignment = $typeof(*ptr).sizeof;
+	$endif
+	return $$atomic_fetch_sub(ptr, y, $volatile, $ordering.ordinal, $alignment);
+}
+
+<*
+ @param [&in] ptr "the variable or dereferenced pointer to the data."
+ @param [in] y "the value to be added to ptr."
+ @param $ordering "atomic ordering of the load, defaults to SEQ_CONSISTENT"
+ @return "returns the old value of ptr"
+
+ @require types::is_int($typeof(*ptr)) || types::is_float($typeof(*ptr)) "Only integer/float pointers may be used."
+ @require $ordering != AtomicOrdering.NOT_ATOMIC && $ordering != AtomicOrdering.UNORDERED "Acquire ordering is not valid."
+*>
+macro fetch_mul(ptr, y, AtomicOrdering $ordering = SEQ_CONSISTENT)
+{
+	var $load_ordering = $ordering;
+	$if $ordering == RELEASE || $ordering == ACQUIRE_RELEASE:
+	    $load_ordering = AtomicOrdering.SEQ_CONSISTENT;
+	$endif
+	
+	var $StorageType = $typefrom(types::lower_to_atomic_compatible_type($typeof(*ptr)));
+
+	$StorageType* storage_ptr = ($StorageType*)ptr;
+
+	$typeof(*ptr) old_value;
+	$typeof(*ptr) new_value;
+
+	$StorageType storage_old_value;
+	$StorageType storage_new_value;
+
+
+	do {
+		storage_old_value = $$atomic_load(storage_ptr, false, $load_ordering.ordinal);
+		old_value = bitcast(storage_old_value, $typeof(*ptr));
+		new_value = old_value * y;
+		storage_new_value = bitcast(new_value, $StorageType);
+	} while (mem::compare_exchange(storage_ptr, storage_old_value, storage_new_value, $ordering, $load_ordering) != storage_old_value);
+
+	return old_value;
+}
+
+<*
+ @param [&in] ptr "the variable or dereferenced pointer to the data."
+ @param [in] y "the value to be added to ptr."
+ @param $ordering "atomic ordering of the load, defaults to SEQ_CONSISTENT"
+ @return "returns the old value of ptr"
+
+ @require types::is_int($typeof(*ptr)) || types::is_float($typeof(*ptr)) "Only integer/float pointers may be used."
+ @require $ordering != AtomicOrdering.NOT_ATOMIC && $ordering != AtomicOrdering.UNORDERED "Acquire ordering is not valid."
+*>
+macro fetch_div(ptr, y, AtomicOrdering $ordering = SEQ_CONSISTENT)
+{
+	var $load_ordering = $ordering;
+	$if $ordering == RELEASE || $ordering == ACQUIRE_RELEASE:
+	    $load_ordering = AtomicOrdering.SEQ_CONSISTENT;
+	$endif
+	
+	var $StorageType = $typefrom(types::lower_to_atomic_compatible_type($typeof(*ptr)));
+
+	$StorageType* storage_ptr = ($StorageType*)ptr;
+
+	$typeof(*ptr) old_value;
+	$typeof(*ptr) new_value;
+
+	$StorageType storage_old_value;
+	$StorageType storage_new_value;
+	
+	do {
+		storage_old_value = $$atomic_load(storage_ptr, false, $load_ordering.ordinal);
+		old_value = bitcast(storage_old_value, $typeof(*ptr));
+		new_value = old_value / y;
+		storage_new_value = bitcast(new_value, $StorageType);
+	} while (mem::compare_exchange(storage_ptr, storage_old_value, storage_new_value, $ordering, $load_ordering) != storage_old_value);
+
+	return old_value;
+}
+
+<*
+ @param [&in] ptr "the variable or dereferenced pointer to the data."
+ @param [in] y "the value to be added to ptr."
+ @param $ordering "atomic ordering of the load, defaults to SEQ_CONSISTENT"
+ @return "returns the old value of ptr"
+
+ @require !$alignment || math::is_power_of_2($alignment) "Alignment must be a power of two."
+ @require types::is_int($typeof(*ptr)) "Only integer pointers may be used."
+ @require types::is_int($typeof(y)) "The value for or must be an int"
+ @require $ordering != AtomicOrdering.NOT_ATOMIC && $ordering != AtomicOrdering.UNORDERED "Acquire ordering is not valid."
+*>
+macro fetch_or(ptr, y, AtomicOrdering $ordering = SEQ_CONSISTENT, bool $volatile = false, usz $alignment = 0)
+{
+	$if	types::is_int($typeof(*ptr)):
+		return $$atomic_fetch_or(ptr, y, $volatile, $ordering.ordinal, $alignment);
+	$endif
+
+	var $load_ordering = $ordering;
+	$if $ordering == RELEASE || $ordering == ACQUIRE_RELEASE:
+	    $load_ordering = AtomicOrdering.SEQ_CONSISTENT;
+	$endif
+	
+	var $StorageType = $typefrom(types::lower_to_atomic_compatible_type($typeof(*ptr)));
+
+	$StorageType* storage_ptr = ($StorageType*)ptr;
+
+	$typeof(*ptr) old_value;
+	$typeof(*ptr) new_value;
+
+	$StorageType storage_old_value;
+	$StorageType storage_new_value;
+	$StorageType storage_y = ($StorageType)y;
+	
+	do {
+		storage_old_value = $$atomic_load(storage_ptr, false, $load_ordering.ordinal);
+		old_value = bitcast(storage_old_value, $typeof(*ptr));
+		new_value = storage_old_value | storage_y;
+		storage_new_value = bitcast(new_value, $StorageType);
+	} while (mem::compare_exchange(storage_ptr, storage_old_value, storage_new_value, $ordering, $load_ordering) != storage_old_value);
+
+	return old_value;
+}
+
+<*
+ @param [&in] ptr "the variable or dereferenced pointer to the data."
+ @param [in] y "the value to be added to ptr."
+ @param $ordering "atomic ordering of the load, defaults to SEQ_CONSISTENT"
+ @return "returns the old value of ptr"
+
+ @require !$alignment || math::is_power_of_2($alignment) "Alignment must be a power of two."
+ @require types::is_int($typeof(*ptr)) "Only integer pointers may be used."
+ @require types::is_int($typeof(y)) "The value for or must be an int"
+ @require $ordering != AtomicOrdering.NOT_ATOMIC && $ordering != AtomicOrdering.UNORDERED "Acquire ordering is not valid."
+*>
+macro fetch_xor(ptr, y, AtomicOrdering $ordering = SEQ_CONSISTENT, bool $volatile = false, usz $alignment = 0)
+{
+	$if	types::is_int($typeof(*ptr)):
+		return $$atomic_fetch_xor(ptr, y, $volatile, $ordering.ordinal, $alignment);
+	$endif
+
+	var $load_ordering = $ordering;
+	$if $ordering == RELEASE || $ordering == ACQUIRE_RELEASE:
+	    $load_ordering = AtomicOrdering.SEQ_CONSISTENT;
+	$endif
+	
+	var $StorageType = $typefrom(types::lower_to_atomic_compatible_type($typeof(*ptr)));
+
+	$StorageType* storage_ptr = ($StorageType*)ptr;
+
+	$typeof(*ptr) old_value;
+	$typeof(*ptr) new_value;
+
+	$StorageType storage_old_value;
+	$StorageType storage_new_value;
+	$StorageType storage_y = ($StorageType)y;
+	
+	do {
+		storage_old_value = $$atomic_load(storage_ptr, false, $load_ordering.ordinal);
+		old_value = bitcast(storage_old_value, $typeof(*ptr));
+		new_value = storage_old_value ^ storage_y;
+		storage_new_value = bitcast(new_value, $StorageType);
+	} while (mem::compare_exchange(storage_ptr, storage_old_value, storage_new_value, $ordering, $load_ordering) != storage_old_value);
+
+	return old_value;
+}
+
+<*
+ @param [&in] ptr "the variable or dereferenced pointer to the data."
+ @param [in] y "the value to be added to ptr."
+ @param $ordering "atomic ordering of the load, defaults to SEQ_CONSISTENT"
+ @return "returns the old value of ptr"
+
+ @require !$alignment || math::is_power_of_2($alignment) "Alignment must be a power of two."
+ @require types::is_int($typeof(*ptr)) "Only integer pointers may be used."
+ @require types::is_int($typeof(y)) "The value for or must be an int"
+ @require $ordering != AtomicOrdering.NOT_ATOMIC && $ordering != AtomicOrdering.UNORDERED "Acquire ordering is not valid."
+*>
+macro fetch_and(ptr, y, AtomicOrdering $ordering = SEQ_CONSISTENT, bool $volatile = false, usz $alignment = 0)
+{
+	$if	types::is_int($typeof(*ptr)):
+		return $$atomic_fetch_and(ptr, y, $volatile, $ordering.ordinal, $alignment);
+	$endif
+
+	var $load_ordering = $ordering;
+	$if $ordering == RELEASE || $ordering == ACQUIRE_RELEASE:
+	    $load_ordering = AtomicOrdering.SEQ_CONSISTENT;
+	$endif
+	
+	var $StorageType = $typefrom(types::lower_to_atomic_compatible_type($typeof(*ptr)));
+
+	$StorageType* storage_ptr = ($StorageType*)ptr;
+
+	$typeof(*ptr) old_value;
+	$typeof(*ptr) new_value;
+
+	$StorageType storage_old_value;
+	$StorageType storage_new_value;
+	$StorageType storage_y = ($StorageType)y;
+	
+	do {
+		storage_old_value = $$atomic_load(storage_ptr, false, $load_ordering.ordinal);
+		old_value = bitcast(storage_old_value, $typeof(*ptr));
+		new_value = storage_old_value & storage_y;
+		storage_new_value = bitcast(new_value, $StorageType);
+	} while (mem::compare_exchange(storage_ptr, storage_old_value, storage_new_value, $ordering, $load_ordering) != storage_old_value);
+
+	return old_value;
+}
+
+<*
+ @param [&in] ptr "the variable or dereferenced pointer to the data."
+ @param [in] y "the value to be added to ptr."
+ @param $ordering "atomic ordering of the load, defaults to SEQ_CONSISTENT"
+ @return "returns the old value of ptr"
+
+ @require types::is_int($typeof(*ptr)) "Only integer pointers may be used."
+ @require types::is_int($typeof(y)) "The value for or must be an int"
+ @require $ordering != AtomicOrdering.NOT_ATOMIC && $ordering != AtomicOrdering.UNORDERED "Acquire ordering is not valid."
+*>
+macro fetch_shift_right(ptr, y, AtomicOrdering $ordering = SEQ_CONSISTENT)
+{
+	var $load_ordering = $ordering;
+	$if $ordering == RELEASE || $ordering == ACQUIRE_RELEASE:
+	    $load_ordering = AtomicOrdering.SEQ_CONSISTENT;
+	$endif
+	
+	var $StorageType = $typefrom(types::lower_to_atomic_compatible_type($typeof(*ptr)));
+
+	$StorageType* storage_ptr = ($StorageType*)ptr;
+
+	$typeof(*ptr) old_value;
+	$typeof(*ptr) new_value;
+
+	$StorageType storage_old_value;
+	$StorageType storage_new_value;
+	$StorageType storage_y = ($StorageType)y;
+	
+	do {
+		storage_old_value = $$atomic_load(storage_ptr, false, $load_ordering.ordinal);
+		old_value = bitcast(storage_old_value, $typeof(*ptr));
+		new_value = storage_old_value >> storage_y;
+		storage_new_value = bitcast(new_value, $StorageType);
+	} while (mem::compare_exchange(storage_ptr, storage_old_value, storage_new_value, $ordering, $load_ordering) != storage_old_value);
+
+	return old_value;
+}
+
+<*
+ @param [&in] ptr "the variable or dereferenced pointer to the data."
+ @param [in] y "the value to be added to ptr."
+ @param $ordering "atomic ordering of the load, defaults to SEQ_CONSISTENT"
+ @return "returns the old value of ptr"
+
+ @require types::is_int($typeof(*ptr)) "Only integer pointers may be used."
+ @require types::is_int($typeof(y)) "The value for or must be an int"
+ @require $ordering != AtomicOrdering.NOT_ATOMIC && $ordering != AtomicOrdering.UNORDERED "Acquire ordering is not valid."
+*>
+macro fetch_shift_left(ptr, y, AtomicOrdering $ordering = SEQ_CONSISTENT)
+{
+	var $load_ordering = $ordering;
+	$if $ordering == RELEASE || $ordering == ACQUIRE_RELEASE:
+	    $load_ordering = AtomicOrdering.SEQ_CONSISTENT;
+	$endif
+	
+	var $StorageType = $typefrom(types::lower_to_atomic_compatible_type($typeof(*ptr)));
+
+	$StorageType* storage_ptr = ($StorageType*)ptr;
+
+	$typeof(*ptr) old_value;
+	$typeof(*ptr) new_value;
+
+	$StorageType storage_old_value;
+	$StorageType storage_new_value;
+	$StorageType storage_y = ($StorageType)y;
+	
+	do {
+		storage_old_value = $$atomic_load(storage_ptr, false, $load_ordering.ordinal);
+		old_value = bitcast(storage_old_value, $typeof(*ptr));
+		new_value = storage_old_value << storage_y;
+		storage_new_value = bitcast(new_value, $StorageType);
+	} while (mem::compare_exchange(storage_ptr, storage_old_value, storage_new_value, $ordering, $load_ordering) != storage_old_value);
+
+	return old_value;
+}
+
+<*
+ @param [&in] ptr "the variable or dereferenced pointer to the data."
+ @param $ordering "atomic ordering of the load, defaults to SEQ_CONSISTENT"
+ @return "returns the old value of ptr"
+
+ @require types::is_int($typeof(*ptr)) "Only integer pointers may be used."
+ @require $ordering != AtomicOrdering.NOT_ATOMIC && $ordering != AtomicOrdering.UNORDERED "Acquire ordering is not valid."
+*>
+macro flag_set(ptr, AtomicOrdering $ordering = SEQ_CONSISTENT)
+{
+	$typeof(*ptr) old_value;
+	$typeof(*ptr) new_value = true;
+
+	do {
+		old_value = $$atomic_load(ptr, false, $ordering.ordinal);
+	} while (mem::compare_exchange(ptr, old_value, new_value, $ordering, $load_ordering) != old_value);
+	
+	return old_value;
+}
+
+<*
+ @param [&in] ptr "the variable or dereferenced pointer to the data."
+ @param $ordering "atomic ordering of the load, defaults to SEQ_CONSISTENT"
+ @return "returns the old value of ptr"
+
+ @require types::is_int($typeof(*ptr)) "Only integer pointers may be used."
+ @require $ordering != AtomicOrdering.NOT_ATOMIC && $ordering != AtomicOrdering.UNORDERED "Acquire ordering is not valid."
+*>
+macro flag_clear(ptr, AtomicOrdering $ordering = SEQ_CONSISTENT)
+{
+	$typeof(*ptr) old_value;
+	$typeof(*ptr) new_value = false;
+
+	do {
+		old_value = $$atomic_load(ptr, false, $ordering.ordinal);
+	} while (mem::compare_exchange(ptr, old_value, new_value, $ordering, $load_ordering) != old_value);
+
+	return old_value;
+}
+
+<*
+ @param [&in] ptr "the variable or dereferenced pointer to the data."
+ @param [in] y "the value to be added to ptr."
+ @param $ordering "atomic ordering of the load, defaults to SEQ_CONSISTENT"
+ @return "returns the old value of ptr"
+
+ @require types::is_int($typeof(*ptr)) || types::is_float($typeof(*ptr)) "Only integer/float pointers may be used."
+ @require $ordering != AtomicOrdering.NOT_ATOMIC && $ordering != AtomicOrdering.UNORDERED "Acquire ordering is not valid."
+*>
+macro fetch_max(ptr, y, AtomicOrdering $ordering = SEQ_CONSISTENT, bool $volatile = false, usz $alignment = 0)
+{
+	$if $alignment == 0:
+		$alignment = $typeof(*ptr).sizeof;
+	$endif
+	return $$atomic_fetch_max(ptr, y, $volatile, $ordering.ordinal, $alignment);
+}
+
+<*
+ @param [&in] ptr "the variable or dereferenced pointer to the data."
+ @param [in] y "the value to be added to ptr."
+ @param $ordering "atomic ordering of the load, defaults to SEQ_CONSISTENT"
+ @return "returns the old value of ptr"
+
+ @require types::is_int($typeof(*ptr)) || types::is_float($typeof(*ptr)) "Only integer/float pointers may be used."
+ @require $ordering != AtomicOrdering.NOT_ATOMIC && $ordering != AtomicOrdering.UNORDERED "Acquire ordering is not valid."
+*>
+macro fetch_min(ptr, y, AtomicOrdering $ordering = SEQ_CONSISTENT, bool $volatile = false, usz $alignment = 0)
+{
+	$if $alignment == 0:
+		$alignment = $typeof(*ptr).sizeof;
+	$endif
+	return $$atomic_fetch_min(ptr, y, $volatile, $ordering.ordinal, $alignment);
+}

lib/std/atomic_nolibc.c3

@@ -0,0 +1,63 @@
+// Copyright (c) 2023 Eduardo José Gómez Hernández. All rights reserved.
+// Use of this source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+module std::atomic;
+
+macro @__atomic_compare_exchange_ordering_failure(ptr, expected, desired, $success, failure, $alignment) {
+	switch(failure)
+	{
+		case AtomicOrdering.RELAXED.ordinal: return $$compare_exchange(ptr, expected, desired, false, false, $success, AtomicOrdering.RELAXED.ordinal, $alignment);
+		case AtomicOrdering.ACQUIRE.ordinal: return $$compare_exchange(ptr, expected, desired, false, false, $success, AtomicOrdering.ACQUIRE.ordinal, $alignment);
+		case AtomicOrdering.SEQ_CONSISTENT.ordinal: return $$compare_exchange(ptr, expected, desired, false, false, $success, AtomicOrdering.SEQ_CONSISTENT.ordinal, $alignment);
+		default: unreachable("Unrecognized failure ordering");
+	}
+	return 0;
+}
+
+macro @__atomic_compare_exchange_ordering_success(ptr, expected, desired, success, failure, $alignment)
+{
+	switch(success)
+	{
+		case AtomicOrdering.RELAXED.ordinal: return @__atomic_compare_exchange_ordering_failure(ptr, expected, desired, AtomicOrdering.RELAXED.ordinal, failure, $alignment);
+		case AtomicOrdering.ACQUIRE.ordinal: return @__atomic_compare_exchange_ordering_failure(ptr, expected, desired, AtomicOrdering.ACQUIRE.ordinal, failure, $alignment);
+		case AtomicOrdering.RELEASE.ordinal: return @__atomic_compare_exchange_ordering_failure(ptr, expected, desired, AtomicOrdering.RELEASE.ordinal, failure, $alignment);
+		case AtomicOrdering.ACQUIRE_RELEASE.ordinal: return @__atomic_compare_exchange_ordering_failure(ptr, expected, desired, AtomicOrdering.ACQUIRE_RELEASE.ordinal, failure, $alignment);
+		case AtomicOrdering.SEQ_CONSISTENT.ordinal: return @__atomic_compare_exchange_ordering_failure(ptr, expected, desired, AtomicOrdering.SEQ_CONSISTENT.ordinal, failure, $alignment);
+		default: unreachable("Unrecognized success ordering");
+	}
+	return 0;
+}
+
+fn CInt __atomic_compare_exchange(CInt size, any ptr, any expected, any desired, CInt success, CInt failure) @extern("__atomic_compare_exchange") @export
+{
+	switch (size)
+	{
+		case 1:
+			char* pt = (char*)ptr;
+			char ex = *(char*)expected;
+			char de = *(char*)desired;
+			if (ex == @__atomic_compare_exchange_ordering_success(pt, ex, de, success, failure, 1)) return 1;
+		case 2:
+			short* pt = (short*)ptr;
+			short ex = *(short*)expected;
+			short de = *(short*)desired;
+			if (ex == @__atomic_compare_exchange_ordering_success(pt, ex, de, success, failure, 2)) return 1;
+		case 4:
+			int* pt = (int*)ptr;
+			int ex = *(int*)expected;
+			int de = *(int*)desired;
+			if (ex == @__atomic_compare_exchange_ordering_success(pt, ex, de, success, failure, 4)) return 1;
+		case 8:
+			$if iptr.sizeof >= 8:
+				long* pt = (long*)ptr;
+				long ex = *(long*)expected;
+				long de = *(long*)desired;
+				if (ex == @__atomic_compare_exchange_ordering_success(pt, ex, de, success, failure, 8)) return 1;
+			$else
+				nextcase;
+			$endif
+		default:
+			unreachable("Unsuported size (%d) for atomic_compare_exchange", size);
+	}
+	return 0;
+}

lib/std/bits.c3

@@ -0,0 +1,171 @@
+module std::bits;
+
+<*
+ @require types::is_intlike($typeof(i)) `The input must be an integer or integer vector`
+*>
+macro reverse(i) => $$bitreverse(i);
+
+<*
+ @require types::is_intlike($typeof(i)) `The input must be an integer or integer vector`
+*>
+macro bswap(i) @builtin => $$bswap(i);
+
+macro uint[<?>].popcount(self) => $$popcount(self);
+macro uint[<?>].ctz(self) => $$ctz(self);
+macro uint[<?>].clz(self) => $$clz(self);
+macro uint[<?>] uint[<?>].fshl(hi, uint[<?>] lo, uint[<?>] shift) => $$fshl(hi, lo, shift);
+macro uint[<?>] uint[<?>].fshr(hi, uint[<?>] lo, uint[<?>] shift) => $$fshr(hi, lo, shift);
+macro uint[<?>] uint[<?>].rotl(self, uint[<?>] shift) => $$fshl(self, self, shift);
+macro uint[<?>] uint[<?>].rotr(self, uint[<?>] shift) => $$fshr(self, self, shift);
+
+macro int[<?>].popcount(self) => $$popcount(self);
+macro int[<?>].ctz(self) => $$ctz(self);
+macro int[<?>].clz(self) => $$clz(self);
+macro int[<?>] int[<?>].fshl(hi, int[<?>] lo, int[<?>] shift) => $$fshl(hi, lo, shift);
+macro int[<?>] int[<?>].fshr(hi, int[<?>] lo, int[<?>] shift) => $$fshr(hi, lo, shift);
+macro int[<?>] int[<?>].rotl(self, int[<?>] shift) => $$fshl(self, self, shift);
+macro int[<?>] int[<?>].rotr(self, int[<?>] shift) => $$fshr(self, self, shift);
+
+macro ushort[<?>].popcount(self) => $$popcount(self);
+macro ushort[<?>].ctz(self) => $$ctz(self);
+macro ushort[<?>].clz(self) => $$clz(self);
+macro ushort[<?>] ushort[<?>].fshl(hi, ushort[<?>] lo, ushort[<?>] shift) => $$fshl(hi, lo, shift);
+macro ushort[<?>] ushort[<?>].fshr(hi, ushort[<?>] lo, ushort[<?>] shift) => $$fshr(hi, lo, shift);
+macro ushort[<?>] ushort[<?>].rotl(self, ushort[<?>] shift) => $$fshl(self, self, shift);
+macro ushort[<?>] ushort[<?>].rotr(self, ushort[<?>] shift) => $$fshr(self, self, shift);
+
+macro short[<?>].popcount(self) => $$popcount(self);
+macro short[<?>].ctz(self) => $$ctz(self);
+macro short[<?>].clz(self) => $$clz(self);
+macro short[<?>] short[<?>].fshl(hi, short[<?>] lo, short[<?>] shift) => $$fshl(hi, lo, shift);
+macro short[<?>] short[<?>].fshr(hi, short[<?>] lo, short[<?>] shift) => $$fshr(hi, lo, shift);
+macro short[<?>] short[<?>].rotl(self, short[<?>] shift) => $$fshl(self, self, shift);
+macro short[<?>] short[<?>].rotr(self, short[<?>] shift) => $$fshr(self, self, shift);
+
+macro char[<?>].popcount(self) => $$popcount(self);
+macro char[<?>].ctz(self) => $$ctz(self);
+macro char[<?>].clz(self) => $$clz(self);
+macro char[<?>] char[<?>].fshl(hi, char[<?>] lo, char[<?>] shift) => $$fshl(hi, lo, shift);
+macro char[<?>] char[<?>].fshr(hi, char[<?>] lo, char[<?>] shift) => $$fshr(hi, lo, shift);
+macro char[<?>] char[<?>].rotl(self, char[<?>] shift) => $$fshl(self, self, shift);
+macro char[<?>] char[<?>].rotr(self, char[<?>] shift) => $$fshr(self, self, shift);
+
+macro ichar[<?>].popcount(self) => $$popcount(self);
+macro ichar[<?>].ctz(self) => $$ctz(self);
+macro ichar[<?>].clz(self) => $$clz(self);
+macro ichar[<?>] ichar[<?>].fshl(hi, ichar[<?>] lo, ichar[<?>] shift) => $$fshl(hi, lo, shift);
+macro ichar[<?>] ichar[<?>].fshr(hi, ichar[<?>] lo, ichar[<?>] shift) => $$fshr(hi, lo, shift);
+macro ichar[<?>] ichar[<?>].rotl(self, ichar[<?>] shift) => $$fshl(self, self, shift);
+macro ichar[<?>] ichar[<?>].rotr(self, ichar[<?>] shift) => $$fshr(self, self, shift);
+
+macro ulong[<?>].popcount(self) => $$popcount(self);
+macro ulong[<?>].ctz(self) => $$ctz(self);
+macro ulong[<?>].clz(self) => $$clz(self);
+macro ulong[<?>] ulong[<?>].fshl(hi, ulong[<?>] lo, ulong[<?>] shift) => $$fshl(hi, lo, shift);
+macro ulong[<?>] ulong[<?>].fshr(hi, ulong[<?>] lo, ulong[<?>] shift) => $$fshr(hi, lo, shift);
+macro ulong[<?>] ulong[<?>].rotl(self, ulong[<?>] shift) => $$fshl(self, self, shift);
+macro ulong[<?>] ulong[<?>].rotr(self, ulong[<?>] shift) => $$fshr(self, self, shift);
+
+macro long[<?>].popcount(self) => $$popcount(self);
+macro long[<?>].ctz(self) => $$ctz(self);
+macro long[<?>].clz(self) => $$clz(self);
+macro long[<?>] long[<?>].fshl(hi, long[<?>] lo, long[<?>] shift) => $$fshl(hi, lo, shift);
+macro long[<?>] long[<?>].fshr(hi, long[<?>] lo, long[<?>] shift) => $$fshr(hi, lo, shift);
+macro long[<?>] long[<?>].rotl(self, long[<?>] shift) => $$fshl(self, self, shift);
+macro long[<?>] long[<?>].rotr(self, long[<?>] shift) => $$fshr(self, self, shift);
+
+macro uint128[<?>].popcount(self) => $$popcount(self);
+macro uint128[<?>].ctz(self) => $$ctz(self);
+macro uint128[<?>].clz(self) => $$clz(self);
+macro uint128[<?>] uint128[<?>].fshl(hi, uint128[<?>] lo, uint128[<?>] shift) => $$fshl(hi, lo, shift);
+macro uint128[<?>] uint128[<?>].fshr(hi, uint128[<?>] lo, uint128[<?>] shift) => $$fshr(hi, lo, shift);
+macro uint128[<?>] uint128[<?>].rotl(self, uint128[<?>] shift) => $$fshl(self, self, shift);
+macro uint128[<?>] uint128[<?>].rotr(self, uint128[<?>] shift) => $$fshr(self, self, shift);
+
+macro int128[<?>].popcount(self) => $$popcount(self);
+macro int128[<?>].ctz(self) => $$ctz(self);
+macro int128[<?>].clz(self) => $$clz(self);
+macro int128[<?>] int128[<?>].fshl(hi, int128[<?>] lo, int128[<?>] shift) => $$fshl(hi, lo, shift);
+macro int128[<?>] int128[<?>].fshr(hi, int128[<?>] lo, int128[<?>] shift) => $$fshr(hi, lo, shift);
+macro int128[<?>] int128[<?>].rotl(self, int128[<?>] shift) => $$fshl(self, self, shift);
+macro int128[<?>] int128[<?>].rotr(self, int128[<?>] shift) => $$fshr(self, self, shift);
+
+macro uint.popcount(self) => $$popcount(self);
+macro uint.ctz(self) => $$ctz(self);
+macro uint.clz(self) => $$clz(self);
+macro uint uint.fshl(hi, uint lo, uint shift) => $$fshl(hi, lo, shift);
+macro uint uint.fshr(hi, uint lo, uint shift) => $$fshr(hi, lo, shift);
+macro uint uint.rotl(self, uint shift) => $$fshl(self, self, shift);
+macro uint uint.rotr(self, uint shift) => $$fshr(self, self, shift);
+
+macro int.popcount(self) => $$popcount(self);
+macro int.ctz(self) => $$ctz(self);
+macro int.clz(self) => $$clz(self);
+macro int int.fshl(hi, int lo, int shift) => $$fshl(hi, lo, shift);
+macro int int.fshr(hi, int lo, int shift) => $$fshr(hi, lo, shift);
+macro int int.rotl(self, int shift) => $$fshl(self, self, shift);
+macro int int.rotr(self, int shift) => $$fshr(self, self, shift);
+
+macro ushort.popcount(self) => $$popcount(self);
+macro ushort.ctz(self) => $$ctz(self);
+macro ushort.clz(self) => $$clz(self);
+macro ushort ushort.fshl(hi, ushort lo, ushort shift) => $$fshl(hi, lo, shift);
+macro ushort ushort.fshr(hi, ushort lo, ushort shift) => $$fshr(hi, lo, shift);
+macro ushort ushort.rotl(self, ushort shift) => $$fshl(self, self, shift);
+macro ushort ushort.rotr(self, ushort shift) => $$fshr(self, self, shift);
+
+macro short.popcount(self) => $$popcount(self);
+macro short.ctz(self) => $$ctz(self);
+macro short.clz(self) => $$clz(self);
+macro short short.fshl(hi, short lo, short shift) => $$fshl(hi, lo, shift);
+macro short short.fshr(hi, short lo, short shift) => $$fshr(hi, lo, shift);
+macro short short.rotl(self, short shift) => $$fshl(self, self, shift);
+macro short short.rotr(self, short shift) => $$fshr(self, self, shift);
+
+macro char.popcount(self) => $$popcount(self);
+macro char.ctz(self) => $$ctz(self);
+macro char.clz(self) => $$clz(self);
+macro char char.fshl(hi, char lo, char shift) => $$fshl(hi, lo, shift);
+macro char char.fshr(hi, char lo, char shift) => $$fshr(hi, lo, shift);
+macro char char.rotl(self, char shift) => $$fshl(self, self, shift);
+macro char char.rotr(self, char shift) => $$fshr(self, self, shift);
+
+macro ichar.popcount(self) => $$popcount(self);
+macro ichar.ctz(self) => $$ctz(self);
+macro ichar.clz(self) => $$clz(self);
+macro ichar ichar.fshl(hi, ichar lo, ichar shift) => $$fshl(hi, lo, shift);
+macro ichar ichar.fshr(hi, ichar lo, ichar shift) => $$fshr(hi, lo, shift);
+macro ichar ichar.rotl(self, ichar shift) => $$fshl(self, self, shift);
+macro ichar ichar.rotr(self, ichar shift) => $$fshr(self, self, shift);
+
+macro ulong.popcount(self) => $$popcount(self);
+macro ulong.ctz(self) => $$ctz(self);
+macro ulong.clz(self) => $$clz(self);
+macro ulong ulong.fshl(hi, ulong lo, ulong shift) => $$fshl(hi, lo, shift);
+macro ulong ulong.fshr(hi, ulong lo, ulong shift) => $$fshr(hi, lo, shift);
+macro ulong ulong.rotl(self, ulong shift) => $$fshl(self, self, shift);
+macro ulong ulong.rotr(self, ulong shift) => $$fshr(self, self, shift);
+
+macro long.popcount(self) => $$popcount(self);
+macro long.ctz(self) => $$ctz(self);
+macro long.clz(self) => $$clz(self);
+macro long long.fshl(hi, long lo, long shift) => $$fshl(hi, lo, shift);
+macro long long.fshr(hi, long lo, long shift) => $$fshr(hi, lo, shift);
+macro long long.rotl(self, long shift) => $$fshl(self, self, shift);
+macro long long.rotr(self, long shift) => $$fshr(self, self, shift);
+
+macro uint128.popcount(self) => $$popcount(self);
+macro uint128.ctz(self) => $$ctz(self);
+macro uint128.clz(self) => $$clz(self);
+macro uint128 uint128.fshl(hi, uint128 lo, uint128 shift) => $$fshl(hi, lo, shift);
+macro uint128 uint128.fshr(hi, uint128 lo, uint128 shift) => $$fshr(hi, lo, shift);
+macro uint128 uint128.rotl(self, uint128 shift) => $$fshl(self, self, shift);
+macro uint128 uint128.rotr(self, uint128 shift) => $$fshr(self, self, shift);
+
+macro int128.popcount(self) => $$popcount(self);
+macro int128.ctz(self) => $$ctz(self);
+macro int128.clz(self) => $$clz(self);
+macro int128 int128.fshl(hi, int128 lo, int128 shift) => $$fshl(hi, lo, shift);
+macro int128 int128.fshr(hi, int128 lo, int128 shift) => $$fshr(hi, lo, shift);
+macro int128 int128.rotl(self, int128 shift) => $$fshl(self, self, shift);
+macro int128 int128.rotr(self, int128 shift) => $$fshr(self, self, shift);

lib/std/builtin.c3

@@ -1,176 +0,0 @@
-// Copyright (c) 2021 Christoffer Lerno. All rights reserved.
-// Use of this source code is governed by the MIT license
-// a copy of which can be found in the LICENSE_STDLIB file.
-module std::builtin;
-
-fault VarCastResult
-{
-	TYPE_MISMATCH
-}
-
-/**
- * Stores a variable on the stack, then restores it at the end of the
- * macro scope.
- *
- * @param variable `the variable to store and restore`
- **/
-macro void scope(&variable; @body) @autoimport
-{
-	$typeof(variable) temp = variable;
-	defer variable = temp;
-	@body();
-}
-
-/**
- * Convert a variant type to a type, returning an failure if there is a type mismatch.
- *
- * @param v `the variant to convert to the given type.`
- * @param $Type `the type to convert to`
- * @return `The variant.ptr converted to its type.`
- **/
-macro varcast(variant v, $Type) @autoimport
-{
-	if (v.type != $Type.typeid) return VarCastResult.TYPE_MISMATCH!;
-	return ($Type*)v.ptr;
-}
-
-extern fn void printf(char*, ...);
-
-struct CallstackElement
-{
-	CallstackElement* prev;
-	char* function;
-	char* file;
-	uint line;
-}
-fn void panic(char* message, char *file, char *function, uint line) @autoimport
-{
-	CallstackElement* stack = $$stacktrace();
-	$if ($defined(libc::stderr) && $defined(libc::fprintf)):
-
-	if (stack) stack = stack.prev;
-	if (stack)
-	{
-		libc::fprintf(@libc::stderr(), "\nERROR: '%s'\n", message);
-	}
-	else
-	{
-		libc::fprintf(@libc::stderr(), "\nERROR: '%s', function %s (%s:%d)\n", message, function, file, line);
-	}
-	while (stack)
-	{
-		libc::fprintf(@libc::stderr(), "  at function %s (%s:%u)\n", stack.function, stack.file, stack.line);
-		if (stack == stack.prev) break;
-		stack = stack.prev;
-	}
-
-	$endif;
-
-	$$trap();
-}
-
-macro unreachable($string = "Unreachable statement reached.") @autoimport @noreturn
-{
-	panic($string, $$FILE, $$FUNC, $$LINE);
-	$$unreachable();
-}
-
-/*
-enum TypeKind
-{
-    VOID,
-    BOOL,
-    FLOAT,
-    INTEGER,
-    STRUCT,
-    UNION,
-    ERROR,
-    ENUM,
-    ARRAY,
-    POINTER,
-    VAR_ARRAY,
-    SUBARRAY,
-    OPAQUE
-    // ALIAS,
-}
-
-struct TypeData
-{
-    typeid typeId;
-    TypeKind kind;
-    int size;
-    int alignment;
-    char* name;
-    char* fullName;
-}
-
-struct TypeAlias
-{
-    TypeData data;
-    typeid aliasType;
-}
-
-struct TypeError
-{
-    TypeData data;
-    TypeErrorValue[] errors;
-}
-
-struct TypeArray
-{
-    TypeData data;
-    typeid elementType;
-    ulong elements;
-}
-
-struct TypeVarArray
-{
-    TypeData data;
-    typeid elementType;
-}
-
-struct TypeSubarray
-{
-    TypeData data;
-    typeid elementType;
-}
-
-struct TypePointer
-{
-    TypeData data;
-    typeid baseType;
-}
-
-struct TypeStruct
-{
-    TypeData data;
-    TypeData*[] fields;
-}
-
-struct TypeUnion
-{
-    TypeData data;
-    TypeData*[] variants;
-}
-
-struct TypeEnum
-{
-    TypeData data;
-    typeid valueType;
-    TypeData*[] associated_value_types;
-}
-
-
-struct TypeEnumValue
-{
-    char* name;
-    ulong value;
-    void*[] associated_values;
-}
-
-struct TypeErrorValue
-{
-    char* name;
-    ulong value;
-}
-*/

lib/std/cinterop.c3

@@ -1,87 +0,0 @@
-// Copyright (c) 2021 Christoffer Lerno. All rights reserved.
-// Use of this source code is governed by the MIT license
-// a copy of which can be found in the LICENSE_STDLIB file.
-module std::cinterop;
-
-const C_INT_SIZE = $$C_INT_SIZE;
-const C_LONG_SIZE = $$C_LONG_SIZE;
-const C_SHORT_SIZE = $$C_SHORT_SIZE;
-const C_LONG_LONG_SIZE = $$C_LONG_LONG_SIZE;
-
-$assert (C_SHORT_SIZE < 32);
-$assert (C_INT_SIZE < 128);
-$assert (C_LONG_SIZE < 128);
-$assert (C_LONG_LONG_SIZE <= 128);
-$assert (C_SHORT_SIZE <= C_INT_SIZE);
-$assert (C_INT_SIZE <= C_LONG_SIZE);
-$assert (C_LONG_SIZE <= C_LONG_LONG_SIZE);
-
-$switch ($$C_INT_SIZE):
-$case 64:
-    define CInt = long;
-    define CUInt = ulong;
-$case 32:
-    define CInt = int;
-    define CUInt = uint;
-$case 16:
-    define CInt = short;
-    define CUInt = ushort;
-$default:
-    $assert(false, "Invalid C int size");
-$endswitch;
-
-$switch ($$C_LONG_SIZE):
-$case 64:
-    define CLong = long;
-    define CULong = ulong;
-$case 32:
-    define CLong = int;
-    define CULong = uint;
-$case 16:
-    define CLong = short;
-    define CULong = ushort;
-$default:
-    $assert(false, "Invalid C long size");
-$endswitch;
-
-$switch ($$C_SHORT_SIZE):
-$case 32:
-    define CShort = int;
-    define CUShort = uint;
-$case 16:
-    define CShort = short;
-    define CUShort = ushort;
-$case 8:
-    define CShort = ichar;
-    define CUShort = char;
-$default:
-    $assert(false, "Invalid C short size");
-$endswitch;
-
-$switch ($$C_LONG_LONG_SIZE):
-$case 128:
-    define CLongLong = int128;
-    define CULongLong = uint128;
-$case 64:
-    define CLongLong = long;
-    define CULongLong = ulong;
-$case 32:
-    define CLongLong = int;
-    define CULongLong = uint;
-$case 16:
-    define CLongLong = short;
-    define CULongLong = ushort;
-$default:
-    $assert(false, "Invalid C long long size");
-$endswitch;
-
-
-
-define CSChar = ichar;
-define CUChar = char;
-
-$if ($$C_CHAR_IS_SIGNED):
-    define CChar = ichar;
-$else:
-    define CChar = char;
-$endif;

lib/std/collections/anylist.c3

@@ -0,0 +1,514 @@
+// Copyright (c) 2024 Christoffer Lerno. All rights reserved.
+// Use of self source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+module std::collections::anylist;
+import std::io,std::math;
+
+def AnyPredicate = fn bool(any value);
+def AnyTest = fn bool(any type, any context);
+
+struct AnyList (Printable)
+{
+	usz size;
+	usz capacity;
+	Allocator allocator;
+	any* entries;
+}
+
+
+<*
+ Use `init` for to use a custom allocator.
+
+ @param initial_capacity "The initial capacity to reserve"
+*>
+fn AnyList* AnyList.new_init(&self, usz initial_capacity = 16, Allocator allocator = null) @deprecated("Use init(mem)")
+{
+	return self.init(allocator ?: allocator::heap(), initial_capacity) @inline;
+}
+
+<*
+ @param [&inout] allocator "The allocator to use"
+ @param initial_capacity "The initial capacity to reserve"
+*>
+fn AnyList* AnyList.init(&self, Allocator allocator, usz initial_capacity = 16)
+{
+	self.allocator = allocator;
+	self.size = 0;
+	if (initial_capacity > 0)
+	{
+		initial_capacity = math::next_power_of_2(initial_capacity);
+		self.entries = allocator::alloc_array(allocator, any, initial_capacity);
+	}
+	else
+	{
+		self.entries = null;
+	}
+	self.capacity = initial_capacity;
+	return self;
+}
+
+<*
+ Initialize the list using the temp allocator.
+
+ @param initial_capacity "The initial capacity to reserve"
+*>
+fn AnyList* AnyList.temp_init(&self, usz initial_capacity = 16) @deprecated("Use tinit")
+{
+	return self.init(allocator::temp(), initial_capacity) @inline;
+}
+
+
+<*
+ Initialize the list using the temp allocator.
+
+ @param initial_capacity "The initial capacity to reserve"
+*>
+fn AnyList* AnyList.tinit(&self, usz initial_capacity = 16)
+{
+	return self.init(allocator::temp(), initial_capacity) @inline;
+}
+
+fn usz! AnyList.to_format(&self, Formatter* formatter) @dynamic
+{
+	switch (self.size)
+	{
+		case 0:
+			return formatter.print("[]")!;
+		case 1:
+			return formatter.printf("[%s]", self.entries[0])!;
+		default:
+			usz n = formatter.print("[")!;
+			foreach (i, element : self.entries[:self.size])
+			{
+				if (i != 0) formatter.print(", ")!;
+				n += formatter.printf("%s", element)!;
+			}
+			n += formatter.print("]")!;
+			return n;
+	}
+}
+
+fn String AnyList.to_new_string(&self, Allocator allocator = null) @dynamic
+{
+	return string::format("%s", *self, allocator: allocator ?: allocator::heap());
+}
+
+
+fn String AnyList.to_string(&self, Allocator allocator) @dynamic
+{
+	return string::format("%s", *self, allocator: allocator);
+}
+
+fn String AnyList.to_tstring(&self) => string::tformat("%s", *self);
+
+<*
+ Push an element on the list by cloning it.
+*>
+macro void AnyList.push(&self, element)
+{
+	if (!self.allocator) self.allocator = allocator::heap();
+	self.append_internal(allocator::clone(self.allocator, element));
+}
+
+fn void AnyList.append_internal(&self, any element) @local
+{
+	self.ensure_capacity();
+	self.entries[self.size++] = element;
+}
+
+<*
+ Free a retained element removed using *_retained.
+*>
+fn void AnyList.free_element(&self, any element) @inline
+{
+	allocator::free(self.allocator, element.ptr);
+}
+
+<*
+ Pop a value who's type is known. If the type is incorrect, this
+ will still pop the element.
+
+ @return! CastResult.TYPE_MISMATCH, IteratorResult.NO_MORE_ELEMENT
+*>
+macro AnyList.pop(&self, $Type)
+{
+	if (!self.size) return IteratorResult.NO_MORE_ELEMENT?;
+	defer self.free_element(self.entries[self.size]);
+	return *anycast(self.entries[--self.size], $Type);
+}
+
+<*
+ Pop the last value and allocate the copy using the given allocator.
+ @return! IteratorResult.NO_MORE_ELEMENT
+*>
+fn any! AnyList.copy_pop(&self, Allocator allocator = allocator::heap())
+{
+	if (!self.size) return IteratorResult.NO_MORE_ELEMENT?;
+	defer self.free_element(self.entries[self.size]);
+	return allocator::clone_any(allocator, self.entries[--self.size]);
+}
+
+<*
+ Pop the last value and allocate the copy using the given allocator.
+ @return! IteratorResult.NO_MORE_ELEMENT
+ @deprecated `use copy_pop`
+*>
+fn any! AnyList.new_pop(&self, Allocator allocator = allocator::heap())
+{
+	return self.copy_pop(allocator);
+}
+
+<*
+ Pop the last value and allocate the copy using the temp allocator
+ @return! IteratorResult.NO_MORE_ELEMENT
+ @deprecated `use tcopy_pop`
+*>
+fn any! AnyList.temp_pop(&self) => self.copy_pop(allocator::temp());
+
+<*
+ Pop the last value and allocate the copy using the temp allocator
+ @return! IteratorResult.NO_MORE_ELEMENT
+*>
+fn any! AnyList.tcopy_pop(&self) => self.copy_pop(allocator::temp());
+
+<*
+ Pop the last value. It must later be released using list.free_element()
+ @return! IteratorResult.NO_MORE_ELEMENT
+*>
+fn any! AnyList.pop_retained(&self)
+{
+	if (!self.size) return IteratorResult.NO_MORE_ELEMENT?;
+	return self.entries[--self.size];
+}
+
+fn void AnyList.clear(&self)
+{
+	for (usz i = 0; i < self.size; i++)
+	{
+		self.free_element(self.entries[i]);
+	}
+	self.size = 0;
+}
+
+<*
+ Same as pop() but pops the first value instead.
+*>
+macro AnyList.pop_first(&self, $Type)
+{
+	if (!self.size) return IteratorResult.NO_MORE_ELEMENT?;
+	defer self.remove_at(0);
+	return *anycast(self.entries[0], $Type);
+}
+
+<*
+ Same as pop_retained() but pops the first value instead.
+*>
+fn any! AnyList.pop_first_retained(&self)
+{
+	if (!self.size) return IteratorResult.NO_MORE_ELEMENT?;
+	defer self.remove_at(0);
+	return self.entries[0];
+}
+
+<*
+ Same as new_pop() but pops the first value instead.
+ @deprecated `use copy_pop_first`
+*>
+fn any! AnyList.new_pop_first(&self, Allocator allocator = allocator::heap())
+{
+	return self.copy_pop_first(allocator) @inline;
+}
+
+<*
+ Same as new_pop() but pops the first value instead.
+*>
+fn any! AnyList.copy_pop_first(&self, Allocator allocator = allocator::heap())
+{
+	if (!self.size) return IteratorResult.NO_MORE_ELEMENT?;
+	defer self.free_element(self.entries[self.size]);
+	defer self.remove_at(0);
+	return allocator::clone_any(allocator, self.entries[0]);
+}
+
+<*
+ Same as temp_pop() but pops the first value instead.
+*>
+fn any! AnyList.tcopy_pop_first(&self) => self.copy_pop_first(allocator::temp());
+
+<*
+ Same as temp_pop() but pops the first value instead.
+ @deprecated `use tcopy_pop_first`
+*>
+fn any! AnyList.temp_pop_first(&self) => self.new_pop_first(allocator::temp());
+
+<*
+ @require index < self.size
+*>
+fn void AnyList.remove_at(&self, usz index)
+{
+	if (!--self.size || index == self.size) return;
+	self.free_element(self.entries[index]);
+	self.entries[index .. self.size - 1] = self.entries[index + 1 .. self.size];
+}
+
+fn void AnyList.add_all(&self, AnyList* other_list)
+{
+	if (!other_list.size) return;
+	self.reserve(other_list.size);
+	foreach (value : other_list)
+	{
+		self.entries[self.size++] = allocator::clone_any(self.allocator, value);
+	}
+}
+
+<*
+ Reverse the elements in a list.
+*>
+fn void AnyList.reverse(&self)
+{
+	if (self.size < 2) return;
+	usz half = self.size / 2U;
+	usz end = self.size - 1;
+	for (usz i = 0; i < half; i++)
+	{
+		self.swap(i, end - i);
+	}
+}
+
+fn any[] AnyList.array_view(&self)
+{
+	return self.entries[:self.size];
+}
+
+<*
+ Push an element to the front of the list.
+*>
+macro void AnyList.push_front(&self, type)
+{
+	self.insert_at(0, type);
+}
+
+<*
+ @require index < self.size
+*>
+macro void AnyList.insert_at(&self, usz index, type) @local
+{
+	any value = allocator::copy(self.allocator, type);
+	self.insert_at_internal(self, index, value);
+}
+
+<*
+ @require index < self.size
+*>
+fn void AnyList.insert_at_internal(&self, usz index, any value) @local
+{
+	self.ensure_capacity();
+	for (usz i = self.size; i > index; i--)
+	{
+		self.entries[i] = self.entries[i - 1];
+	}
+	self.size++;
+	self.entries[index] = value;
+}
+
+
+<*
+ @require self.size > 0
+*>
+fn void AnyList.remove_last(&self)
+{
+	self.free_element(self.entries[--self.size]);
+}
+
+<*
+ @require self.size > 0
+*>
+fn void AnyList.remove_first(&self)
+{
+	self.remove_at(0);
+}
+
+macro AnyList.first(&self, $Type)
+{
+	return *anycast(self.first_any(), $Type);
+}
+
+fn any! AnyList.first_any(&self) @inline
+{
+	return self.size ? self.entries[0] : IteratorResult.NO_MORE_ELEMENT?;
+}
+
+macro AnyList.last(&self, $Type)
+{
+	return *anycast(self.last_any(), $Type);
+}
+
+fn any! AnyList.last_any(&self) @inline
+{
+	return self.size ? self.entries[self.size - 1] : IteratorResult.NO_MORE_ELEMENT?;
+}
+
+fn bool AnyList.is_empty(&self) @inline
+{
+	return !self.size;
+}
+
+fn usz AnyList.len(&self) @operator(len) @inline
+{
+	return self.size;
+}
+
+<*
+ @require index < self.size "Index out of range"
+*>
+macro AnyList.get(&self, usz index, $Type)
+{
+	return *anycast(self.entries[index], $Type);
+}
+
+<*
+ @require index < self.size "Index out of range"
+*>
+fn any AnyList.get_any(&self, usz index) @inline
+{
+	return self.entries[index];
+}
+
+fn void AnyList.free(&self)
+{
+	if (!self.allocator) return;
+	self.clear();
+	allocator::free(self.allocator, self.entries);
+	self.capacity = 0;
+	self.entries = null;
+}
+
+fn void AnyList.swap(&self, usz i, usz j)
+{
+	any temp = self.entries[i];
+	self.entries[i] = self.entries[j];
+	self.entries[j] = temp;
+}
+
+<*
+ @param filter "The function to determine if it should be removed or not"
+ @return "the number of deleted elements"
+*>
+fn usz AnyList.remove_if(&self, AnyPredicate filter)
+{
+	return self._remove_if(filter, false);
+}
+
+<*
+ @param selection "The function to determine if it should be kept or not"
+ @return "the number of deleted elements"
+*>
+fn usz AnyList.retain_if(&self, AnyPredicate selection)
+{
+	return self._remove_if(selection, true);
+}
+
+macro usz AnyList._remove_if(&self, AnyPredicate filter, bool $invert) @local
+{
+	usz size = self.size;
+	for (usz i = size, usz k = size; k > 0; k = i)
+	{
+		// Find last index of item to be deleted.
+		$if $invert:
+			while (i > 0 && !filter(&self.entries[i - 1])) i--;
+		$else
+			while (i > 0 && filter(&self.entries[i - 1])) i--;
+		$endif
+		// Remove the items from this index up to the one not to be deleted.
+		usz n = self.size - k;
+		for (usz j = i; j < k; j++) self.free_element(self.entries[j]);
+		self.entries[i:n] = self.entries[k:n];
+		self.size -= k - i;
+		// Find last index of item not to be deleted.
+		$if $invert:
+			while (i > 0 && filter(&self.entries[i - 1])) i--;
+		$else
+			while (i > 0 && !filter(&self.entries[i - 1])) i--;
+		$endif
+	}
+	return size - self.size;
+}
+
+fn usz AnyList.remove_using_test(&self, AnyTest filter, any context)
+{
+	return self._remove_using_test(filter, false, context);
+}
+
+fn usz AnyList.retain_using_test(&self, AnyTest filter, any context)
+{
+	return self._remove_using_test(filter, true, context);
+}
+
+macro usz AnyList._remove_using_test(&self, AnyTest filter, bool $invert, ctx) @local
+{
+	usz size = self.size;
+	for (usz i = size, usz k = size; k > 0; k = i)
+	{
+		// Find last index of item to be deleted.
+		$if $invert:
+			while (i > 0 && !filter(&self.entries[i - 1], ctx)) i--;
+		$else
+			while (i > 0 && filter(&self.entries[i - 1], ctx)) i--;
+		$endif
+		// Remove the items from this index up to the one not to be deleted.
+		usz n = self.size - k;
+		for (usz j = i; j < k; j++) self.free_element(self.entries[j]);
+		self.entries[i:n] = self.entries[k:n];
+		self.size -= k - i;
+		// Find last index of item not to be deleted.
+		$if $invert:
+			while (i > 0 && filter(&self.entries[i - 1], ctx)) i--;
+		$else
+			while (i > 0 && !filter(&self.entries[i - 1], ctx)) i--;
+		$endif
+	}
+	return size - self.size;
+}
+
+<*
+ Reserve at least min_capacity
+*>
+fn void AnyList.reserve(&self, usz min_capacity)
+{
+	if (!min_capacity) return;
+	if (self.capacity >= min_capacity) return;
+	if (!self.allocator) self.allocator = allocator::heap();
+	min_capacity = math::next_power_of_2(min_capacity);
+	self.entries = allocator::realloc(self.allocator, self.entries, any.sizeof * min_capacity);
+	self.capacity = min_capacity;
+}
+
+macro any AnyList.@item_at(&self, usz index) @operator([])
+{
+	return self.entries[index];
+}
+
+<*
+ @require index <= self.size "Index out of range"
+*>
+macro void AnyList.set(&self, usz index, value)
+{
+	if (index == self.size)
+	{
+		self.push(value);
+		return;
+	}
+	self.free_element(self.entries[index]);
+	self.entries[index] = allocator::copy(self.allocator, value);
+}
+
+fn void AnyList.ensure_capacity(&self, usz added = 1) @inline @private
+{
+	usz new_size = self.size + added;
+	if (self.capacity >= new_size) return;
+
+	assert(new_size < usz.max / 2U);
+	usz new_capacity = self.capacity ? 2U * self.capacity : 16U;
+	while (new_capacity < new_size) new_capacity *= 2U;
+	self.reserve(new_capacity);
+}

lib/std/collections/bitset.c3

@@ -0,0 +1,170 @@
+<*
+ @require SIZE > 0
+*>
+module std::collections::bitset(<SIZE>);
+
+def Type = uint;
+
+const BITS = Type.sizeof * 8;
+const SZ = (SIZE + BITS - 1) / BITS;
+
+struct BitSet
+{
+	Type[SZ] data;
+}
+
+fn usz BitSet.cardinality(&self)
+{
+	usz n;
+	foreach (x : self.data)
+	{
+		n += x.popcount();
+	}
+	return n;
+}
+
+<*
+ @require i < SIZE
+*>
+fn void BitSet.set(&self, usz i)
+{
+	usz q = i / BITS;
+	usz r = i % BITS;
+	self.data[q] |= 1 << r;
+}
+
+<*
+ @require i < SIZE
+*>
+fn void BitSet.unset(&self, usz i)
+{
+	usz q = i / BITS;
+	usz r = i % BITS;
+	self.data[q] &= ~(1 << r);
+}
+
+<*
+ @require i < SIZE
+*>
+fn bool BitSet.get(&self, usz i) @operator([]) @inline
+{
+	usz q = i / BITS;
+	usz r = i % BITS;
+	return self.data[q] & (1 << r) != 0;
+}
+
+fn usz BitSet.len(&self) @operator(len) @inline
+{
+	return SZ * BITS;
+}
+
+<*
+ @require i < SIZE
+*>
+fn void BitSet.set_bool(&self, usz i, bool value) @operator([]=) @inline
+{
+	if (value) return self.set(i);
+	self.unset(i);
+}
+
+<*
+ @require Type.kindof == UNSIGNED_INT
+*>
+module std::collections::growablebitset(<Type>);
+import std::collections::list;
+
+const BITS = Type.sizeof * 8;
+
+def GrowableBitSetList = List(<Type>);
+
+struct GrowableBitSet
+{
+	GrowableBitSetList data;
+}
+
+<*
+ @param initial_capacity
+ @param [&inout] allocator "The allocator to use, defaults to the heap allocator"
+*>
+fn GrowableBitSet* GrowableBitSet.new_init(&self, usz initial_capacity = 1, Allocator allocator = allocator::heap()) @deprecated("Use init(mem)")
+{
+	self.data.init(allocator, initial_capacity);
+	return self;
+}
+
+<*
+ @param initial_capacity
+ @param [&inout] allocator "The allocator to use, defaults to the heap allocator"
+*>
+fn GrowableBitSet* GrowableBitSet.init(&self, Allocator allocator, usz initial_capacity = 1)
+{
+	self.data.init(allocator, initial_capacity);
+	return self;
+}
+
+fn GrowableBitSet* GrowableBitSet.temp_init(&self, usz initial_capacity = 1) @deprecated("Use tinit()")
+{
+	return self.init(allocator::temp(), initial_capacity) @inline;
+}
+
+fn GrowableBitSet* GrowableBitSet.tinit(&self, usz initial_capacity = 1)
+{
+	return self.init(allocator::temp(), initial_capacity) @inline;
+}
+
+fn void GrowableBitSet.free(&self)
+{
+	self.data.free();
+}
+
+fn usz GrowableBitSet.cardinality(&self)
+{
+	usz n;
+	foreach (x : self.data)
+	{
+		n += x.popcount();
+	}
+	return n;
+}
+
+fn void GrowableBitSet.set(&self, usz i)
+{
+	usz q = i / BITS;
+	usz r = i % BITS;
+	usz current_len = self.data.len();
+	while (q >= current_len)
+	{
+		self.data.push(0);
+		current_len++;
+	}
+	self.data.set(q, self.data[q] | (1 << r));
+}
+
+fn void GrowableBitSet.unset(&self, usz i)
+{
+	usz q = i / BITS;
+	usz r = i % BITS;
+	if (q >= self.data.len()) return;
+	self.data.set(q, self.data[q] &~ (1 << r));
+}
+
+fn bool GrowableBitSet.get(&self, usz i) @operator([]) @inline
+{
+	usz q = i / BITS;
+	usz r = i % BITS;
+	if (q >= self.data.len()) return false;
+	return self.data[q] & (1 << r) != 0;
+}
+
+fn usz GrowableBitSet.len(&self) @operator(len)
+{
+	usz n = self.data.len() * BITS;
+	if (n > 0) n -= (usz)self.data[^1].clz();
+	return n;
+}
+
+fn void GrowableBitSet.set_bool(&self, usz i, bool value) @operator([]=) @inline
+{
+	if (value) return self.set(i);
+	self.unset(i);
+}

lib/std/collections/elastic_array.c3

@@ -0,0 +1,455 @@
+// Copyright (c) 2021-2024 Christoffer Lerno. All rights reserved.
+// Use of self source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+<*
+ @require MAX_SIZE >= 1 `The size must be at least 1 element big.`
+*>
+module std::collections::elastic_array(<Type, MAX_SIZE>);
+import std::io, std::math, std::collections::list_common;
+
+def ElementPredicate = fn bool(Type *type);
+def ElementTest = fn bool(Type *type, any context);
+const ELEMENT_IS_EQUATABLE = types::is_equatable_type(Type);
+const ELEMENT_IS_POINTER = Type.kindof == POINTER;
+macro type_is_overaligned() => Type.alignof > mem::DEFAULT_MEM_ALIGNMENT;
+
+struct ElasticArray (Printable)
+{
+	usz size;
+	Type[MAX_SIZE] entries;
+}
+
+fn usz! ElasticArray.to_format(&self, Formatter* formatter) @dynamic
+{
+	switch (self.size)
+	{
+		case 0:
+			return formatter.print("[]")!;
+		case 1:
+			return formatter.printf("[%s]", self.entries[0])!;
+		default:
+			usz n = formatter.print("[")!;
+			foreach (i, element : self.entries[:self.size])
+			{
+				if (i != 0) formatter.print(", ")!;
+				n += formatter.printf("%s", element)!;
+			}
+			n += formatter.print("]")!;
+			return n;
+	}
+}
+
+fn String ElasticArray.to_string(&self, Allocator allocator) @dynamic
+{
+	return string::format("%s", *self, allocator: allocator);
+}
+
+fn String ElasticArray.to_new_string(&self, Allocator allocator = nul) @dynamic
+{
+	return string::format("%s", *self, allocator: allocator ?: allocator::heap());
+}
+
+fn String ElasticArray.to_tstring(&self)
+{
+	return string::tformat("%s", *self);
+}
+
+fn void! ElasticArray.push_try(&self, Type element) @inline
+{
+	if (self.size == MAX_SIZE) return AllocationFailure.OUT_OF_MEMORY?;
+	self.entries[self.size++] = element;
+}
+
+<*
+ @require self.size < MAX_SIZE `Tried to exceed the max size`
+*>
+fn void ElasticArray.push(&self, Type element) @inline
+{
+	self.entries[self.size++] = element;
+}
+
+fn Type! ElasticArray.pop(&self)
+{
+	if (!self.size) return IteratorResult.NO_MORE_ELEMENT?;
+	return self.entries[--self.size];
+}
+
+fn void ElasticArray.clear(&self)
+{
+	self.size = 0;
+}
+
+<*
+ @require self.size > 0
+*>
+fn Type! ElasticArray.pop_first(&self)
+{
+	if (!self.size) return IteratorResult.NO_MORE_ELEMENT?;
+	defer self.remove_at(0);
+	return self.entries[0];
+}
+
+<*
+ @require index < self.size
+*>
+fn void ElasticArray.remove_at(&self, usz index)
+{
+	if (!--self.size || index == self.size) return;
+	self.entries[index .. self.size - 1] = self.entries[index + 1 .. self.size];
+}
+
+<*
+ @require other_list.size + self.size <= MAX_SIZE
+*>
+fn void ElasticArray.add_all(&self, ElasticArray* other_list)
+{
+	if (!other_list.size) return;
+	foreach (&value : other_list)
+	{
+		self.entries[self.size++] = *value;
+	}
+}
+
+<*
+ Add as many elements as possible to the new array,
+ returning the number of elements that didn't fit.
+*>
+fn usz ElasticArray.add_all_to_limit(&self, ElasticArray* other_list)
+{
+	if (!other_list.size) return 0;
+	foreach (i, &value : other_list)
+	{
+		if (self.size == MAX_SIZE) return other_list.size - i;
+		self.entries[self.size++] = *value;
+	}
+	return 0;
+}
+
+<*
+ Add as many values from this array as possible, returning the
+ number of elements that didn't fit.
+
+ @param [in] array
+*>
+fn usz ElasticArray.add_array_to_limit(&self, Type[] array)
+{
+	if (!array.len) return 0;
+	foreach (i, &value : array)
+	{
+		if (self.size == MAX_SIZE) return array.len - i;
+		self.entries[self.size++] = *value;
+	}
+	return 0;
+}
+
+<*
+ Add the values of an array to this list.
+
+ @param [in] array
+ @require array.len + self.size <= MAX_SIZE `Size would exceed max.`
+ @ensure self.size >= array.len
+*>
+fn void ElasticArray.add_array(&self, Type[] array)
+{
+	if (!array.len) return;
+	foreach (&value : array)
+	{
+		self.entries[self.size++] = *value;
+	}
+}
+
+
+
+<*
+ IMPORTANT The returned array must be freed using free_aligned.
+*>
+fn Type[] ElasticArray.to_new_aligned_array(&self)
+{
+	return list_common::list_to_new_aligned_array(Type, self, allocator::heap());
+}
+
+<*
+ IMPORTANT The returned array must be freed using free_aligned.
+*>
+fn Type[] ElasticArray.to_aligned_array(&self, Allocator allocator)
+{
+	return list_common::list_to_new_aligned_array(Type, self, allocator);
+}
+
+<*
+ @require !type_is_overaligned() : "This function is not available on overaligned types"
+*>
+macro Type[] ElasticArray.to_new_array(&self)
+{
+	return list_common::list_to_array(Type, self, allocator::heap());
+}
+
+<*
+ @require !type_is_overaligned() : "This function is not available on overaligned types"
+*>
+macro Type[] ElasticArray.to_array(&self, Allocator allocator)
+{
+	return list_common::list_to_new_array(Type, self, allocator);
+}
+
+fn Type[] ElasticArray.to_tarray(&self)
+{
+	$if type_is_overaligned():
+	return self.to_aligned_array(allocator::temp());
+	$else
+	return self.to_array(allocator::temp());
+	$endif;
+}
+
+<*
+ Reverse the elements in a list.
+*>
+fn void ElasticArray.reverse(&self)
+{
+	list_common::list_reverse(self);
+}
+
+fn Type[] ElasticArray.array_view(&self)
+{
+	return self.entries[:self.size];
+}
+
+<*
+ @require self.size < MAX_SIZE `List would exceed max size`
+*>
+fn void ElasticArray.push_front(&self, Type type) @inline
+{
+	self.insert_at(0, type);
+}
+
+<*
+ @require self.size < MAX_SIZE `List would exceed max size`
+*>
+fn void! ElasticArray.push_front_try(&self, Type type) @inline
+{
+	return self.insert_at_try(0, type);
+}
+
+<*
+ @require index <= self.size
+*>
+fn void! ElasticArray.insert_at_try(&self, usz index, Type value)
+{
+	if (self.size == MAX_SIZE) return AllocationFailure.OUT_OF_MEMORY?;
+	self.insert_at(index, value);
+}
+
+<*
+ @require self.size < MAX_SIZE `List would exceed max size`
+ @require index <= self.size
+*>
+fn void ElasticArray.insert_at(&self, usz index, Type type)
+{
+	for (usz i = self.size; i > index; i--)
+	{
+		self.entries[i] = self.entries[i - 1];
+	}
+	self.size++;
+	self.entries[index] = type;
+}
+
+<*
+ @require index < self.size
+*>
+fn void ElasticArray.set_at(&self, usz index, Type type)
+{
+	self.entries[index] = type;
+}
+
+fn void! ElasticArray.remove_last(&self) @maydiscard
+{
+	if (!self.size) return IteratorResult.NO_MORE_ELEMENT?;
+	self.size--;
+}
+
+fn void! ElasticArray.remove_first(&self) @maydiscard
+{
+	if (!self.size) return IteratorResult.NO_MORE_ELEMENT?;
+	self.remove_at(0);
+}
+
+fn Type! ElasticArray.first(&self)
+{
+	if (!self.size) return IteratorResult.NO_MORE_ELEMENT?;
+	return self.entries[0];
+}
+
+fn Type! ElasticArray.last(&self)
+{
+	if (!self.size) return IteratorResult.NO_MORE_ELEMENT?;
+	return self.entries[self.size - 1];
+}
+
+fn bool ElasticArray.is_empty(&self) @inline
+{
+	return !self.size;
+}
+
+fn usz ElasticArray.byte_size(&self) @inline
+{
+	return Type.sizeof * self.size;
+}
+
+fn usz ElasticArray.len(&self) @operator(len) @inline
+{
+	return self.size;
+}
+
+fn Type ElasticArray.get(&self, usz index) @inline
+{
+	return self.entries[index];
+}
+
+fn void ElasticArray.swap(&self, usz i, usz j)
+{
+	@swap(self.entries[i], self.entries[j]);
+}
+
+<*
+ @param filter "The function to determine if it should be removed or not"
+ @return "the number of deleted elements"
+*>
+fn usz ElasticArray.remove_if(&self, ElementPredicate filter)
+{
+	return list_common::list_remove_if(self, filter, false);
+}
+
+<*
+ @param selection "The function to determine if it should be kept or not"
+ @return "the number of deleted elements"
+*>
+fn usz ElasticArray.retain_if(&self, ElementPredicate selection)
+{
+	return list_common::list_remove_if(self, selection, true);
+}
+
+fn usz ElasticArray.remove_using_test(&self, ElementTest filter, any context)
+{
+	return list_common::list_remove_using_test(self, filter, false, context);
+}
+
+fn usz ElasticArray.retain_using_test(&self, ElementTest filter, any context)
+{
+	return list_common::list_remove_using_test(self, filter, true, context);
+}
+
+
+macro Type ElasticArray.@item_at(&self, usz index) @operator([])
+{
+	return self.entries[index];
+}
+
+fn Type* ElasticArray.get_ref(&self, usz index) @operator(&[]) @inline
+{
+	return &self.entries[index];
+}
+
+fn void ElasticArray.set(&self, usz index, Type value) @operator([]=)
+{
+	self.entries[index] = value;
+}
+
+
+// Functions for equatable types
+fn usz! ElasticArray.index_of(&self, Type type) @if(ELEMENT_IS_EQUATABLE)
+{
+	foreach (i, v : self)
+	{
+		if (equals(v, type)) return i;
+	}
+	return SearchResult.MISSING?;
+}
+
+fn usz! ElasticArray.rindex_of(&self, Type type) @if(ELEMENT_IS_EQUATABLE)
+{
+	foreach_r (i, v : self)
+	{
+		if (equals(v, type)) return i;
+	}
+	return SearchResult.MISSING?;
+}
+
+fn bool ElasticArray.equals(&self, ElasticArray other_list) @if(ELEMENT_IS_EQUATABLE)
+{
+	if (self.size != other_list.size) return false;
+	foreach (i, v : self)
+	{
+		if (!equals(v, other_list.entries[i])) return false;
+	}
+	return true;
+}
+
+<*
+ Check for presence of a value in a list.
+
+ @param [&in] self "the list to find elements in"
+ @param value "The value to search for"
+ @return "True if the value is found, false otherwise"
+*>
+fn bool ElasticArray.contains(&self, Type value) @if(ELEMENT_IS_EQUATABLE)
+{
+	foreach (i, v : self)
+	{
+		if (equals(v, value)) return true;
+	}
+	return false;
+}
+
+<*
+ @param [&inout] self "The list to remove elements from"
+ @param value "The value to remove"
+ @return "true if the value was found"
+*>
+fn bool ElasticArray.remove_last_item(&self, Type value) @if(ELEMENT_IS_EQUATABLE)
+{
+	return @ok(self.remove_at(self.rindex_of(value)));
+}
+
+<*
+ @param [&inout] self "The list to remove elements from"
+ @param value "The value to remove"
+ @return "true if the value was found"
+*>
+fn bool ElasticArray.remove_first_item(&self, Type value) @if(ELEMENT_IS_EQUATABLE)
+{
+	return @ok(self.remove_at(self.index_of(value)));
+}
+
+<*
+ @param [&inout] self "The list to remove elements from"
+ @param value "The value to remove"
+ @return "the number of deleted elements."
+*>
+fn usz ElasticArray.remove_item(&self, Type value) @if(ELEMENT_IS_EQUATABLE)
+{
+	return list_common::list_remove_item(self, value);
+}
+
+
+
+fn void ElasticArray.remove_all_from(&self, ElasticArray* other_list) @if(ELEMENT_IS_EQUATABLE)
+{
+	if (!other_list.size) return;
+	foreach (v : other_list) self.remove_item(v);
+}
+
+<*
+ @param [&in] self
+ @return "The number non-null values in the list"
+*>
+fn usz ElasticArray.compact_count(&self) @if(ELEMENT_IS_POINTER)
+{
+	usz vals = 0;
+	foreach (v : self) if (v) vals++;
+	return vals;
+}
+
+fn usz ElasticArray.compact(&self) @if(ELEMENT_IS_POINTER)
+{
+	return list_common::list_compact(self);
+}

lib/std/collections/enummap.c3

@@ -0,0 +1,71 @@
+<*
+ @require Enum.kindof == TypeKind.ENUM : "Only enums may be used with an enummap"
+*>
+module std::collections::enummap(<Enum, ValueType>);
+import std::io;
+struct EnumMap (Printable)
+{
+	ValueType[Enum.len] values;
+}
+
+fn void EnumMap.init(&self, ValueType init_value)
+{
+	foreach (&a : self.values)
+	{
+		*a = init_value;
+	}
+}
+
+fn usz! EnumMap.to_format(&self, Formatter* formatter) @dynamic
+{
+	usz n = formatter.print("{ ")!;
+	foreach (i, &value : self.values)
+	{
+		if (i != 0) formatter.print(", ")!;
+		n += formatter.printf("%s: %s", Enum.from_ordinal(i), *value)!;
+	}
+	n += formatter.print(" }")!;
+	return n;
+}
+
+fn String EnumMap.to_string(&self, Allocator allocator) @dynamic
+{
+	return string::format("%s", *self, allocator: allocator);
+}
+
+fn String EnumMap.to_new_string(&self, Allocator allocator = null) @dynamic
+{
+	return string::format("%s", *self, allocator: allocator ?: allocator::heap());
+}
+
+fn String EnumMap.to_tstring(&self) @dynamic
+{
+	return string::tformat("%s", *self);
+}
+
+<*
+ @return "The total size of this map, which is the same as the number of enum values"
+ @pure
+*>
+fn usz EnumMap.len(&self) @operator(len) @inline
+{
+	return self.values.len;
+}
+
+<*
+ @return "Retrieve a value given the underlying enum, if there is no entry, then the zero value for the value is returned."
+*>
+fn ValueType EnumMap.get(&self, Enum key) @operator([]) @inline
+{
+	return self.values[key.ordinal];
+}
+
+fn ValueType* EnumMap.get_ref(&self, Enum key) @operator(&[]) @inline
+{
+	return &self.values[key.ordinal];
+}
+
+fn void EnumMap.set(&self, Enum key, ValueType value) @operator([]=) @inline
+{
+	self.values[key.ordinal] = value;
+}

lib/std/collections/enumset.c3

@@ -0,0 +1,177 @@
+// Copyright (c) 2021-2024 Christoffer Lerno. All rights reserved.
+// Use of self source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+
+<*
+ @require Enum.kindof == TypeKind.ENUM : "Only enums may be used with an enumset"
+*>
+module std::collections::enumset(<Enum>);
+import std::io;
+
+def EnumSetType = $typefrom(private::type_for_enum_elements(Enum.elements)) @private;
+
+const IS_CHAR_ARRAY = Enum.elements > 128;
+distinct EnumSet (Printable) = EnumSetType;
+
+fn void EnumSet.add(&self, Enum v)
+{
+	$if IS_CHAR_ARRAY:
+		(*self)[(usz)v.ordinal / 8] |= (char)(1u << ((usz)v.ordinal % 8));
+	$else
+		*self = (EnumSet)((EnumSetType)*self | 1u << (EnumSetType)v.ordinal);
+	$endif
+}
+
+fn void EnumSet.clear(&self)
+{
+	$if IS_CHAR_ARRAY:
+		*self = {};
+	$else
+		*self = 0;
+	$endif
+}
+
+fn bool EnumSet.remove(&self, Enum v)
+{
+	$if IS_CHAR_ARRAY:
+		if (!self.has(v) @inline) return false;
+		(*self)[(usz)v.ordinal / 8] &= (char)~(1u << ((usz)v.ordinal % 8));
+		return true;
+	$else
+		EnumSetType old = (EnumSetType)*self;
+		EnumSetType new = old & ~(1u << (EnumSetType)v.ordinal);
+		*self = (EnumSet)new;
+		return old != new;
+	$endif
+}
+
+fn bool EnumSet.has(&self, Enum v)
+{
+	$if IS_CHAR_ARRAY:
+		return (bool)(((*self)[(usz)v.ordinal / 8] << ((usz)v.ordinal % 8)) & 0x01);
+	$else
+		return ((EnumSetType)*self & (1u << (EnumSetType)v.ordinal)) != 0;
+	$endif
+}
+
+fn void EnumSet.add_all(&self, EnumSet s)
+{
+	$if IS_CHAR_ARRAY:
+		foreach (i, c : s) (*self)[i] |= c;
+	$else
+		*self = (EnumSet)((EnumSetType)*self | (EnumSetType)s);
+	$endif
+}
+
+fn void EnumSet.retain_all(&self, EnumSet s)
+{
+	$if IS_CHAR_ARRAY:
+		foreach (i, c : s) (*self)[i] &= c;
+	$else
+		*self = (EnumSet)((EnumSetType)*self & (EnumSetType)s);
+	$endif
+}
+
+fn void EnumSet.remove_all(&self, EnumSet s)
+{
+	$if IS_CHAR_ARRAY:
+		foreach (i, c : s) (*self)[i] &= ~c;
+	$else
+		*self = (EnumSet)((EnumSetType)*self & ~(EnumSetType)s);
+	$endif
+}
+
+fn EnumSet EnumSet.and_of(&self, EnumSet s)
+{
+	$if IS_CHAR_ARRAY:
+		EnumSet copy = *self;
+		copy.retain_all(s);
+		return copy;
+	$else
+		return (EnumSet)((EnumSetType)*self & (EnumSetType)s);
+	$endif
+}
+
+fn EnumSet EnumSet.or_of(&self, EnumSet s)
+{
+	$if IS_CHAR_ARRAY:
+		EnumSet copy = *self;
+		copy.add_all(s);
+		return copy;
+	$else
+		return (EnumSet)((EnumSetType)*self | (EnumSetType)s);
+	$endif
+}
+
+
+fn EnumSet EnumSet.diff_of(&self, EnumSet s)
+{
+	$if IS_CHAR_ARRAY:
+		EnumSet copy = *self;
+		copy.remove_all(s);
+		return copy;
+	$else
+		return (EnumSet)((EnumSetType)*self & ~(EnumSetType)s);
+	$endif
+}
+
+fn EnumSet EnumSet.xor_of(&self, EnumSet s)
+{
+	$if IS_CHAR_ARRAY:
+		EnumSet copy = *self;
+		foreach (i, c : s) copy[i] ^= c;
+		return copy;
+	$else
+		return (EnumSet)((EnumSetType)*self ^ (EnumSetType)s);
+	$endif
+}
+
+fn usz! EnumSet.to_format(&set, Formatter* formatter) @dynamic
+{
+	usz n = formatter.print("[")!;
+	bool found;
+	foreach (value : Enum.values)
+	{
+		if (!set.has(value)) continue;
+		if (found) n += formatter.print(", ")!;
+		found = true;
+		n += formatter.printf("%s", value)!;
+	}
+	n += formatter.print("]")!;
+	return n;
+}
+
+fn String EnumSet.to_new_string(&set, Allocator allocator = allocator::heap()) @dynamic
+{
+	return string::format("%s", *set, allocator: allocator);
+}
+
+fn String EnumSet.to_string(&set, Allocator allocator) @dynamic
+{
+	return string::format("%s", *set, allocator: allocator);
+}
+
+fn String EnumSet.to_tstring(&set) @dynamic
+{
+	return string::tformat("%s", *set);
+}
+
+module std::collections::enumset::private;
+
+macro typeid type_for_enum_elements(usz $elements)
+{
+	$switch
+		$case ($elements > 128):
+			return char[($elements + 7) / 8].typeid;
+		$case ($elements > 64):
+			return uint128.typeid;
+		$case ($elements > 32 || $$C_INT_SIZE > 32):
+			return ulong.typeid;
+		$case ($elements > 16 || $$C_INT_SIZE > 16):
+			return uint.typeid;
+		$case ($elements > 8 || $$C_INT_SIZE > 8):
+			return ushort.typeid;
+		$default:
+			return char.typeid;
+	$endswitch
+}

lib/std/collections/hashmap.c3

@@ -0,0 +1,669 @@
+// Copyright (c) 2023 Christoffer Lerno. All rights reserved.
+// Use of this source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+<*
+ @require $defined((Key){}.hash()) `No .hash function found on the key`
+*>
+module std::collections::map(<Key, Value>);
+import std::math;
+import std::io @norecurse;
+
+struct HashMap (Printable)
+{
+	Entry*[] table;
+	Allocator allocator;
+	uint count; // Number of elements
+	uint threshold; // Resize limit
+	float load_factor;
+}
+
+<*
+ @param [&inout] allocator "The allocator to use"
+ @require capacity > 0 "The capacity must be 1 or higher"
+ @require load_factor > 0.0 "The load factor must be higher than 0"
+ @require !self.allocator "Map was already initialized"
+ @require capacity < MAXIMUM_CAPACITY "Capacity cannot exceed maximum"
+*>
+fn HashMap* HashMap.new_init(&self, uint capacity = DEFAULT_INITIAL_CAPACITY, float load_factor = DEFAULT_LOAD_FACTOR, Allocator allocator = null) @deprecated("Use init(mem)")
+{
+	return self.init(allocator ?: allocator::heap(), capacity, load_factor);
+}
+
+<*
+ @param [&inout] allocator "The allocator to use"
+ @require capacity > 0 "The capacity must be 1 or higher"
+ @require load_factor > 0.0 "The load factor must be higher than 0"
+ @require !self.allocator "Map was already initialized"
+ @require capacity < MAXIMUM_CAPACITY "Capacity cannot exceed maximum"
+*>
+fn HashMap* HashMap.init(&self, Allocator allocator, uint capacity = DEFAULT_INITIAL_CAPACITY, float load_factor = DEFAULT_LOAD_FACTOR)
+{
+	capacity = math::next_power_of_2(capacity);
+	self.allocator = allocator;
+	self.load_factor = load_factor;
+	self.threshold = (uint)(capacity * load_factor);
+	self.table = allocator::new_array(allocator, Entry*, capacity);
+	return self;
+}
+
+<*
+ @require capacity > 0 "The capacity must be 1 or higher"
+ @require load_factor > 0.0 "The load factor must be higher than 0"
+ @require !self.allocator "Map was already initialized"
+ @require capacity < MAXIMUM_CAPACITY "Capacity cannot exceed maximum"
+*>
+fn HashMap* HashMap.temp_init(&self, uint capacity = DEFAULT_INITIAL_CAPACITY, float load_factor = DEFAULT_LOAD_FACTOR) @deprecated("Use tinit()")
+{
+	return self.init(allocator::temp(), capacity, load_factor) @inline;
+}
+
+<*
+ @require capacity > 0 "The capacity must be 1 or higher"
+ @require load_factor > 0.0 "The load factor must be higher than 0"
+ @require !self.allocator "Map was already initialized"
+ @require capacity < MAXIMUM_CAPACITY "Capacity cannot exceed maximum"
+*>
+fn HashMap* HashMap.tinit(&self, uint capacity = DEFAULT_INITIAL_CAPACITY, float load_factor = DEFAULT_LOAD_FACTOR)
+{
+	return self.init(allocator::temp(), capacity, load_factor) @inline;
+}
+
+<*
+ @param [&inout] allocator "The allocator to use"
+ @require $vacount % 2 == 0 "There must be an even number of arguments provided for keys and values"
+ @require capacity > 0 "The capacity must be 1 or higher"
+ @require load_factor > 0.0 "The load factor must be higher than 0"
+ @require !self.allocator "Map was already initialized"
+ @require capacity < MAXIMUM_CAPACITY "Capacity cannot exceed maximum"
+*>
+macro HashMap* HashMap.new_init_with_key_values(&self, ..., uint capacity = DEFAULT_INITIAL_CAPACITY, float load_factor = DEFAULT_LOAD_FACTOR, Allocator allocator = allocator::heap()) @deprecated("Use init_with_key_values(mem)")
+{
+	self.init(capacity, load_factor, allocator);
+	$for (var $i = 0; $i < $vacount; $i += 2)
+		self.set($vaarg[$i], $vaarg[$i+1]);
+	$endfor
+	return self;
+}
+
+<*
+ @param [in] keys "The keys for the HashMap entries"
+ @param [in] values "The values for the HashMap entries"
+ @param [&inout] allocator "The allocator to use"
+ @require keys.len == values.len "Both keys and values arrays must be the same length"
+ @require capacity > 0 "The capacity must be 1 or higher"
+ @require load_factor > 0.0 "The load factor must be higher than 0"
+ @require !self.allocator "Map was already initialized"
+ @require capacity < MAXIMUM_CAPACITY "Capacity cannot exceed maximum"
+*>
+fn HashMap* HashMap.new_init_from_keys_and_values(&self, Key[] keys, Value[] values, uint capacity = DEFAULT_INITIAL_CAPACITY, float load_factor = DEFAULT_LOAD_FACTOR, Allocator allocator = allocator::heap()) @deprecated("Use init_from_keys_and_values(mem)")
+{
+	assert(keys.len == values.len);
+	self.init(allocator, capacity, load_factor);
+	for (usz i = 0; i < keys.len; i++)
+	{
+		self.set(keys[i], values[i]);
+	}
+	return self;
+}
+
+<*
+ @require $vacount % 2 == 0 "There must be an even number of arguments provided for keys and values"
+ @require capacity > 0 "The capacity must be 1 or higher"
+ @require load_factor > 0.0 "The load factor must be higher than 0"
+ @require !self.allocator "Map was already initialized"
+ @require capacity < MAXIMUM_CAPACITY "Capacity cannot exceed maximum"
+*>
+macro HashMap* HashMap.temp_init_with_key_values(&self, ..., uint capacity = DEFAULT_INITIAL_CAPACITY, float load_factor = DEFAULT_LOAD_FACTOR) @deprecated("Use tinit_with_key_values")
+{
+	self.tinit(capacity, load_factor);
+	$for (var $i = 0; $i < $vacount; $i += 2)
+		self.set($vaarg[$i], $vaarg[$i+1]);
+	$endfor
+	return self;
+}
+
+<*
+ @require $vacount % 2 == 0 "There must be an even number of arguments provided for keys and values"
+ @require capacity > 0 "The capacity must be 1 or higher"
+ @require load_factor > 0.0 "The load factor must be higher than 0"
+ @require !self.allocator "Map was already initialized"
+ @require capacity < MAXIMUM_CAPACITY "Capacity cannot exceed maximum"
+*>
+macro HashMap* HashMap.tinit_with_key_values(&self, ..., uint capacity = DEFAULT_INITIAL_CAPACITY, float load_factor = DEFAULT_LOAD_FACTOR)
+{
+	self.tinit(capacity, load_factor);
+	$for (var $i = 0; $i < $vacount; $i += 2)
+		self.set($vaarg[$i], $vaarg[$i+1]);
+	$endfor
+	return self;
+}
+
+<*
+ @param [in] keys "The keys for the HashMap entries"
+ @param [in] values "The values for the HashMap entries"
+ @param [&inout] allocator "The allocator to use"
+ @require keys.len == values.len "Both keys and values arrays must be the same length"
+ @require capacity > 0 "The capacity must be 1 or higher"
+ @require load_factor > 0.0 "The load factor must be higher than 0"
+ @require !self.allocator "Map was already initialized"
+ @require capacity < MAXIMUM_CAPACITY "Capacity cannot exceed maximum"
+*>
+fn HashMap* HashMap.temp_init_from_keys_and_values(&self, Key[] keys, Value[] values, uint capacity = DEFAULT_INITIAL_CAPACITY, float load_factor = DEFAULT_LOAD_FACTOR, Allocator allocator = allocator::heap()) @deprecated("Use tinit_from_keys_and_values")
+{
+	assert(keys.len == values.len);
+	self.tinit(capacity, load_factor);
+	for (usz i = 0; i < keys.len; i++)
+	{
+		self.set(keys[i], values[i]);
+	}
+	return self;
+}
+
+<*
+ @param [in] keys "The keys for the HashMap entries"
+ @param [in] values "The values for the HashMap entries"
+ @param [&inout] allocator "The allocator to use"
+ @require keys.len == values.len "Both keys and values arrays must be the same length"
+ @require capacity > 0 "The capacity must be 1 or higher"
+ @require load_factor > 0.0 "The load factor must be higher than 0"
+ @require !self.allocator "Map was already initialized"
+ @require capacity < MAXIMUM_CAPACITY "Capacity cannot exceed maximum"
+*>
+fn HashMap* HashMap.tinit_from_keys_and_values(&self, Key[] keys, Value[] values, uint capacity = DEFAULT_INITIAL_CAPACITY, float load_factor = DEFAULT_LOAD_FACTOR, Allocator allocator = allocator::heap())
+{
+	assert(keys.len == values.len);
+	self.tinit(capacity, load_factor);
+	for (usz i = 0; i < keys.len; i++)
+	{
+		self.set(keys[i], values[i]);
+	}
+	return self;
+}
+
+<*
+ Has this hash map been initialized yet?
+
+ @param [&in] map "The hash map we are testing"
+ @return "Returns true if it has been initialized, false otherwise"
+*>
+fn bool HashMap.is_initialized(&map)
+{
+	return (bool)map.allocator;
+}
+
+<*
+ @param [&in] other_map "The map to copy from."
+*>
+fn HashMap* HashMap.new_init_from_map(&self, HashMap* other_map) @deprecated("Use init_from_map(mem, map)")
+{
+	return self.init_from_map(allocator::heap(), other_map) @inline;
+}
+
+<*
+ @param [&inout] allocator "The allocator to use"
+ @param [&in] other_map "The map to copy from."
+*>
+fn HashMap* HashMap.init_from_map(&self, Allocator allocator, HashMap* other_map)
+{
+	self.init(allocator, other_map.table.len, other_map.load_factor);
+	self.put_all_for_create(other_map);
+	return self;
+}
+
+<*
+ @param [&in] other_map "The map to copy from."
+*>
+fn HashMap* HashMap.temp_init_from_map(&map, HashMap* other_map) @deprecated("Use tinit_from_map")
+{
+	return map.init_from_map(allocator::temp(), other_map) @inline;
+}
+
+<*
+ @param [&in] other_map "The map to copy from."
+*>
+fn HashMap* HashMap.tinit_from_map(&map, HashMap* other_map)
+{
+	return map.init_from_map(allocator::temp(), other_map) @inline;
+}
+
+fn bool HashMap.is_empty(&map) @inline
+{
+	return !map.count;
+}
+
+fn usz HashMap.len(&map) @inline
+{
+	return map.count;
+}
+
+fn Value*! HashMap.get_ref(&map, Key key)
+{
+	if (!map.count) return SearchResult.MISSING?;
+	uint hash = rehash(key.hash());
+	for (Entry *e = map.table[index_for(hash, map.table.len)]; e != null; e = e.next)
+	{
+		if (e.hash == hash && equals(key, e.key)) return &e.value;
+	}
+	return SearchResult.MISSING?;
+}
+
+fn Entry*! HashMap.get_entry(&map, Key key)
+{
+	if (!map.count) return SearchResult.MISSING?;
+	uint hash = rehash(key.hash());
+	for (Entry *e = map.table[index_for(hash, map.table.len)]; e != null; e = e.next)
+	{
+		if (e.hash == hash && equals(key, e.key)) return e;
+	}
+	return SearchResult.MISSING?;
+}
+
+<*
+ Get the value or update and
+ @require $assignable(#expr, Value)
+*>
+macro Value HashMap.@get_or_set(&map, Key key, Value #expr)
+{
+	if (!map.count)
+	{
+		Value val = #expr;
+		map.set(key, val);
+		return val;
+	}
+	uint hash = rehash(key.hash());
+	uint index = index_for(hash, map.table.len);
+	for (Entry *e = map.table[index]; e != null; e = e.next)
+	{
+		if (e.hash == hash && equals(key, e.key)) return e.value;
+	}
+	Value val = #expr;
+	map.add_entry(hash, key, val, index);
+	return val;
+}
+
+fn Value! HashMap.get(&map, Key key) @operator([])
+{
+	return *map.get_ref(key) @inline;
+}
+
+fn bool HashMap.has_key(&map, Key key)
+{
+	return @ok(map.get_ref(key));
+}
+
+fn bool HashMap.set(&map, Key key, Value value) @operator([]=)
+{
+	// If the map isn't initialized, use the defaults to initialize it.
+	if (!map.allocator)
+	{
+		map.init(allocator::heap());
+	}
+	uint hash = rehash(key.hash());
+	uint index = index_for(hash, map.table.len);
+	for (Entry *e = map.table[index]; e != null; e = e.next)
+	{
+		if (e.hash == hash && equals(key, e.key))
+		{
+			e.value = value;
+			return true;
+		}
+	}
+	map.add_entry(hash, key, value, index);
+	return false;
+}
+
+fn void! HashMap.remove(&map, Key key) @maydiscard
+{
+	if (!map.remove_entry_for_key(key)) return SearchResult.MISSING?;
+}
+
+fn void HashMap.clear(&map)
+{
+	if (!map.count) return;
+	foreach (Entry** &entry_ref : map.table)
+	{
+		Entry* entry = *entry_ref;
+		if (!entry) continue;
+		Entry *next = entry.next;
+		while (next)
+		{
+			Entry *to_delete = next;
+			next = next.next;
+			map.free_entry(to_delete);
+		}
+		map.free_entry(entry);
+		*entry_ref = null;
+	}
+	map.count = 0;
+}
+
+fn void HashMap.free(&map)
+{
+	if (!map.allocator) return;
+	map.clear();
+	map.free_internal(map.table.ptr);
+	map.table = {};
+}
+
+fn Key[] HashMap.tcopy_keys(&map)
+{
+	return map.copy_keys(allocator::temp()) @inline;
+}
+
+fn Key[] HashMap.key_tlist(&map) @deprecated("Use 'tcopy_keys'")
+{
+	return map.copy_keys(allocator::temp()) @inline;
+}
+
+<*
+ @deprecated "use copy_keys"
+*>
+fn Key[] HashMap.key_new_list(&map, Allocator allocator = allocator::heap())
+{
+	return map.copy_keys(allocator) @inline;
+}
+
+fn Key[] HashMap.copy_keys(&map, Allocator allocator = allocator::heap())
+{
+	if (!map.count) return {};
+
+	Key[] list = allocator::alloc_array(allocator, Key, map.count);
+	usz index = 0;
+	foreach (Entry* entry : map.table)
+	{
+		while (entry)
+		{
+			$if COPY_KEYS:
+				list[index++] = entry.key.copy(allocator);
+			$else
+				list[index++] = entry.key;
+			$endif
+			entry = entry.next;
+		}
+	}
+	return list;
+}
+
+macro HashMap.@each(map; @body(key, value))
+{
+	map.@each_entry(; Entry* entry) {
+		@body(entry.key, entry.value);
+	};
+}
+
+macro HashMap.@each_entry(map; @body(entry))
+{
+	if (map.count)
+	{
+		foreach (Entry* entry : map.table)
+		{
+			while (entry)
+			{
+				@body(entry);
+				entry = entry.next;
+			}
+		}
+	}
+}
+
+<*
+ @deprecated `use tcopy_values`
+*>
+fn Value[] HashMap.value_tlist(&map)
+{
+	return map.copy_values(allocator::temp()) @inline;
+}
+
+fn Value[] HashMap.tcopy_values(&map)
+{
+	return map.copy_values(allocator::temp()) @inline;
+}
+
+<*
+ @deprecated `use copy_values`
+*>
+fn Value[] HashMap.value_new_list(&map, Allocator allocator = allocator::heap())
+{
+	return map.copy_values(allocator);
+}
+
+fn Value[] HashMap.copy_values(&map, Allocator allocator = allocator::heap())
+{
+	if (!map.count) return {};
+	Value[] list = allocator::alloc_array(allocator, Value, map.count);
+	usz index = 0;
+	foreach (Entry* entry : map.table)
+	{
+		while (entry)
+		{
+			list[index++] = entry.value;
+			entry = entry.next;
+		}
+	}
+	return list;
+}
+
+fn bool HashMap.has_value(&map, Value v) @if(VALUE_IS_EQUATABLE)
+{
+	if (!map.count) return false;
+	foreach (Entry* entry : map.table)
+	{
+		while (entry)
+		{
+			if (equals(v, entry.value)) return true;
+			entry = entry.next;
+		}
+	}
+	return false;
+}
+
+fn HashMapIterator HashMap.iter(&self)
+{
+	return { .map = self, .index = -1 };
+}
+
+fn HashMapValueIterator HashMap.value_iter(&self)
+{
+	return { .map = self, .index = -1 };
+}
+
+fn HashMapKeyIterator HashMap.key_iter(&self)
+{
+	return { .map = self, .index = -1 };
+}
+
+// --- private methods
+
+fn void HashMap.add_entry(&map, uint hash, Key key, Value value, uint bucket_index) @private
+{
+	$if COPY_KEYS:
+	key = key.copy(map.allocator);
+	$endif
+	Entry* entry = allocator::new(map.allocator, Entry, { .hash = hash, .key = key, .value = value, .next = map.table[bucket_index] });
+	map.table[bucket_index] = entry;
+	if (map.count++ >= map.threshold)
+	{
+		map.resize(map.table.len * 2);
+	}
+}
+
+fn void HashMap.resize(&map, uint new_capacity) @private
+{
+	Entry*[] old_table = map.table;
+	uint old_capacity = old_table.len;
+	if (old_capacity == MAXIMUM_CAPACITY)
+	{
+		map.threshold = uint.max;
+		return;
+	}
+	Entry*[] new_table = allocator::new_array(map.allocator, Entry*, new_capacity);
+	map.transfer(new_table);
+	map.table = new_table;
+	map.free_internal(old_table.ptr);
+	map.threshold = (uint)(new_capacity * map.load_factor);
+}
+
+fn usz! HashMap.to_format(&self, Formatter* f) @dynamic
+{
+	usz len;
+	len += f.print("{ ")!;
+	self.@each_entry(; Entry* entry)
+	{
+		if (len > 2) len += f.print(", ")!;
+		len += f.printf("%s: %s", entry.key, entry.value)!;
+    };
+	return len + f.print(" }");
+}
+
+fn void HashMap.transfer(&map, Entry*[] new_table) @private
+{
+	Entry*[] src = map.table;
+	uint new_capacity = new_table.len;
+	foreach (uint j, Entry *e : src)
+	{
+		if (!e) continue;
+		do
+		{
+			Entry* next = e.next;
+			uint i = index_for(e.hash, new_capacity);
+			e.next = new_table[i];
+			new_table[i] = e;
+			e = next;
+		}
+		while (e);
+	}
+}
+
+fn void HashMap.put_all_for_create(&map, HashMap* other_map) @private
+{
+	if (!other_map.count) return;
+	foreach (Entry *e : other_map.table)
+	{
+		while (e)
+		{
+			map.put_for_create(e.key, e.value);
+			e = e.next;
+		}
+	}
+}
+
+fn void HashMap.put_for_create(&map, Key key, Value value) @private
+{
+	uint hash = rehash(key.hash());
+	uint i = index_for(hash, map.table.len);
+	for (Entry *e = map.table[i]; e != null; e = e.next)
+	{
+		if (e.hash == hash && equals(key, e.key))
+		{
+			e.value = value;
+			return;
+		}
+	}
+	map.create_entry(hash, key, value, i);
+}
+
+fn void HashMap.free_internal(&map, void* ptr) @inline @private
+{
+	allocator::free(map.allocator, ptr);
+}
+
+fn bool HashMap.remove_entry_for_key(&map, Key key) @private
+{
+	if (!map.count) return false;
+	uint hash = rehash(key.hash());
+	uint i = index_for(hash, map.table.len);
+	Entry* prev = map.table[i];
+	Entry* e = prev;
+	while (e)
+	{
+		Entry *next = e.next;
+		if (e.hash == hash && equals(key, e.key))
+		{
+			map.count--;
+			if (prev == e)
+			{
+				map.table[i] = next;
+			}
+			else
+			{
+				prev.next = next;
+			}
+			map.free_entry(e);
+			return true;
+		}
+		prev = e;
+		e = next;
+	}
+	return false;
+}
+
+fn void HashMap.create_entry(&map, uint hash, Key key, Value value, int bucket_index) @private
+{
+	Entry *e = map.table[bucket_index];
+	$if COPY_KEYS:
+	key = key.copy(map.allocator);
+	$endif
+	Entry* entry = allocator::new(map.allocator, Entry, { .hash = hash, .key = key, .value = value, .next = map.table[bucket_index] });
+	map.table[bucket_index] = entry;
+	map.count++;
+}
+
+fn void HashMap.free_entry(&self, Entry *entry) @local
+{
+	$if COPY_KEYS:
+	allocator::free(self.allocator, entry.key);
+	$endif
+	self.free_internal(entry);
+}
+
+
+struct HashMapIterator
+{
+	HashMap* map;
+	int top_index;
+	int index;
+	Entry* current_entry;
+}
+
+distinct HashMapValueIterator = HashMapIterator;
+distinct HashMapKeyIterator = HashMapIterator;
+
+
+<*
+ @require idx < self.map.count
+*>
+fn Entry HashMapIterator.get(&self, usz idx) @operator([])
+{
+	if (idx < self.index)
+	{
+		self.top_index = 0;
+		self.current_entry = null;
+		self.index = -1;
+	}
+	while (self.index != idx)
+	{
+		if (self.current_entry)
+		{
+			self.current_entry = self.current_entry.next;
+			if (self.current_entry) self.index++;
+			continue;
+		}
+		self.current_entry = self.map.table[self.top_index++];
+		if (self.current_entry) self.index++;
+	}
+	return *self.current_entry;
+}
+
+fn Value HashMapValueIterator.get(&self, usz idx) @operator([])
+{
+	return ((HashMapIterator*)self).get(idx).value;
+}
+
+fn Key HashMapKeyIterator.get(&self, usz idx) @operator([])
+{
+	return ((HashMapIterator*)self).get(idx).key;
+}
+
+fn usz HashMapValueIterator.len(self) @operator(len) => self.map.count;
+fn usz HashMapKeyIterator.len(self) @operator(len) => self.map.count;
+fn usz HashMapIterator.len(self) @operator(len) => self.map.count;

lib/std/collections/linkedlist.c3

@@ -0,0 +1,348 @@
+// Copyright (c) 2021-2024 Christoffer Lerno. All rights reserved.
+// Use of self source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+module std::collections::linkedlist(<Type>);
+
+const ELEMENT_IS_EQUATABLE = types::is_equatable_type(Type);
+
+struct Node @private
+{
+	Node *next;
+	Node *prev;
+	Type value;
+}
+
+struct LinkedList
+{
+	Allocator allocator;
+	usz size;
+	Node *_first;
+	Node *_last;
+}
+
+<*
+ @param [&inout] allocator "The allocator to use, defaults to the heap allocator"
+ @return "the initialized list"
+*>
+fn LinkedList* LinkedList.init(&self, Allocator allocator)
+{
+	*self = { .allocator = allocator };
+	return self;
+}
+
+<*
+ @return "the initialized list"
+*>
+fn LinkedList* LinkedList.new_init(&self) @deprecated("Use init(mem)")
+{
+	return self.init(allocator::heap()) @inline;
+}
+
+
+fn LinkedList* LinkedList.temp_init(&self) @deprecated("Use tinit()")
+{
+	return self.init(allocator::temp()) @inline;
+}
+
+fn LinkedList* LinkedList.tinit(&self)
+{
+	return self.init(allocator::temp()) @inline;
+}
+
+<*
+ @require self.allocator != null
+*>
+macro void LinkedList.free_node(&self, Node* node) @private
+{
+	allocator::free(self.allocator, node);
+}
+
+macro Node* LinkedList.alloc_node(&self) @private
+{
+	if (!self.allocator) self.allocator = allocator::heap();
+	return allocator::alloc(self.allocator, Node);
+}
+
+fn void LinkedList.push_front(&self, Type value)
+{
+	Node *first = self._first;
+	Node *new_node = self.alloc_node();
+	*new_node = { .next = first, .value = value };
+	self._first = new_node;
+	if (!first)
+	{
+		self._last = new_node;
+	}
+	else
+	{
+		first.prev = new_node;
+	}
+	self.size++;
+}
+
+fn void LinkedList.push(&self, Type value)
+{
+	Node *last = self._last;
+	Node *new_node = self.alloc_node();
+	*new_node = { .prev = last, .value = value };
+	self._last = new_node;
+	if (!last)
+	{
+		self._first = new_node;
+	}
+	else
+	{
+		last.next = new_node;
+	}
+	self.size++;
+}
+
+fn Type! LinkedList.peek(&self) => self.first() @inline;
+fn Type! LinkedList.peek_last(&self) => self.last() @inline;
+
+fn Type! LinkedList.first(&self)
+{
+	if (!self._first) return IteratorResult.NO_MORE_ELEMENT?;
+	return self._first.value;
+}
+
+fn Type! LinkedList.last(&self)
+{
+	if (!self._last) return IteratorResult.NO_MORE_ELEMENT?;
+	return self._last.value;
+}
+
+fn void LinkedList.free(&self) => self.clear() @inline;
+
+fn void LinkedList.clear(&self)
+{
+	for (Node* node = self._first; node != null;)
+	{
+		Node* next = node.next;
+		self.free_node(node);
+		node = next;
+	}
+	self._first = null;
+	self._last = null;
+	self.size = 0;
+}
+
+fn usz LinkedList.len(&self) @inline => self.size;
+
+<*
+ @require index < self.size
+*>
+macro Node* LinkedList.node_at_index(&self, usz index)
+{
+	if (index * 2 >= self.size)
+	{
+		Node* node = self._last;
+		index = self.size - index - 1;
+		while (index--) node = node.prev;
+		return node;
+	}
+	Node* node = self._first;
+	while (index--) node = node.next;
+	return node;
+}
+<*
+ @require index < self.size
+*>
+fn Type LinkedList.get(&self, usz index)
+{
+	return self.node_at_index(index).value;
+}
+
+<*
+ @require index < self.size
+*>
+fn void LinkedList.set(&self, usz index, Type element)
+{
+	self.node_at_index(index).value = element;
+}
+
+<*
+ @require index < self.size
+*>
+fn void LinkedList.remove_at(&self, usz index)
+{
+	self.unlink(self.node_at_index(index));
+}
+
+<*
+ @require index <= self.size
+*>
+fn void LinkedList.insert_at(&self, usz index, Type element)
+{
+	switch (index)
+	{
+		case 0:
+			self.push_front(element);
+		case self.size:
+			self.push(element);
+		default:
+			self.link_before(self.node_at_index(index), element);
+	}
+}
+<*
+ @require succ != null
+*>
+fn void LinkedList.link_before(&self, Node *succ, Type value) @private
+{
+	Node* pred = succ.prev;
+	Node* new_node = self.alloc_node();
+	*new_node = { .prev = pred, .next = succ, .value = value };
+	succ.prev = new_node;
+	if (!pred)
+	{
+		self._first = new_node;
+	}
+	else
+	{
+		pred.next = new_node;
+	}
+	self.size++;
+}
+
+<*
+ @require self._first != null
+*>
+fn void LinkedList.unlink_first(&self) @private
+{
+	Node* f = self._first;
+	Node* next = f.next;
+	self.free_node(f);
+	self._first = next;
+	if (!next)
+	{
+		self._last = null;
+	}
+	else
+	{
+		next.prev = null;
+	}
+	self.size--;
+}
+
+fn usz LinkedList.remove(&self, Type t) @if(ELEMENT_IS_EQUATABLE)
+{
+	usz start = self.size;
+	Node* node = self._first;
+	while (node)
+	{
+		switch
+		{
+			case equals(node.value, t):
+				Node* next = node.next;
+				self.unlink(node);
+				node = next;
+			default:
+				node = node.next;
+		}
+	}
+	return start - self.size;
+}
+
+fn Type! LinkedList.pop(&self)
+{
+	if (!self._last) return IteratorResult.NO_MORE_ELEMENT?;
+	defer self.unlink_last();
+	return self._last.value;
+}
+
+fn bool LinkedList.is_empty(&self)
+{
+	return !self._first;
+}
+
+fn Type! LinkedList.pop_front(&self)
+{
+	if (!self._first) return IteratorResult.NO_MORE_ELEMENT?;
+	defer self.unlink_first();
+	return self._first.value;
+}
+
+fn void! LinkedList.remove_last(&self) @maydiscard
+{
+	if (!self._first) return IteratorResult.NO_MORE_ELEMENT?;
+	self.unlink_last();
+}
+
+fn void! LinkedList.remove_first(&self) @maydiscard
+{
+	if (!self._first) return IteratorResult.NO_MORE_ELEMENT?;
+	self.unlink_first();
+}
+
+
+fn bool LinkedList.remove_first_match(&self, Type t) @if(ELEMENT_IS_EQUATABLE)
+{
+	for (Node* node = self._first; node != null; node = node.next)
+	{
+		if (node.value == t)
+		{
+			self.unlink(node);
+			return true;
+		}
+	}
+	return false;
+}
+
+fn bool LinkedList.remove_last_match(&self, Type t)  @if(ELEMENT_IS_EQUATABLE)
+{
+	for (Node* node = self._last; node != null; node = node.prev)
+	{
+		if (node.value == t)
+		{
+			self.unlink(node);
+			return true;
+		}
+	}
+	return false;
+}
+<*
+ @require self._last != null
+*>
+fn void LinkedList.unlink_last(&self) @inline @private
+{
+	Node* l = self._last;
+	Node* prev = l.prev;
+	self._last = prev;
+	self.free_node(l);
+	if (!prev)
+	{
+		self._first = null;
+	}
+	else
+	{
+		prev.next = null;
+	}
+	self.size--;
+}
+
+<*
+ @require x != null
+*>
+fn void LinkedList.unlink(&self, Node* x) @private
+{
+	Node* next = x.next;
+	Node* prev = x.prev;
+	if (!prev)
+	{
+		self._first = next;
+	}
+	else
+	{
+		prev.next = next;
+	}
+	if (!next)
+	{
+		self._last = prev;
+	}
+	else
+	{
+		next.prev = prev;
+	}
+	self.free_node(x);
+	self.size--;
+}

lib/std/collections/list.c3

@@ -0,0 +1,627 @@
+// Copyright (c) 2021-2024 Christoffer Lerno. All rights reserved.
+// Use of self source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+module std::collections::list(<Type>);
+import std::io, std::math, std::collections::list_common;
+
+def ElementPredicate = fn bool(Type *type);
+def ElementTest = fn bool(Type *type, any context);
+const ELEMENT_IS_EQUATABLE = types::is_equatable_type(Type);
+const ELEMENT_IS_POINTER = Type.kindof == POINTER;
+
+macro type_is_overaligned() => Type.alignof > mem::DEFAULT_MEM_ALIGNMENT;
+
+struct List (Printable)
+{
+	usz size;
+	usz capacity;
+	Allocator allocator;
+	Type *entries;
+}
+
+<*
+ @param initial_capacity "The initial capacity to reserve"
+ @param [&inout] allocator "The allocator to use, defaults to the heap allocator"
+*>
+fn List* List.init(&self, Allocator allocator, usz initial_capacity = 16)
+{
+	self.allocator = allocator;
+	self.size = 0;
+	self.capacity = 0;
+	self.entries = null;
+	self.reserve(initial_capacity);
+	return self;
+}
+
+<*
+ @param initial_capacity "The initial capacity to reserve"
+ @param [&inout] allocator "The allocator to use, defaults to the heap allocator"
+*>
+fn List* List.new_init(&self, usz initial_capacity = 16, Allocator allocator = allocator::heap()) @deprecated("Use init(mem)")
+{
+	self.allocator = allocator;
+	self.size = 0;
+	self.capacity = 0;
+	self.entries = null;
+	self.reserve(initial_capacity);
+	return self;
+}
+
+<*
+ Initialize the list using the temp allocator.
+
+ @param initial_capacity "The initial capacity to reserve"
+*>
+fn List* List.temp_init(&self, usz initial_capacity = 16) @deprecated("Use tinit()")
+{
+	return self.init(allocator::temp(), initial_capacity) @inline;
+}
+
+<*
+ Initialize the list using the temp allocator.
+
+ @param initial_capacity "The initial capacity to reserve"
+*>
+fn List* List.tinit(&self, usz initial_capacity = 16)
+{
+	return self.init(allocator::temp(), initial_capacity) @inline;
+}
+
+<*
+ Initialize a new list with an array.
+
+ @param [in] values `The values to initialize the list with.`
+ @require self.size == 0 "The List must be empty"
+*>
+fn List* List.new_init_with_array(&self, Type[] values, Allocator allocator = allocator::heap()) @deprecated("Use init_with_array(mem)")
+{
+	return self.init_with_array(allocator, values);
+}
+
+<*
+ Initialize a new list with an array.
+
+ @param [in] values `The values to initialize the list with.`
+ @require self.size == 0 "The List must be empty"
+*>
+fn List* List.init_with_array(&self, Allocator allocator, Type[] values)
+{
+	self.init(allocator, values.len) @inline;
+	self.add_array(values) @inline;
+	return self;
+}
+
+<*
+ Initialize a temporary list with an array.
+
+ @param [in] values `The values to initialize the list with.`
+ @require self.size == 0 "The List must be empty"
+*>
+fn List* List.temp_init_with_array(&self, Type[] values) @deprecated("Use tinit_with_array()")
+{
+	self.tinit(values.len) @inline;
+	self.add_array(values) @inline;
+	return self;
+}
+
+<*
+ Initialize a temporary list with an array.
+
+ @param [in] values `The values to initialize the list with.`
+ @require self.size == 0 "The List must be empty"
+*>
+fn List* List.tinit_with_array(&self, Type[] values)
+{
+	self.tinit(values.len) @inline;
+	self.add_array(values) @inline;
+	return self;
+}
+
+<*
+ @require self.capacity == 0 "The List must not be allocated"
+*>
+fn void List.init_wrapping_array(&self, Type[] types, Allocator allocator = allocator::heap())
+{
+	self.allocator = allocator;
+	self.capacity = types.len;
+	self.entries = types.ptr;
+	self.set_size(types.len);
+}
+
+fn usz! List.to_format(&self, Formatter* formatter) @dynamic
+{
+	switch (self.size)
+	{
+		case 0:
+			return formatter.print("[]")!;
+		case 1:
+			return formatter.printf("[%s]", self.entries[0])!;
+		default:
+			usz n = formatter.print("[")!;
+			foreach (i, element : self.entries[:self.size])
+			{
+				if (i != 0) formatter.print(", ")!;
+				n += formatter.printf("%s", element)!;
+			}
+			n += formatter.print("]")!;
+			return n;
+	}
+}
+
+fn String List.to_new_string(&self, Allocator allocator = allocator::heap()) @dynamic
+{
+	return string::format("%s", *self, allocator: allocator);
+}
+
+fn String List.to_tstring(&self)
+{
+	return string::tformat("%s", *self);
+}
+
+fn void List.push(&self, Type element) @inline
+{
+	self.reserve(1);
+	self.entries[self.set_size(self.size + 1)] = element;
+}
+
+fn Type! List.pop(&self)
+{
+	if (!self.size) return IteratorResult.NO_MORE_ELEMENT?;
+	defer self.set_size(self.size - 1);
+	return self.entries[self.size - 1];
+}
+
+fn void List.clear(&self)
+{
+	self.set_size(0);
+}
+
+fn Type! List.pop_first(&self)
+{
+	if (!self.size) return IteratorResult.NO_MORE_ELEMENT?;
+	defer self.remove_at(0);
+	return self.entries[0];
+}
+
+<*
+ @require index < self.size `Removed element out of bounds`
+*>
+fn void List.remove_at(&self, usz index)
+{
+	self.set_size(self.size - 1);
+	if (!self.size || index == self.size) return;
+	self.entries[index .. self.size - 1] = self.entries[index + 1 .. self.size];
+}
+
+fn void List.add_all(&self, List* other_list)
+{
+	if (!other_list.size) return;
+	self.reserve(other_list.size);
+	usz index = self.set_size(self.size + other_list.size);
+	foreach (&value : other_list)
+	{
+		self.entries[index++] = *value;
+	}
+}
+
+
+<*
+ IMPORTANT The returned array must be freed using free_aligned.
+*>
+fn Type[] List.to_new_aligned_array(&self, Allocator allocator = allocator::heap())
+{
+	return list_common::list_to_new_aligned_array(Type, self, allocator);
+}
+
+<*
+ @require !type_is_overaligned() : "This function is not available on overaligned types"
+*>
+macro Type[] List.to_new_array(&self, Allocator allocator = allocator::heap())
+{
+	return list_common::list_to_new_array(Type, self, allocator);
+}
+
+fn Type[] List.to_tarray(&self)
+{
+	$if type_is_overaligned():
+	return self.to_new_aligned_array(allocator::temp());
+	$else
+	return self.to_new_array(allocator::temp());
+	$endif;
+}
+
+<*
+ Reverse the elements in a list.
+*>
+fn void List.reverse(&self)
+{
+	list_common::list_reverse(self);
+}
+
+fn Type[] List.array_view(&self)
+{
+	return self.entries[:self.size];
+}
+
+<*
+ Add the values of an array to this list.
+
+ @param [in] array
+ @ensure self.size >= array.len
+*>
+fn void List.add_array(&self, Type[] array)
+{
+	if (!array.len) return;
+	self.reserve(array.len);
+	usz index = self.set_size(self.size + array.len);
+	self.entries[index : array.len] = array[..];
+}
+
+fn void List.push_front(&self, Type type) @inline
+{
+	self.insert_at(0, type);
+}
+
+<*
+ @require index <= self.size `Insert was out of bounds`
+*>
+fn void List.insert_at(&self, usz index, Type type)
+{
+	self.reserve(1);
+	self.set_size(self.size + 1);
+	for (isz i = self.size - 1; i > index; i--)
+	{
+		self.entries[i] = self.entries[i - 1];
+	}
+	self.entries[index] = type;
+}
+
+<*
+ @require index < self.size
+*>
+fn void List.set_at(&self, usz index, Type type)
+{
+	self.entries[index] = type;
+}
+
+fn void! List.remove_last(&self) @maydiscard
+{
+	if (!self.size) return IteratorResult.NO_MORE_ELEMENT?;
+	self.set_size(self.size - 1);
+}
+
+fn void! List.remove_first(&self) @maydiscard
+{
+	if (!self.size) return IteratorResult.NO_MORE_ELEMENT?;
+	self.remove_at(0);
+}
+
+fn Type! List.first(&self)
+{
+	if (!self.size) return IteratorResult.NO_MORE_ELEMENT?;
+	return self.entries[0];
+}
+
+fn Type! List.last(&self)
+{
+	if (!self.size) return IteratorResult.NO_MORE_ELEMENT?;
+	return self.entries[self.size - 1];
+}
+
+fn bool List.is_empty(&self) @inline
+{
+	return !self.size;
+}
+
+fn usz List.byte_size(&self) @inline
+{
+	return Type.sizeof * self.size;
+}
+
+fn usz List.len(&self) @operator(len) @inline
+{
+	return self.size;
+}
+
+<*
+ @require index < self.size `Access out of bounds`
+*>
+fn Type List.get(&self, usz index) @inline
+{
+	return self.entries[index];
+}
+
+fn void List.free(&self)
+{
+	if (!self.allocator || !self.capacity) return;
+
+	self.pre_free(); // Remove sanitizer annotation
+
+	$if type_is_overaligned():
+		allocator::free_aligned(self.allocator, self.entries);
+	$else
+		allocator::free(self.allocator, self.entries);
+	$endif;
+	self.capacity = 0;
+	self.size = 0;
+	self.entries = null;
+}
+
+<*
+ @require i < self.size && j < self.size `Access out of bounds`
+*>
+fn void List.swap(&self, usz i, usz j)
+{
+	@swap(self.entries[i], self.entries[j]);
+}
+
+<*
+ @param filter "The function to determine if it should be removed or not"
+ @return "the number of deleted elements"
+*>
+fn usz List.remove_if(&self, ElementPredicate filter)
+{
+	return list_common::list_remove_if(self, filter, false);
+}
+
+<*
+ @param selection "The function to determine if it should be kept or not"
+ @return "the number of deleted elements"
+*>
+fn usz List.retain_if(&self, ElementPredicate selection)
+{
+	return list_common::list_remove_if(self, selection, true);
+}
+
+fn usz List.remove_using_test(&self, ElementTest filter, any context)
+{
+	usz old_size = self.size;
+	defer
+	{
+		if (old_size != self.size) self._update_size_change(old_size, self.size);
+	}
+	return list_common::list_remove_using_test(self, filter, false, context);
+}
+
+
+
+fn usz List.retain_using_test(&self, ElementTest filter, any context)
+{
+	usz old_size = self.size;
+	defer {
+		if (old_size != self.size) self._update_size_change(old_size, self.size);
+	}
+	return list_common::list_remove_using_test(self, filter, true, context);
+}
+
+fn void List.ensure_capacity(&self, usz min_capacity) @local
+{
+	if (!min_capacity) return;
+	if (self.capacity >= min_capacity) return;
+	if (!self.allocator) self.allocator = allocator::heap();
+
+	self.pre_free(); // Remove sanitizer annotation
+
+	min_capacity = math::next_power_of_2(min_capacity);
+	$if type_is_overaligned():
+		self.entries = allocator::realloc_aligned(self.allocator, self.entries, Type.sizeof * min_capacity, alignment: Type[1].alignof)!!;
+	$else
+		self.entries = allocator::realloc(self.allocator, self.entries, Type.sizeof * min_capacity);
+	$endif;
+	self.capacity = min_capacity;
+
+	self.post_alloc(); // Add sanitizer annotation
+}
+
+<*
+ @require index < self.size `Access out of bounds`
+*>
+macro Type List.@item_at(&self, usz index) @operator([])
+{
+	return self.entries[index];
+}
+
+<*
+ @require index < self.size `Access out of bounds`
+*>
+fn Type* List.get_ref(&self, usz index) @operator(&[]) @inline
+{
+	return &self.entries[index];
+}
+
+<*
+ @require index < self.size `Access out of bounds`
+*>
+fn void List.set(&self, usz index, Type value) @operator([]=)
+{
+	self.entries[index] = value;
+}
+
+fn void List.reserve(&self, usz added)
+{
+	usz new_size = self.size + added;
+	if (self.capacity >= new_size) return;
+
+	assert(new_size < usz.max / 2U);
+	usz new_capacity = self.capacity ? 2U * self.capacity : 16U;
+	while (new_capacity < new_size) new_capacity *= 2U;
+	self.ensure_capacity(new_capacity);
+}
+
+fn void List._update_size_change(&self,usz old_size, usz new_size)
+{
+	if (old_size == new_size) return;
+	sanitizer::annotate_contiguous_container(self.entries,
+	                                         &self.entries[self.capacity],
+	                                         &self.entries[old_size],
+	                                         &self.entries[new_size]);
+}
+<*
+ @require new_size == 0 || self.capacity != 0
+*>
+fn usz List.set_size(&self, usz new_size) @inline @private
+{
+	usz old_size = self.size;
+	self._update_size_change(old_size, new_size);
+	self.size = new_size;
+	return old_size;
+}
+
+macro void List.pre_free(&self) @private
+{
+	if (!self.capacity) return;
+	self._update_size_change(self.size, self.capacity);
+}
+
+<*
+ @require self.capacity > 0
+*>
+macro void List.post_alloc(&self) @private
+{
+	self._update_size_change(self.capacity, self.size);
+}
+
+// Functions for equatable types
+
+
+fn usz! List.index_of(&self, Type type) @if(ELEMENT_IS_EQUATABLE)
+{
+	foreach (i, v : self)
+	{
+		if (equals(v, type)) return i;
+	}
+	return SearchResult.MISSING?;
+}
+
+fn usz! List.rindex_of(&self, Type type) @if(ELEMENT_IS_EQUATABLE)
+{
+	foreach_r (i, v : self)
+	{
+		if (equals(v, type)) return i;
+	}
+	return SearchResult.MISSING?;
+}
+
+fn bool List.equals(&self, List other_list) @if(ELEMENT_IS_EQUATABLE)
+{
+	if (self.size != other_list.size) return false;
+	foreach (i, v : self)
+	{
+		if (!equals(v, other_list.entries[i])) return false;
+	}
+	return true;
+}
+
+<*
+ Check for presence of a value in a list.
+
+ @param [&in] self "the list to find elements in"
+ @param value "The value to search for"
+ @return "True if the value is found, false otherwise"
+*>
+fn bool List.contains(&self, Type value) @if(ELEMENT_IS_EQUATABLE)
+{
+	foreach (i, v : self)
+	{
+		if (equals(v, value)) return true;
+	}
+	return false;
+}
+
+<*
+ @param [&inout] self "The list to remove elements from"
+ @param value "The value to remove"
+ @return "true if the value was found"
+*>
+fn bool List.remove_last_item(&self, Type value) @if(ELEMENT_IS_EQUATABLE)
+{
+	return @ok(self.remove_at(self.rindex_of(value)));
+}
+
+<*
+ @param [&inout] self "The list to remove elements from"
+ @param value "The value to remove"
+ @return "true if the value was found"
+*>
+fn bool List.remove_first_item(&self, Type value) @if(ELEMENT_IS_EQUATABLE)
+{
+	return @ok(self.remove_at(self.index_of(value)));
+}
+<*
+ @param [&inout] self "The list to remove elements from"
+ @param value "The value to remove"
+ @return "the number of deleted elements."
+*>
+fn usz List.remove_item(&self, Type value) @if(ELEMENT_IS_EQUATABLE)
+{
+	usz old_size = self.size;
+	defer {
+		if (old_size != self.size) self._update_size_change(old_size, self.size);
+	}
+	return list_common::list_remove_item(self, value);
+}
+
+
+
+fn void List.remove_all_from(&self, List* other_list) @if(ELEMENT_IS_EQUATABLE)
+{
+	if (!other_list.size) return;
+	usz old_size = self.size;
+	defer {
+		if (old_size != self.size) self._update_size_change(old_size, self.size);
+	}
+	foreach (v : other_list) self.remove_item(v);
+}
+
+<*
+ @param [&in] self
+ @return "The number non-null values in the list"
+*>
+fn usz List.compact_count(&self) @if(ELEMENT_IS_POINTER)
+{
+	usz vals = 0;
+	foreach (v : self) if (v) vals++;
+	return vals;
+}
+
+fn usz List.compact(&self) @if(ELEMENT_IS_POINTER)
+{
+	usz old_size = self.size;
+	defer {
+		if (old_size != self.size) self._update_size_change(old_size, self.size);
+	}
+	return list_common::list_compact(self);
+}
+
+// --> Deprecated
+
+<*
+ @param [&inout] self "The list to remove elements from"
+ @param value "The value to remove"
+ @return "true if the value was found"
+*>
+fn bool List.remove_last_match(&self, Type value) @if(ELEMENT_IS_EQUATABLE) @deprecated
+{
+	return self.remove_last_item(value) @inline;
+}
+
+<*
+ @param [&inout] self "The list to remove elements from"
+ @param value "The value to remove"
+ @return "true if the value was found"
+*>
+fn bool List.remove_first_match(&self, Type value) @if(ELEMENT_IS_EQUATABLE) @deprecated
+{
+	return self.remove_first_item(value) @inline;
+}
+
+
+<*
+ @param [&inout] self "The list to remove elements from"
+ @param value "The value to remove"
+ @return "the number of deleted elements."
+*>
+fn usz List.remove_all_matches(&self, Type value) @if(ELEMENT_IS_EQUATABLE) @deprecated
+{
+	return self.remove_item(value) @inline;
+}

lib/std/collections/list_common.c3

@@ -0,0 +1,112 @@
+module std::collections::list_common;
+
+<*
+ IMPORTANT The returned array must be freed using free_aligned.
+*>
+macro list_to_new_aligned_array($Type, self, Allocator allocator)
+{
+	if (!self.size) return ($Type[]){};
+	$Type[] result = allocator::alloc_array_aligned(allocator, $Type, self.size);
+	result[..] = self.entries[:self.size];
+	return result;
+}
+
+macro list_to_new_array($Type, self, Allocator allocator)
+{
+	if (!self.size) return ($Type[]){};
+	$Type[] result = allocator::alloc_array(allocator, $Type, self.size);
+	result[..] = self.entries[:self.size];
+	return result;
+}
+
+macro void list_reverse(self)
+{
+	if (self.size < 2) return;
+	usz half = self.size / 2U;
+	usz end = self.size - 1;
+	for (usz i = 0; i < half; i++)
+	{
+		@swap(self.entries[i], self.entries[end - i]);
+	}
+}
+
+macro usz list_remove_using_test(self, filter, bool $invert, ctx)
+{
+	usz size = self.size;
+	for (usz i = size, usz k = size; k > 0; k = i)
+	{
+		// Find last index of item to be deleted.
+		$if $invert:
+			while (i > 0 && !filter(&self.entries[i - 1], ctx)) i--;
+		$else
+			while (i > 0 && filter(&self.entries[i - 1], ctx)) i--;
+		$endif
+		// Remove the items from this index up to the one not to be deleted.
+		usz n = self.size - k;
+		self.entries[i:n] = self.entries[k:n];
+		self.size -= k - i;
+		// Find last index of item not to be deleted.
+		$if $invert:
+			while (i > 0 && filter(&self.entries[i - 1], ctx)) i--;
+		$else
+			while (i > 0 && !filter(&self.entries[i - 1], ctx)) i--;
+		$endif
+	}
+	return size - self.size;
+}
+
+macro usz list_compact(self)
+{
+	usz size = self.size;
+	for (usz i = size; i > 0; i--)
+	{
+		if (self.entries[i - 1]) continue;
+		for (usz j = i; j < size; j++)
+		{
+			self.entries[j - 1] = self.entries[j];
+		}
+		self.size--;
+	}
+	return size - self.size;
+}
+
+macro usz list_remove_item(self, value)
+{
+	usz size = self.size;
+	for (usz i = size; i > 0; i--)
+	{
+		if (!equals(self.entries[i - 1], value)) continue;
+		for (usz j = i; j < self.size; j++)
+		{
+			self.entries[j - 1] = self.entries[j];
+		}
+		self.size--;
+	}
+	return size - self.size;
+}
+
+
+macro usz list_remove_if(self, filter, bool $invert)
+{
+	usz size = self.size;
+	for (usz i = size, usz k = size; k > 0; k = i)
+	{
+		// Find last index of item to be deleted.
+		$if $invert:
+			while (i > 0 && !filter(&self.entries[i - 1])) i--;
+		$else
+			while (i > 0 && filter(&self.entries[i - 1])) i--;
+		$endif
+		// Remove the items from this index up to the one not to be deleted.
+		usz n = self.size - k;
+		self.entries[i:n] = self.entries[k:n];
+		self.size -= k - i;
+		// Find last index of item not to be deleted.
+		$if $invert:
+			while (i > 0 && filter(&self.entries[i - 1])) i--;
+		$else
+			while (i > 0 && !filter(&self.entries[i - 1])) i--;
+		$endif
+	}
+	return size - self.size;
+}

lib/std/collections/map.c3

@@ -0,0 +1,508 @@
+// Copyright (c) 2023 Christoffer Lerno. All rights reserved.
+// Use of this source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+module std::collections::map(<Key, Value>);
+import std::math;
+
+const uint DEFAULT_INITIAL_CAPACITY = 16;
+const uint MAXIMUM_CAPACITY = 1u << 31;
+const float DEFAULT_LOAD_FACTOR = 0.75;
+const VALUE_IS_EQUATABLE = Value.is_eq;
+const bool COPY_KEYS = types::implements_copy(Key);
+
+distinct Map = void*;
+
+struct MapImpl
+{
+	Entry*[] table;
+	Allocator allocator;
+	uint count; // Number of elements
+	uint threshold; // Resize limit
+	float load_factor;
+}
+
+<*
+ @require capacity > 0 "The capacity must be 1 or higher"
+ @require load_factor > 0.0 "The load factor must be higher than 0"
+ @require capacity < MAXIMUM_CAPACITY "Capacity cannot exceed maximum"
+*>
+fn Map new(uint capacity = DEFAULT_INITIAL_CAPACITY, float load_factor = DEFAULT_LOAD_FACTOR, Allocator allocator = allocator::heap()) @deprecated("Map is deprecated")
+{
+	MapImpl* map = allocator::alloc(allocator, MapImpl);
+	_init(map, capacity, load_factor, allocator);
+	return (Map)map;
+}
+
+<*
+ @require capacity > 0 "The capacity must be 1 or higher"
+ @require load_factor > 0.0 "The load factor must be higher than 0"
+ @require capacity < MAXIMUM_CAPACITY "Capacity cannot exceed maximum"
+*>
+fn Map temp(uint capacity = DEFAULT_INITIAL_CAPACITY, float load_factor = DEFAULT_LOAD_FACTOR)
+{
+	MapImpl* map = mem::temp_alloc(MapImpl);
+	_init(map, capacity, load_factor, allocator::temp());
+	return (Map)map;
+}
+
+<*
+ @param [&inout] allocator "The allocator to use"
+ @require $vacount % 2 == 0 "There must be an even number of arguments provided for keys and values"
+ @require capacity > 0 "The capacity must be 1 or higher"
+ @require load_factor > 0.0 "The load factor must be higher than 0"
+ @require capacity < MAXIMUM_CAPACITY "Capacity cannot exceed maximum"
+*>
+macro Map new_init_with_key_values(..., uint capacity = DEFAULT_INITIAL_CAPACITY, float load_factor = DEFAULT_LOAD_FACTOR, Allocator allocator = allocator::heap())
+{
+	Map map = new(capacity, load_factor, allocator);
+	$for (var $i = 0; $i < $vacount; $i += 2)
+		map.set($vaarg[$i], $vaarg[$i+1]);
+	$endfor
+	return map;
+}
+
+<*
+ @param [in] keys "Array of keys for the Map entries"
+ @param [in] values "Array of values for the Map entries"
+ @param [&inout] allocator "The allocator to use"
+ @require keys.len == values.len "Both keys and values arrays must be the same length"
+ @require capacity > 0 "The capacity must be 1 or higher"
+ @require load_factor > 0.0 "The load factor must be higher than 0"
+ @require capacity < MAXIMUM_CAPACITY "Capacity cannot exceed maximum"
+*>
+fn Map new_init_from_keys_and_values(Key[] keys, Value[] values, uint capacity = DEFAULT_INITIAL_CAPACITY, float load_factor = DEFAULT_LOAD_FACTOR, Allocator allocator = allocator::heap())
+{
+	assert(keys.len == values.len);
+	Map map = new(capacity, load_factor, allocator);
+	for (usz i = 0; i < keys.len; i++)
+	{
+		map.set(keys[i], values[i]);
+	}
+	return map;
+}
+
+<*
+ @require $vacount % 2 == 0 "There must be an even number of arguments provided for keys and values"
+ @require capacity > 0 "The capacity must be 1 or higher"
+ @require load_factor > 0.0 "The load factor must be higher than 0"
+ @require capacity < MAXIMUM_CAPACITY "Capacity cannot exceed maximum"
+*>
+macro Map temp_new_with_key_values(..., uint capacity = DEFAULT_INITIAL_CAPACITY, float load_factor = DEFAULT_LOAD_FACTOR)
+{
+	Map map = temp(capacity, load_factor);
+	$for (var $i = 0; $i < $vacount; $i += 2)
+		map.set($vaarg[$i], $vaarg[$i+1]);
+	$endfor
+	return map;
+}
+
+<*
+ @param [in] keys "The keys for the HashMap entries"
+ @param [in] values "The values for the HashMap entries"
+ @param [&inout] allocator "The allocator to use"
+ @require keys.len == values.len "Both keys and values arrays must be the same length"
+ @require capacity > 0 "The capacity must be 1 or higher"
+ @require load_factor > 0.0 "The load factor must be higher than 0"
+ @require capacity < MAXIMUM_CAPACITY "Capacity cannot exceed maximum"
+*>
+fn Map temp_init_from_keys_and_values(Key[] keys, Value[] values, uint capacity = DEFAULT_INITIAL_CAPACITY, float load_factor = DEFAULT_LOAD_FACTOR, Allocator allocator = allocator::heap())
+{
+	assert(keys.len == values.len);
+	Map map = temp(capacity, load_factor);
+	for (usz i = 0; i < keys.len; i++)
+	{
+		map.set(keys[i], values[i]);
+	}
+	return map;
+}
+
+<*
+ @param [&in] other_map "The map to copy from."
+*>
+fn Map new_from_map(Map other_map, Allocator allocator = null)
+{
+	MapImpl* other_map_impl = (MapImpl*)other_map;
+	if (!other_map_impl)
+	{
+		if (allocator) return new(allocator: allocator);
+		return null;
+	}
+	MapImpl* map = (MapImpl*)new(other_map_impl.table.len, other_map_impl.load_factor, allocator ?: allocator::heap());
+	if (!other_map_impl.count) return (Map)map;
+	foreach (Entry *e : other_map_impl.table)
+	{
+		while (e)
+		{
+			map._put_for_create(e.key, e.value);
+			e = e.next;
+		}
+	}
+	return (Map)map;
+}
+
+<*
+ @param [&in] other_map "The map to copy from."
+*>
+fn Map temp_from_map(Map other_map)
+{
+	return new_from_map(other_map, allocator::temp());
+}
+
+fn bool Map.is_empty(map) @inline
+{
+	return !map || !((MapImpl*)map).count;
+}
+
+fn usz Map.len(map) @inline
+{
+	return map ? ((MapImpl*)map).count : 0;
+}
+
+fn Value*! Map.get_ref(self, Key key)
+{
+	MapImpl *map = (MapImpl*)self;
+	if (!map || !map.count) return SearchResult.MISSING?;
+	uint hash = rehash(key.hash());
+	for (Entry *e = map.table[index_for(hash, map.table.len)]; e != null; e = e.next)
+	{
+		if (e.hash == hash && equals(key, e.key)) return &e.value;
+	}
+	return SearchResult.MISSING?;
+}
+
+fn Entry*! Map.get_entry(map, Key key)
+{
+	MapImpl *map_impl = (MapImpl*)map;
+	if (!map_impl || !map_impl.count) return SearchResult.MISSING?;
+	uint hash = rehash(key.hash());
+	for (Entry *e = map_impl.table[index_for(hash, map_impl.table.len)]; e != null; e = e.next)
+	{
+		if (e.hash == hash && equals(key, e.key)) return e;
+	}
+	return SearchResult.MISSING?;
+}
+
+<*
+ Get the value or update and
+ @require $assignable(#expr, Value)
+*>
+macro Value Map.@get_or_set(&self, Key key, Value #expr)
+{
+	MapImpl *map = (MapImpl*)*self;
+	if (!map || !map.count)
+	{
+		Value val = #expr;
+		map.set(key, val);
+		return val;
+	}
+	uint hash = rehash(key.hash());
+	uint index = index_for(hash, map.table.len);
+	for (Entry *e = map.table[index]; e != null; e = e.next)
+	{
+		if (e.hash == hash && equals(key, e.key)) return e.value;
+	}
+	Value val = #expr;
+	map.add_entry(hash, key, val, index);
+	return val;
+}
+
+fn Value! Map.get(map, Key key) @operator([])
+{
+	return *map.get_ref(key) @inline;
+}
+
+fn bool Map.has_key(map, Key key)
+{
+	return @ok(map.get_ref(key));
+}
+
+macro Value Map.set_value_return(&map, Key key, Value value) @operator([]=)
+{
+	map.set(key, value);
+	return value;
+}
+
+fn bool Map.set(&self, Key key, Value value)
+{
+	// If the map isn't initialized, use the defaults to initialize it.
+	if (!*self) *self = new();
+	MapImpl* map = (MapImpl*)*self;
+	uint hash = rehash(key.hash());
+	uint index = index_for(hash, map.table.len);
+	for (Entry *e = map.table[index]; e != null; e = e.next)
+	{
+		if (e.hash == hash && equals(key, e.key))
+		{
+			e.value = value;
+			return true;
+		}
+	}
+	map._add_entry(hash, key, value, index);
+	return false;
+}
+
+fn void! Map.remove(map, Key key) @maydiscard
+{
+	if (!map || !((MapImpl*)map)._remove_entry_for_key(key)) return SearchResult.MISSING?;
+}
+
+fn void Map.clear(self)
+{
+	MapImpl* map = (MapImpl*)self;
+	if (!map || !map.count) return;
+	foreach (Entry** &entry_ref : map.table)
+	{
+		Entry* entry = *entry_ref;
+		if (!entry) continue;
+		Entry *next = entry.next;
+		while (next)
+		{
+			Entry *to_delete = next;
+			next = next.next;
+			map._free_entry(to_delete);
+		}
+		map._free_entry(entry);
+		*entry_ref = null;
+	}
+	map.count = 0;
+}
+
+fn void Map.free(self)
+{
+	if (!self) return;
+	MapImpl* map = (MapImpl*)self;
+	self.clear();
+	map._free_internal(map.table.ptr);
+	map.table = {};
+	allocator::free(map.allocator, map);
+}
+
+fn Key[] Map.temp_keys_list(map)
+{
+	return map.new_keys_list(allocator::temp()) @inline;
+}
+
+fn Key[] Map.new_keys_list(self, Allocator allocator = allocator::heap())
+{
+	MapImpl* map = (MapImpl*)self;
+	if (!map || !map.count) return {};
+
+	Key[] list = allocator::alloc_array(allocator, Key, map.count);
+	usz index = 0;
+	foreach (Entry* entry : map.table)
+	{
+		while (entry)
+		{
+			list[index++] = entry.key;
+			entry = entry.next;
+		}
+	}
+	return list;
+}
+
+macro Map.@each(map; @body(key, value))
+{
+	map.@each_entry(; Entry* entry) {
+		@body(entry.key, entry.value);
+	};
+}
+
+macro Map.@each_entry(self; @body(entry))
+{
+	MapImpl *map = (MapImpl*)self;
+	if (!map || !map.count) return;
+	foreach (Entry* entry : map.table)
+	{
+		while (entry)
+		{
+			@body(entry);
+			entry = entry.next;
+		}
+	}
+}
+
+fn Value[] Map.temp_values_list(map)
+{
+	return map.new_values_list(allocator::temp()) @inline;
+}
+
+fn Value[] Map.new_values_list(self, Allocator allocator = allocator::heap())
+{
+	MapImpl* map = (MapImpl*)self;
+	if (!map || !map.count) return {};
+	Value[] list = allocator::alloc_array(allocator, Value, map.count);
+	usz index = 0;
+	foreach (Entry* entry : map.table)
+	{
+		while (entry)
+		{
+			list[index++] = entry.value;
+			entry = entry.next;
+		}
+	}
+	return list;
+}
+
+fn bool Map.has_value(self, Value v) @if(VALUE_IS_EQUATABLE)
+{
+	MapImpl* map = (MapImpl*)self;
+	if (!map || !map.count) return false;
+	foreach (Entry* entry : map.table)
+	{
+		while (entry)
+		{
+			if (equals(v, entry.value)) return true;
+			entry = entry.next;
+		}
+	}
+	return false;
+}
+
+// --- private methods
+
+fn void MapImpl._add_entry(&map, uint hash, Key key, Value value, uint bucket_index) @private
+{
+	$if COPY_KEYS:
+	key = key.copy(map.allocator);
+	$endif
+	Entry* entry = allocator::new(map.allocator, Entry, { .hash = hash, .key = key, .value = value, .next = map.table[bucket_index] });
+	map.table[bucket_index] = entry;
+	if (map.count++ >= map.threshold)
+	{
+		map._resize(map.table.len * 2);
+	}
+}
+
+fn void MapImpl._resize(&map, uint new_capacity) @private
+{
+	Entry*[] old_table = map.table;
+	uint old_capacity = old_table.len;
+	if (old_capacity == MAXIMUM_CAPACITY)
+	{
+		map.threshold = uint.max;
+		return;
+	}
+	Entry*[] new_table = allocator::new_array(map.allocator, Entry*, new_capacity);
+	map._transfer(new_table);
+	map.table = new_table;
+	map._free_internal(old_table.ptr);
+	map.threshold = (uint)(new_capacity * map.load_factor);
+}
+
+fn uint rehash(uint hash) @inline @private
+{
+	hash ^= (hash >> 20) ^ (hash >> 12);
+	return hash ^ ((hash >> 7) ^ (hash >> 4));
+}
+
+macro uint index_for(uint hash, uint capacity) @private
+{
+	return hash & (capacity - 1);
+}
+
+fn void MapImpl._transfer(&map, Entry*[] new_table) @private
+{
+	Entry*[] src = map.table;
+	uint new_capacity = new_table.len;
+	foreach (uint j, Entry *e : src)
+	{
+		if (!e) continue;
+		do
+		{
+			Entry* next = e.next;
+			uint i = index_for(e.hash, new_capacity);
+			e.next = new_table[i];
+			new_table[i] = e;
+			e = next;
+		}
+		while (e);
+	}
+}
+
+fn void _init(MapImpl* impl, uint capacity, float load_factor, Allocator allocator) @private
+{
+	capacity = math::next_power_of_2(capacity);
+	*impl = {
+		.allocator = allocator,
+		.load_factor = load_factor,
+		.threshold = (uint)(capacity * load_factor),
+		.table = allocator::new_array(allocator, Entry*, capacity)
+	};
+}
+
+fn void MapImpl._put_for_create(&map, Key key, Value value) @private
+{
+	uint hash = rehash(key.hash());
+	uint i = index_for(hash, map.table.len);
+	for (Entry *e = map.table[i]; e != null; e = e.next)
+	{
+		if (e.hash == hash && equals(key, e.key))
+		{
+			e.value = value;
+			return;
+		}
+	}
+	map._create_entry(hash, key, value, i);
+}
+
+fn void MapImpl._free_internal(&map, void* ptr) @inline @private
+{
+	allocator::free(map.allocator, ptr);
+}
+
+fn bool MapImpl._remove_entry_for_key(&map, Key key) @private
+{
+	if (!map.count) return false;
+	uint hash = rehash(key.hash());
+	uint i = index_for(hash, map.table.len);
+	Entry* prev = map.table[i];
+	Entry* e = prev;
+	while (e)
+	{
+		Entry *next = e.next;
+		if (e.hash == hash && equals(key, e.key))
+		{
+			map.count--;
+			if (prev == e)
+			{
+				map.table[i] = next;
+			}
+			else
+			{
+				prev.next = next;
+			}
+			map._free_entry(e);
+			return true;
+		}
+		prev = e;
+		e = next;
+	}
+	return false;
+}
+
+fn void MapImpl._create_entry(&map, uint hash, Key key, Value value, int bucket_index) @private
+{
+	Entry *e = map.table[bucket_index];
+	$if COPY_KEYS:
+	key = key.copy(map.allocator);
+	$endif
+	Entry* entry = allocator::new(map.allocator, Entry, { .hash = hash, .key = key, .value = value, .next = map.table[bucket_index] });
+	map.table[bucket_index] = entry;
+	map.count++;
+}
+
+fn void MapImpl._free_entry(&self, Entry *entry) @local
+{
+	$if COPY_KEYS:
+	allocator::free(self.allocator, entry.key);
+	$endif
+	self._free_internal(entry);
+}
+
+struct Entry
+{
+	uint hash;
+	Key key;
+	Value value;
+	Entry* next;
+}

lib/std/collections/maybe.c3

@@ -0,0 +1,46 @@
+module std::collections::maybe(<Type>);
+import std::io;
+
+struct Maybe (Printable)
+{
+	Type value;
+	bool has_value;
+}
+
+fn usz! Maybe.to_format(&self, Formatter* f) @dynamic
+{
+	if (self.has_value) return f.printf("[%s]", self.value);
+	return f.printf("[EMPTY]");
+}
+
+fn void Maybe.set(&self, Type val)
+{
+	*self = { .value = val, .has_value = true };
+}
+
+fn void Maybe.reset(&self)
+{
+	*self = {};
+}
+
+fn Maybe value(Type val)
+{
+	return { .value = val, .has_value = true };
+}
+
+fn Maybe Maybe.with_value(Type val) @deprecated("Use maybe::value instead.") @operator(construct)
+{
+	return { .value = val, .has_value = true };
+}
+
+fn Maybe Maybe.empty() @deprecated("Use maybe::EMPTY instead.") @operator(construct)
+{
+	return { };
+}
+
+const Maybe EMPTY = { };
+
+macro Type! Maybe.get(self)
+{
+	return self.has_value ? self.value : SearchResult.MISSING?;
+}

lib/std/collections/object.c3

@@ -0,0 +1,468 @@
+// Copyright (c) 2023 Christoffer Lerno. All rights reserved.
+// Use of this source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+module std::collections::object;
+import std::collections::map, std::collections::list, std::io;
+
+const Object TRUE_OBJECT = { .b = true, .type = bool.typeid };
+const Object FALSE_OBJECT = { .b = false, .type = bool.typeid };
+const Object NULL_OBJECT = { .type = void*.typeid };
+
+struct Object (Printable)
+{
+	typeid type;
+	Allocator allocator;
+	union
+	{
+		uint128 i;
+		double f;
+		bool b;
+		String s;
+		void* other;
+		ObjectInternalList array;
+		ObjectInternalMap map;
+	}
+}
+
+
+fn usz! Object.to_format(&self, Formatter* formatter) @dynamic
+{
+	switch (self.type)
+	{
+		case void:
+			return formatter.printf("{}")!;
+		case void*:
+			return formatter.printf("null")!;
+		case String:
+			return formatter.printf(`"%s"`, self.s)!;
+		case bool:
+			return formatter.printf(self.b ? "true" : "false")!;
+		case ObjectInternalList:
+			usz n = formatter.printf("[")!;
+			foreach (i, ol : self.array)
+			{
+				if (i > 0) n += formatter.printf(",")!;
+				n += ol.to_format(formatter)!;
+			}
+			n += formatter.printf("]")!;
+			return n;
+		case ObjectInternalMap:
+			usz n = formatter.printf("{")!;
+			@stack_mem(1024; Allocator mem)
+			{
+				foreach (i, key : self.map.copy_keys(mem))
+				{
+					if (i > 0) n += formatter.printf(",")!;
+					n += formatter.printf(`"%s":`, key)!;
+					n += self.map.get(key).to_format(formatter)!;
+				}
+			};
+			n += formatter.printf("}")!;
+			return n;
+		default:
+			switch (self.type.kindof)
+			{
+				case SIGNED_INT:
+					return formatter.printf("%d", (int128)self.i)!;
+				case UNSIGNED_INT:
+					return formatter.printf("%d", (uint128)self.i)!;
+				case FLOAT:
+					return formatter.printf("%g", self.f)!;
+				case ENUM:
+					return formatter.printf("%d", self.i)!;
+				default:
+					return formatter.printf("<>")!;
+			}
+	}
+}
+
+fn Object* new_obj(Allocator allocator)
+{
+	return allocator::new(allocator, Object, { .allocator = allocator, .type = void.typeid });
+}
+
+fn Object* new_null()
+{
+	return &NULL_OBJECT;
+}
+
+fn Object* new_int(int128 i, Allocator allocator)
+{
+	return allocator::new(allocator, Object, { .i = i, .allocator = allocator, .type = int128.typeid });
+}
+
+macro Object* new_enum(e, Allocator allocator)
+{
+	return allocator::new(allocator, Object, { .i = (int128)e, .allocator = allocator, .type = @typeid(e) });
+}
+
+fn Object* new_float(double f, Allocator allocator)
+{
+	return allocator::new(allocator, Object, { .f = f, .allocator = allocator, .type = double.typeid });
+}
+
+fn Object* new_string(String s, Allocator allocator)
+{
+	return allocator::new(allocator, Object, { .s = s.copy(allocator), .allocator = allocator, .type = String.typeid });
+}
+
+
+fn Object* new_bool(bool b)
+{
+	return b ? &TRUE_OBJECT : &FALSE_OBJECT;
+}
+
+fn void Object.free(&self)
+{
+	switch (self.type)
+	{
+		case void:
+			break;
+		case String:
+			allocator::free(self.allocator, self.s);
+		case ObjectInternalList:
+			foreach (ol : self.array)
+			{
+				ol.free();
+			}
+			self.array.free();
+		case ObjectInternalMap:
+			self.map.@each_entry(; ObjectInternalMapEntry* entry) {
+				entry.value.free();
+			};
+			self.map.free();
+		default:
+			break;
+	}
+	if (self.allocator) allocator::free(self.allocator, self);
+}
+
+fn bool Object.is_null(&self) @inline => self == &NULL_OBJECT;
+fn bool Object.is_empty(&self) @inline => self.type == void.typeid;
+fn bool Object.is_map(&self) @inline => self.type == ObjectInternalMap.typeid;
+fn bool Object.is_array(&self) @inline => self.type == ObjectInternalList.typeid;
+fn bool Object.is_bool(&self) @inline => self.type == bool.typeid;
+fn bool Object.is_string(&self) @inline => self.type == String.typeid;
+fn bool Object.is_float(&self) @inline => self.type == double.typeid;
+fn bool Object.is_int(&self) @inline => self.type == int128.typeid;
+fn bool Object.is_keyable(&self) => self.is_empty() || self.is_map();
+fn bool Object.is_indexable(&self) => self.is_empty() || self.is_array();
+
+<*
+ @require self.is_keyable()
+*>
+fn void Object.init_map_if_needed(&self) @private
+{
+	if (self.is_empty())
+	{
+		self.type = ObjectInternalMap.typeid;
+		self.map.init(self.allocator);
+	}
+}
+
+<*
+ @require self.is_indexable()
+*>
+fn void Object.init_array_if_needed(&self) @private
+{
+	if (self.is_empty())
+	{
+		self.type = ObjectInternalList.typeid;
+		self.array.init(self.allocator);
+	}
+}
+
+<*
+ @require self.is_keyable()
+*>
+fn void Object.set_object(&self, String key, Object* new_object) @private
+{
+	self.init_map_if_needed();
+	ObjectInternalMapEntry*! entry = self.map.get_entry(key);
+	defer
+	{
+		(void)entry.value.free();
+	}
+	self.map.set(key, new_object);
+}
+
+macro Object* Object.object_from_value(&self, value) @private
+{
+	var $Type = $typeof(value);
+	$switch
+		$case types::is_int($Type):
+			return new_int(value, self.allocator);
+		$case types::is_float($Type):
+			return new_float(value, self.allocator);
+		$case $Type.typeid == String.typeid:
+			return new_string(value, self.allocator);
+		$case $Type.typeid == bool.typeid:
+			return new_bool(value);
+		$case $Type.typeid == Object*.typeid:
+			return value;
+		$case $Type.typeid == void*.typeid:
+			if (value != null) return CastResult.TYPE_MISMATCH?;
+			return &NULL_OBJECT;
+		$case $assignable(value, String):
+			return new_string(value, self.allocator);
+		$default:
+			$error "Unsupported object type.";
+	$endswitch
+
+}
+
+macro Object* Object.set(&self, String key, value)
+{
+	Object* val = self.object_from_value(value);
+	self.set_object(key, val);
+	return val;
+}
+
+<*
+ @require self.is_indexable()
+*>
+macro Object* Object.set_at(&self, usz index, String key, value)
+{
+	Object* val = self.object_from_value(value);
+	self.set_object_at(key, index, val);
+	return val;
+}
+
+<*
+ @require self.is_indexable()
+ @ensure return != null
+*>
+macro Object* Object.push(&self, value)
+{
+	Object* val = self.object_from_value(value);
+	self.push_object(val);
+	return val;
+}
+
+<*
+ @require self.is_keyable()
+*>
+fn Object*! Object.get(&self, String key) => self.is_empty() ? SearchResult.MISSING? : self.map.get(key);
+
+
+fn bool Object.has_key(&self, String key) => self.is_map() && self.map.has_key(key);
+
+<*
+ @require self.is_indexable()
+*>
+fn Object* Object.get_at(&self, usz index)
+{
+	return self.array.get(index);
+}
+
+<*
+ @require self.is_indexable()
+*>
+fn usz Object.get_len(&self)
+{
+	return self.array.len();
+}
+
+<*
+ @require self.is_indexable()
+*>
+fn void Object.push_object(&self, Object* to_append)
+{
+	self.init_array_if_needed();
+	self.array.push(to_append);
+}
+
+<*
+ @require self.is_indexable()
+*>
+fn void Object.set_object_at(&self, usz index, Object* to_set)
+{
+	self.init_array_if_needed();
+	while (self.array.len() < index)
+	{
+		self.array.push(&NULL_OBJECT);
+	}
+	if (self.array.len() == index)
+	{
+		self.array.push(to_set);
+		return;
+	}
+	self.array.get(index).free();
+	self.array.set_at(index, to_set);
+}
+
+<*
+ @require $Type.kindof.is_int() "Expected an integer type."
+*>
+macro get_integer_value(Object* value, $Type)
+{
+	if (value.is_float())
+	{
+		return ($Type)value.f;
+	}
+	if (value.is_string())
+	{
+		$if $Type.kindof == TypeKind.SIGNED_INT:
+			return ($Type)value.s.to_int128();
+		$else
+			return ($Type)value.s.to_uint128();
+		$endif
+	}
+	if (!value.is_int()) return NumberConversion.MALFORMED_INTEGER?;
+	return ($Type)value.i;
+}
+
+
+<*
+ @require self.is_indexable()
+ @require $Type.kindof.is_int() : "Expected an integer type"
+*>
+macro Object.get_integer_at(&self, $Type, usz index) @private
+{
+	return get_integer_value(self.get_at(index), $Type);
+}
+
+<*
+ @require self.is_keyable()
+ @require $Type.kindof.is_int() : "Expected an integer type"
+*>
+macro Object.get_integer(&self, $Type, String key) @private
+{
+	return get_integer_value(self.get(key), $Type);
+}
+
+fn ichar! Object.get_ichar(&self, String key) => self.get_integer(ichar, key);
+fn short! Object.get_short(&self, String key) => self.get_integer(short, key);
+fn int! Object.get_int(&self, String key) => self.get_integer(int, key);
+fn long! Object.get_long(&self, String key) => self.get_integer(long, key);
+fn int128! Object.get_int128(&self, String key) => self.get_integer(int128, key);
+
+fn ichar! Object.get_ichar_at(&self, usz index) => self.get_integer_at(ichar, index);
+fn short! Object.get_short_at(&self, usz index) => self.get_integer_at(short, index);
+fn int! Object.get_int_at(&self, usz index) => self.get_integer_at(int, index);
+fn long! Object.get_long_at(&self, usz index) => self.get_integer_at(long, index);
+fn int128! Object.get_int128_at(&self, usz index) => self.get_integer_at(int128, index);
+
+fn char! Object.get_char(&self, String key) => self.get_integer(ichar, key);
+fn short! Object.get_ushort(&self, String key) => self.get_integer(ushort, key);
+fn uint! Object.get_uint(&self, String key) => self.get_integer(uint, key);
+fn ulong! Object.get_ulong(&self, String key) => self.get_integer(ulong, key);
+fn uint128! Object.get_uint128(&self, String key) => self.get_integer(uint128, key);
+
+fn char! Object.get_char_at(&self, usz index) => self.get_integer_at(char, index);
+fn ushort! Object.get_ushort_at(&self, usz index) => self.get_integer_at(ushort, index);
+fn uint! Object.get_uint_at(&self, usz index) => self.get_integer_at(uint, index);
+fn ulong! Object.get_ulong_at(&self, usz index) => self.get_integer_at(ulong, index);
+fn uint128! Object.get_uint128_at(&self, usz index) => self.get_integer_at(uint128, index);
+
+<*
+ @require self.is_keyable()
+*>
+fn String! Object.get_string(&self, String key)
+{
+	Object* value = self.get(key)!;
+	if (!value.is_string()) return CastResult.TYPE_MISMATCH?;
+	return value.s;
+}
+
+<*
+ @require self.is_indexable()
+*>
+fn String! Object.get_string_at(&self, usz index)
+{
+	Object* value = self.get_at(index);
+	if (!value.is_string()) return CastResult.TYPE_MISMATCH?;
+	return value.s;
+}
+
+<*
+ @require self.is_keyable()
+*>
+macro String! Object.get_enum(&self, $EnumType, String key)
+{
+	Object value = self.get(key)!;
+	if ($EnumType.typeid != value.type) return CastResult.TYPE_MISMATCH?;
+	return ($EnumType)value.i;
+}
+
+<*
+ @require self.is_indexable()
+*>
+macro String! Object.get_enum_at(&self, $EnumType, usz index)
+{
+	Object value = self.get_at(index);
+	if ($EnumType.typeid != value.type) return CastResult.TYPE_MISMATCH?;
+	return ($EnumType)value.i;
+}
+
+<*
+ @require self.is_keyable()
+*>
+fn bool! Object.get_bool(&self, String key)
+{
+	Object* value = self.get(key)!;
+	if (!value.is_bool()) return CastResult.TYPE_MISMATCH?;
+	return value.b;
+}
+
+
+<*
+ @require self.is_indexable()
+*>
+fn bool! Object.get_bool_at(&self, usz index)
+{
+	Object* value = self.get_at(index);
+	if (!value.is_bool()) return CastResult.TYPE_MISMATCH?;
+	return value.b;
+}
+
+<*
+ @require self.is_keyable()
+*>
+fn double! Object.get_float(&self, String key)
+{
+	Object* value = self.get(key)!;
+	switch (value.type.kindof)
+	{
+		case SIGNED_INT:
+			return (double)value.i;
+		case UNSIGNED_INT:
+			return (double)(uint128)value.i;
+		case FLOAT:
+			return value.f;
+		default:
+			return CastResult.TYPE_MISMATCH?;
+	}
+}
+
+<*
+ @require self.is_indexable()
+*>
+fn double! Object.get_float_at(&self, usz index)
+{
+	Object* value = self.get_at(index);
+	switch (value.type.kindof)
+	{
+		case SIGNED_INT:
+			return (double)value.i;
+		case UNSIGNED_INT:
+			return (double)(uint128)value.i;
+		case FLOAT:
+			return value.f;
+		default:
+			return CastResult.TYPE_MISMATCH?;
+	}
+}
+
+fn Object* Object.get_or_create_obj(&self, String key)
+{
+	if (try obj = self.get(key) && !obj.is_null()) return obj;
+	Object* container = new_obj(self.allocator);
+	self.set(key, container);
+	return container;
+}
+
+def ObjectInternalMap = HashMap(<String, Object*>) @private;
+def ObjectInternalList = List(<Object*>) @private;
+def ObjectInternalMapEntry = Entry(<String, Object*>) @private;
+

lib/std/collections/priorityqueue.c3

@@ -0,0 +1,163 @@
+//  priorityqueue.c3
+//  A priority queue using a classic binary heap for C3.
+//
+//  Copyright (c) 2022 David Kopec
+//
+//  Permission is hereby granted, free of charge, to any person obtaining a copy
+//  of this software and associated documentation files (the "Software"), to deal
+//  in the Software without restriction, including without limitation the rights
+//  to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+//  copies of the Software, and to permit persons to whom the Software is
+//  furnished to do so, subject to the following conditions:
+//
+//  The above copyright notice and this permission notice shall be included in all
+//  copies or substantial portions of the Software.
+//
+//  THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+//  IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+//  FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+//  AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+//  LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+//  OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+//  SOFTWARE.
+module std::collections::priorityqueue(<Type>);
+import std::collections::priorityqueue::private;
+
+distinct PriorityQueue = inline PrivatePriorityQueue(<Type, false>);
+distinct PriorityQueueMax = inline PrivatePriorityQueue(<Type, true>);
+
+module std::collections::priorityqueue::private(<Type, MAX>);
+import std::collections::list, std::io;
+
+def Heap = List(<Type>);
+
+struct PrivatePriorityQueue (Printable)
+{
+	Heap heap;
+}
+
+fn void PrivatePriorityQueue.init(&self, Allocator allocator, usz initial_capacity = 16, ) @inline
+{
+	self.heap.new_init(initial_capacity, allocator);
+}
+
+fn void PrivatePriorityQueue.new_init(&self, usz initial_capacity = 16, Allocator allocator = allocator::heap()) @inline
+{
+	self.heap.new_init(initial_capacity, allocator);
+}
+
+fn void PrivatePriorityQueue.temp_init(&self, usz initial_capacity = 16) @inline
+{
+	self.heap.new_init(initial_capacity, allocator::temp()) @inline;
+}
+
+
+fn void PrivatePriorityQueue.push(&self, Type element)
+{
+	self.heap.push(element);
+	usz i = self.heap.len() - 1;
+	while (i > 0)
+	{
+		usz parent = (i - 1) / 2;
+		Type item = self.heap[i];
+		Type parent_item = self.heap[parent];
+		$if MAX:
+			bool ok = greater(item, parent_item);
+		$else
+			bool ok = less(item, parent_item);
+		$endif
+		if (!ok) break;
+		self.heap.swap(i, parent);
+		i = parent;
+	}
+}
+
+fn void PrivatePriorityQueue.remove_at(&self, usz index)
+{
+	if (index == 0)
+	{
+		self.pop()!!;
+		return;
+	}
+	self.heap.remove_at(index);
+}
+<*
+ @require self != null
+*>
+fn Type! PrivatePriorityQueue.pop(&self)
+{
+	usz i = 0;
+	usz len = self.heap.len();
+	if (!len) return IteratorResult.NO_MORE_ELEMENT?;
+	usz new_count = len - 1;
+	self.heap.swap(0, new_count);
+	while OUTER: ((2 * i + 1) < new_count)
+	{
+		usz j = 2 * i + 1;
+		Type left = self.heap[j];
+		Type item = self.heap[i];
+		switch
+		{
+			case j + 1 < new_count:
+				Type right = self.heap[j + 1];
+				$if MAX:
+				if (!greater(right, left)) nextcase;
+				if (!greater(right, item)) break OUTER;
+				$else
+				if (!greater(left, right)) nextcase;
+				if (!greater(item, right)) break OUTER;
+				$endif
+				j++;
+			default:
+				$if MAX:
+				if (!greater(left, item)) break OUTER;
+				$else
+				if (!greater(item, left)) break OUTER;
+				$endif
+		}
+		self.heap.swap(i, j);
+		i = j;
+	}
+
+	return self.heap.pop();
+}
+
+fn Type! PrivatePriorityQueue.first(&self)
+{
+	if (!self.len()) return IteratorResult.NO_MORE_ELEMENT?;
+	return self.heap.get(0);
+}
+
+fn void PrivatePriorityQueue.free(&self)
+{
+	self.heap.free();
+}
+
+fn usz PrivatePriorityQueue.len(&self) @operator(len)
+{
+	return self.heap.len();
+}
+
+fn bool PrivatePriorityQueue.is_empty(&self)
+{
+	return self.heap.is_empty();
+}
+
+<*
+ @require index < self.len()
+*>
+fn Type PrivatePriorityQueue.get(&self, usz index) @operator([])
+{
+	return self.heap[index];
+}
+
+fn usz! PrivatePriorityQueue.to_format(&self, Formatter* formatter) @dynamic
+{
+	return self.heap.to_format(formatter);
+}
+
+fn String PrivatePriorityQueue.to_new_string(&self, Allocator allocator = allocator::heap()) @dynamic
+{
+	return self.heap.to_new_string(allocator);
+}
+

lib/std/collections/range.c3

@@ -0,0 +1,95 @@
+<*
+ @require Type.is_ordered : "The type must be ordered"
+*>
+module std::collections::range(<Type>);
+import std::io;
+
+struct Range (Printable)
+{
+	Type start;
+	Type end;
+}
+
+fn usz Range.len(&self) @operator(len)
+{
+	if (self.end < self.start) return 0;
+	return (usz)(self.end - self.start) + 1;
+}
+
+fn bool Range.contains(&self, Type value) @inline
+{
+	return value >= self.start && value <= self.end;
+}
+
+<*
+ @require index < self.len() : "Can't index into an empty range"
+*>
+fn Type Range.get(&self, usz index) @operator([])
+{
+	return (Type)(self.start + (usz)index);
+}
+
+fn String Range.to_new_string(&self, Allocator allocator = allocator::heap()) @dynamic @deprecated
+{
+	return string::format("[%s..%s]", self.start, self.end, allocator: allocator);
+}
+
+fn String Range.to_string(&self, Allocator allocator) @dynamic
+{
+	return string::format("[%s..%s]", self.start, self.end, allocator: allocator);
+}
+
+fn String Range.to_tstring(&self)
+{
+	return self.to_string(allocator::temp());
+}
+
+fn usz! Range.to_format(&self, Formatter* formatter) @dynamic
+{
+	return formatter.printf("[%s..%s]", self.start, self.end)!;
+}
+
+struct ExclusiveRange (Printable)
+{
+	Type start;
+	Type end;
+}
+
+fn usz ExclusiveRange.len(&self) @operator(len)
+{
+	if (self.end < self.start) return 0;
+	return (usz)(self.end - self.start);
+}
+
+fn bool ExclusiveRange.contains(&self, Type value) @inline
+{
+	return value >= self.start && value < self.end;
+}
+
+fn usz! ExclusiveRange.to_format(&self, Formatter* formatter) @dynamic
+{
+	return formatter.printf("[%s..<%s]", self.start, self.end)!;
+}
+
+fn String ExclusiveRange.to_new_string(&self, Allocator allocator = null) @dynamic
+{
+	return self.to_string(allocator ?: allocator::heap());
+}
+
+fn String ExclusiveRange.to_string(&self, Allocator allocator) @dynamic
+{
+	return string::format("[%s..<%s]", self.start, self.end, allocator: allocator);
+}
+
+fn String ExclusiveRange.to_tstring(&self)
+{
+	return self.to_new_string(allocator::temp());
+}
+
+<*
+ @require index < self.len() : "Can't index into an empty range"
+*>
+fn Type ExclusiveRange.get(&self, usz index) @operator([])
+{
+	return (Type)(self.start + index);
+}

lib/std/collections/ringbuffer.c3

@@ -0,0 +1,106 @@
+<*
+ @require values::@is_int(SIZE) &&& SIZE > 0 "The size must be positive integer"
+*>
+module std::collections::ringbuffer(<Type, SIZE>);
+
+struct RingBuffer
+{
+	Type[SIZE] buf;
+	usz written;
+	usz head;
+}
+
+fn void RingBuffer.init(&self) @inline
+{
+	*self = {};
+}
+
+fn void RingBuffer.push(&self, Type c)
+{
+	if (self.written < SIZE)
+	{
+		self.buf[self.written] = c;
+		self.written++;
+	}
+	else
+	{
+		self.buf[self.head] = c;
+		self.head = (self.head + 1) % SIZE;
+	}
+}
+
+fn Type RingBuffer.get(&self, usz index) @operator([])
+{
+	index %= SIZE;
+	usz avail = SIZE - self.head;
+	if (index < avail)
+	{
+		return self.buf[self.head + index];
+	}
+	return self.buf[index - avail];
+}
+
+fn Type! RingBuffer.pop(&self)
+{
+	switch
+	{
+		case self.written == 0:
+			return SearchResult.MISSING?;
+		case self.written < SIZE:
+			self.written--;
+			return self.buf[self.written];
+		default:
+			self.head = (self.head - 1) % SIZE;
+			return self.buf[self.head];
+	}
+}
+
+fn usz RingBuffer.read(&self, usz index, Type[] buffer)
+{
+	index %= SIZE;
+	if (self.written < SIZE)
+	{
+		if (index >= self.written) return 0;
+		usz end = self.written - index;
+		usz n = min(end, buffer.len);
+		buffer[:n] = self.buf[index:n];
+		return n;
+	}
+	usz end = SIZE - self.head;
+	if (index >= end)
+	{
+		index -= end;
+		if (index >= self.head) return 0;
+		usz n = min(self.head - index, buffer.len);
+		buffer[:n] = self.buf[index:n];
+		return n;
+	}
+	if (buffer.len <= SIZE - index)
+	{
+		usz n = buffer.len;
+		buffer[:n] = self.buf[self.head + index:n];
+		return n;
+	}
+	usz n1 = SIZE - index;
+	buffer[:n1] = self.buf[self.head + index:n1];
+	buffer = buffer[n1..];
+	index -= n1;
+	usz n2 = min(self.head - index, buffer.len);
+	buffer[:n2] = self.buf[index:n2];
+	return n1 + n2;
+}
+
+fn void RingBuffer.write(&self, Type[] buffer)
+{
+	usz i;
+	while (self.written < SIZE && i < buffer.len)
+	{
+		self.buf[self.written] = buffer[i++];
+		self.written++;
+	}
+	foreach (c : buffer[i..])
+	{
+		self.buf[self.head] = c;
+		self.head = (self.head + 1) % SIZE;
+	}
+}

lib/std/collections/tuple.c3

@@ -0,0 +1,16 @@
+module std::collections::tuple(<Type1, Type2>);
+
+struct Tuple
+{
+	Type1 first;
+	Type2 second;
+}
+
+module std::collections::triple(<Type1, Type2, Type3>);
+
+struct Triple
+{
+	Type1 first;
+	Type2 second;
+	Type3 third;
+}

lib/std/compression/qoi.c3

@@ -0,0 +1,495 @@
+module std::compression::qoi;
+
+const uint PIXELS_MAX = 400000000;
+
+<*
+ Colorspace.
+ Purely informative. It will be saved to the file header,
+ but does not affect how chunks are en-/decoded.
+*>
+enum QOIColorspace : char (char id)
+{
+	SRGB = 0, 	// sRGB with linear alpha
+	LINEAR = 1 	// all channels linear
+}
+
+<*
+ Channels.
+ The channels used in an image.
+ AUTO can be used when decoding to automatically determine
+ the channels from the file's header.
+*>
+enum QOIChannels : char (char id)
+{
+	AUTO = 0,
+	RGB = 3,
+	RGBA = 4
+}
+
+<*
+ Descriptor.
+ Contains information about an image.
+*>
+struct QOIDesc
+{
+	uint width;
+	uint height;
+	QOIChannels channels;
+	QOIColorspace colorspace;
+}
+
+<*
+ QOI Errors.
+ These are all the possible bad outcomes.
+*>
+fault QOIError
+{
+	INVALID_PARAMETERS,
+	FILE_OPEN_FAILED,
+	FILE_WRITE_FAILED,
+	INVALID_DATA,
+	TOO_MANY_PIXELS
+}
+
+
+// Let the user decide if they want to use std::io
+module std::compression::qoi @if(!$feature(QOI_NO_STDIO));
+import std::io;
+
+<*
+ Encode raw RGB or RGBA pixels into a QOI image and write it to the
+ file system.
+
+ The desc struct must be filled with the image width, height, the
+ used channels (QOIChannels.RGB or RGBA) and the colorspace
+ (QOIColorspace.SRGB or LINEAR).
+
+ The function returns an optional, which can either be a QOIError
+ or the number of bytes written on success.
+
+ @param [in] filename `The file's name to write the image to`
+ @param [in] input `The raw RGB or RGBA pixels to encode`
+ @param [&in] desc `The descriptor of the image`
+*>
+fn usz! write(String filename, char[] input, QOIDesc* desc) => @pool()
+{
+	// encode data
+	char[] output = new_encode(input, desc, allocator: allocator::temp())!;
+
+	// open file
+	File! f = file::open(filename, "wb");
+	if (catch f) return QOIError.FILE_OPEN_FAILED?;
+
+	// write data to file and close it
+	usz! written = f.write(output);
+	if (catch written) return QOIError.FILE_WRITE_FAILED?;
+	if (catch f.close()) return QOIError.FILE_WRITE_FAILED?;
+
+	return written;
+}
+
+
+<*
+ Read and decode a QOI image from the file system.
+
+ If channels is set to QOIChannels.AUTO, the function will
+ automatically determine the channels from the file's header.
+ However, if channels is RGB or RGBA, the output format will be
+ forced into this number of channels.
+
+ The desc struct will be filled with the width, height,
+ channels and colorspace of the image.
+
+ The function returns an optional, which can either be a QOIError
+ or a char[] pointing to the decoded pixels on success.
+
+ The returned pixel data should be free()d after use, or the decoding
+ and use of the data should be wrapped in a @pool() { ... }; block.
+
+ @param [in] filename `The file's name to read the image from`
+ @param [&out] desc `The descriptor to fill with the image's info`
+ @param channels `The channels to be used`
+*>
+fn char[]! new_read(String filename, QOIDesc* desc, QOIChannels channels = AUTO, Allocator allocator = allocator::heap()) => @pool(allocator)
+{
+	// read file
+	char[] data = file::load_temp(filename) ?? QOIError.FILE_OPEN_FAILED?!;
+	// pass data to decode function
+	return new_decode(data, desc, channels, allocator);
+}
+
+fn char[]! read(String filename, QOIDesc* desc, QOIChannels channels = AUTO, Allocator allocator = allocator::heap()) @deprecated("Use new_read")
+{
+	return new_read(filename, desc, channels, allocator);
+}
+
+
+// Back to basic non-stdio mode
+module std::compression::qoi;
+import std::bits;
+
+fn char[]! encode(char[] input, QOIDesc* desc, Allocator allocator = allocator::heap()) @deprecated("use encode_new")
+{
+	return new_encode(input, desc, allocator);
+}
+
+<*
+ Encode raw RGB or RGBA pixels into a QOI image in memory.
+
+ The function returns an optional, which can either be a QOIError
+ or a char[] pointing to the encoded data on success.
+
+ The returned qoi data should be free()d after use, or the encoding
+ and use of the data should be wrapped in a @pool() { ... }; block.
+ See the write() function for an example.
+
+ @param [in] input `The raw RGB or RGBA pixels to encode`
+ @param [&in] desc `The descriptor of the image`
+*>
+fn char[]! new_encode(char[] input, QOIDesc* desc, Allocator allocator = allocator::heap()) @nodiscard
+{
+	// check info in desc
+	if (desc.width == 0 || desc.height == 0) return QOIError.INVALID_PARAMETERS?;
+	if (desc.channels == AUTO) return QOIError.INVALID_PARAMETERS?;
+	uint pixels = desc.width * desc.height;
+	if (pixels > PIXELS_MAX) return QOIError.TOO_MANY_PIXELS?;
+	
+	// check input data size
+	uint image_size = pixels * desc.channels.id;
+	if (image_size != input.len) return QOIError.INVALID_DATA?;
+
+	// allocate memory for encoded data (output)
+	// header + chunk tag and RGB(A) data for each pixel + end of stream
+	uint max_size = Header.sizeof + pixels + image_size + END_OF_STREAM.len;
+	char[] output = allocator::alloc_array(allocator, char, max_size); // no need to init
+	defer catch allocator::free(allocator, output);
+
+	// write header
+	*(Header*)output.ptr = {
+		.be_magic = bswap('qoif'),
+		.be_width = bswap(desc.width),
+		.be_height = bswap(desc.height),
+		.channels = desc.channels.id,
+		.colorspace = desc.colorspace.id
+	};
+
+	uint pos = Header.sizeof; 	// Current position in output
+	uint loc;					// Current position in image (top-left corner)
+	uint loc_end = image_size - desc.channels.id; // End of image data
+	char run_length = 0; 		// Length of the current run
+	
+	Pixel[64] palette; // Zero-initialized by default
+	Pixel prev = { 0, 0, 0, 255 };
+	Pixel p = { 0, 0, 0, 255 };
+
+	ichar[<3>] diff; // pre-allocate for diff
+	ichar[<3>] luma; // ...and luma
+
+	// write chunks
+	for (loc = 0; loc < image_size; loc += desc.channels.id)
+	{
+		// set previous pixel
+		prev = p;
+
+		// get current pixel
+		p[:3] = input[loc:3]; // cutesy slices :3
+		if (desc.channels == RGBA) p.a = input[loc + 3];
+
+		// check if we can run the previous pixel
+		if (prev == p)
+		{
+			run_length++;
+			if (run_length == 62 || loc == loc_end)
+			{
+				*@extract(OpRun, output, &pos) = { OP_RUN, run_length - 1 };
+				run_length = 0;
+			}
+			continue;
+		}
+		// end last run if there was one
+		if (run_length > 0)
+		{
+			*@extract(OpRun, output, &pos) = { OP_RUN, run_length - 1 };
+			run_length = 0;
+		}
+
+		switch
+		{
+			// check if we can index the palette
+			case (palette[p.hash()] == p):
+				*@extract(OpIndex, output, &pos) = {
+					OP_INDEX,
+					p.hash()
+				};
+
+			// check if we can use diff or luma
+			case (prev != p && prev.a == p.a):
+				// diff the pixels
+				diff = p.rgb - prev.rgb;
+				if (diff.r > -3 && diff.r < 2
+					&& diff.g > -3 && diff.g < 2
+					&& diff.b > -3 && diff.b < 2)
+				{
+					*@extract(OpDiff, output, &pos) = {
+						OP_DIFF,
+						(char)diff.r + 2,
+						(char)diff.g + 2,
+						(char)diff.b + 2
+					};
+					palette[p.hash()] = p;
+					break;
+				}
+				// check luma eligibility
+				luma = { diff.r - diff.g, diff.g, diff.b - diff.g };
+				if (luma.r >= -8 && luma.r <= 7
+					&& luma.g >= -32 && luma.g <= 31
+					&& luma.b >= -8 && luma.b <= 7)
+				{
+					*@extract(OpLuma, output, &pos) = {
+						OP_LUMA,
+						(char)luma.g + 32,
+						(char)luma.r + 8,
+						(char)luma.b + 8
+					};
+					palette[p.hash()] = p;
+					break;
+				}
+				nextcase;
+
+			// worst case scenario: just encode the raw pixel
+			default:
+				if (prev.a != p.a)
+				{
+					*@extract(OpRGBA, output, &pos) = { OP_RGBA, p.r, p.g, p.b, p.a };
+				}
+				else
+				{
+					*@extract(OpRGB, output, &pos) = { OP_RGB, p.r, p.g, p.b };
+				}
+				palette[p.hash()] = p;
+		}
+	}
+
+	// write end of stream
+	output[pos:END_OF_STREAM.len] = END_OF_STREAM;
+	pos += END_OF_STREAM.len;
+	
+	return output[:pos];
+}
+
+
+fn char[]! decode(char[] data, QOIDesc* desc, QOIChannels channels = AUTO, Allocator allocator = allocator::heap())
+{
+	return new_decode(data, desc, channels, allocator);
+}
+
+<*
+ Decode a QOI image from memory.
+
+ If channels is set to QOIChannels.AUTO, the function will
+ automatically determine the channels from the file's header.
+ However, if channels is RGB or RGBA, the output format will be
+ forced into this number of channels.
+
+ The desc struct will be filled with the width, height,
+ channels and colorspace of the image.
+
+ The function returns an optional, which can either be a QOIError
+ or a char[] pointing to the decoded pixels on success.
+
+ The returned pixel data should be free()d after use, or the decoding
+ and use of the data should be wrapped in a @pool() { ... }; block.
+
+ @param [in] data `The QOI image data to decode`
+ @param [&out] desc `The descriptor to fill with the image's info`
+ @param channels `The channels to be used`
+*>
+fn char[]! new_decode(char[] data, QOIDesc* desc, QOIChannels channels = AUTO, Allocator allocator = allocator::heap()) @nodiscard
+{
+	// check input data
+	if (data.len < Header.sizeof + END_OF_STREAM.len) return QOIError.INVALID_DATA?;
+
+	// get header
+	Header* header = (Header*)data.ptr;
+
+	// check magic bytes (FourCC)
+	if (bswap(header.be_magic) != 'qoif') return QOIError.INVALID_DATA?;
+
+	// copy header data to desc
+	desc.width = bswap(header.be_width);
+	desc.height = bswap(header.be_height);
+	desc.channels = @enumcast(QOIChannels, header.channels)!; 			// Rethrow if invalid
+	desc.colorspace = @enumcast(QOIColorspace, header.colorspace)!;	// Rethrow if invalid
+	if (desc.channels == AUTO) return QOIError.INVALID_DATA?; // Channels must be specified in the header
+
+	// check width and height
+	if (desc.width == 0 || desc.height == 0) return QOIError.INVALID_DATA?;
+
+	// check pixel count
+	ulong pixels = (ulong)desc.width * (ulong)desc.height;
+	if (pixels > PIXELS_MAX) return QOIError.TOO_MANY_PIXELS?;
+
+	uint pos = Header.sizeof; 	// Current position in data
+	uint loc;					// Current position in image (top-left corner)
+	char run_length = 0; 		// Length of the current run
+	char tag;					// Current chunk tag
+
+	Pixel[64] palette; // Zero-initialized by default
+	Pixel p = { 0, 0, 0, 255 };
+
+	if (channels == AUTO) channels = desc.channels;
+	
+	// allocate memory for image data
+	usz image_size = (usz)pixels * channels.id;
+	char[] image = allocator::alloc_array(allocator, char, image_size);
+	defer catch allocator::free(allocator, image);
+
+	for (loc = 0; loc < image_size; loc += channels.id)
+	{
+		// get chunk tag
+		tag = data[pos];
+
+		// check for chunk type
+		switch
+		{
+			case run_length > 0:
+				run_length--;
+
+			case tag == OP_RGB:
+				OpRGB* op = @extract(OpRGB, data, &pos);
+				p = { op.red, op.green, op.blue, p.a };
+				palette[p.hash()] = p;
+
+			case tag == OP_RGBA:
+				OpRGBA* op = @extract(OpRGBA, data, &pos);
+				p = { op.red, op.green, op.blue, op.alpha };
+				palette[p.hash()] = p;
+				
+			case tag >> 6 == OP_INDEX:
+				OpIndex* op = @extract(OpIndex, data, &pos);
+				p = palette[op.index];
+
+			case tag >> 6 == OP_DIFF:
+				OpDiff* op = @extract(OpDiff, data, &pos);
+				p.r += op.diff_red - 2;
+				p.g += op.diff_green - 2;
+				p.b += op.diff_blue - 2;
+				palette[p.hash()] = p;
+
+			case tag >> 6 == OP_LUMA:
+				OpLuma* op = @extract(OpLuma, data, &pos);
+				int diff_green = op.diff_green - 32;
+				p.r += (char)(op.diff_red_minus_green - 8 + diff_green);
+				p.g += (char)(diff_green);
+				p.b += (char)(op.diff_blue_minus_green - 8 + diff_green);
+				palette[p.hash()] = p;
+
+			case tag >> 6 == OP_RUN:
+				OpRun* op = @extract(OpRun, data, &pos);
+				run_length = op.run;
+		}
+
+		// draw the pixel
+		if (channels == RGBA) { image[loc:4] = p.rgba; } else { image[loc:3] = p.rgb; }
+	}
+	
+	return image;
+}
+
+
+
+// ***************************************************************************
+// ***                                                                     ***
+// ***    Main functions are at the top to make the file more readable.    ***
+// ***        From here on, helper functions and types are defined.        ***
+// ***                                                                     ***
+// ***************************************************************************
+module std::compression::qoi @private;
+
+// 8-bit opcodes
+const OP_RGB = 0b11111110;
+const OP_RGBA = 0b11111111;
+// 2-bit opcodes
+const OP_INDEX = 0b00;
+const OP_DIFF = 0b01;
+const OP_LUMA = 0b10;
+const OP_RUN = 0b11;
+
+struct Header @packed
+{
+	uint be_magic;	// magic bytes "qoif"
+	uint be_width;	// image width in pixels (BE)
+	uint be_height;	// image height in pixels (BE)
+
+	// informative fields
+	char channels;	// 3 = RGB, 4 = RGB
+	char colorspace;	// 0 = sRGB with linear alpha, 1 = all channels linear
+}
+
+const char[?] END_OF_STREAM = {0, 0, 0, 0, 0, 0, 0, 1};
+
+// inefficient, but it's only run once at a time
+macro @enumcast($Type, raw)
+{
+	foreach (value : $Type.values)
+	{
+		if (value.id == raw) return value;
+	}
+	return QOIError.INVALID_DATA?;
+}
+
+distinct Pixel = inline char[<4>];
+macro char Pixel.hash(Pixel p)
+{
+	return (p.r * 3 + p.g * 5 + p.b * 7 + p.a * 11) % 64;
+}
+
+struct OpRGB // No need to use @packed here, the alignment is 1 anyways.
+{
+	char tag;
+	char red;
+	char green;
+	char blue;
+}
+struct OpRGBA @packed
+{
+	char tag;
+	char red;
+	char green;
+	char blue;
+	char alpha;
+}
+bitstruct OpIndex : char
+{
+	char tag : 6..7;
+	char index : 0..5;
+}
+bitstruct OpDiff : char
+{
+	char tag : 6..7;
+	char diff_red : 4..5;
+	char diff_green : 2..3;
+	char diff_blue : 0..1;
+}
+bitstruct OpLuma : ushort @align(1)
+{
+	char tag : 6..7;
+	char diff_green : 0..5;
+	char diff_red_minus_green : 12..15;
+	char diff_blue_minus_green : 8..11;
+}
+bitstruct OpRun : char
+{
+	char tag : 6..7;
+	char run : 0..5;
+}
+
+// Macro used to locate chunks in data buffers.
+// Can be used both for reading and writing.
+macro @extract($Type, char[] data, uint* pos)
+{
+	// slice data, then double cast
+	$Type* chunk = ($Type*)data[*pos : $Type.sizeof].ptr;
+	*pos += $Type.sizeof;
+	return chunk;
+}

lib/std/core/allocators/arena_allocator.c3

@@ -0,0 +1,112 @@
+// Copyright (c) 2023 Christoffer Lerno. All rights reserved.
+// Use of this source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+module std::core::mem::allocator;
+import std::math;
+
+struct ArenaAllocator (Allocator)
+{
+	char[] data;
+	usz used;
+}
+
+<*
+ Initialize a memory arena for use using the provided bytes.
+*>
+fn ArenaAllocator* ArenaAllocator.init(&self, char[] data)
+{
+	self.data = data;
+	self.used = 0;
+	return self;
+}
+
+fn void ArenaAllocator.clear(&self)
+{
+	self.used = 0;
+}
+
+struct ArenaAllocatorHeader @local
+{
+	usz size;
+	char[?] data;
+}
+
+macro ArenaAllocator* wrap(char[] bytes)
+{
+	return (ArenaAllocator){}.init(bytes);
+}
+
+<*
+ @require ptr != null
+*>
+fn void ArenaAllocator.release(&self, void* ptr, bool) @dynamic
+{
+	assert((uptr)ptr >= (uptr)self.data.ptr, "Pointer originates from a different allocator.");
+	ArenaAllocatorHeader* header = ptr - ArenaAllocatorHeader.sizeof;
+	// Reclaim memory if it's the last element.
+	if (ptr + header.size == &self.data[self.used])
+	{
+		self.used -= header.size + ArenaAllocatorHeader.sizeof;
+	}
+}
+
+fn usz ArenaAllocator.mark(&self) @dynamic => self.used;
+fn void ArenaAllocator.reset(&self, usz mark) @dynamic => self.used = mark;
+
+<*
+ @require !alignment || math::is_power_of_2(alignment)
+ @require alignment <= mem::MAX_MEMORY_ALIGNMENT `alignment too big`
+ @require size > 0
+*>
+fn void*! ArenaAllocator.acquire(&self, usz size, AllocInitType init_type, usz alignment) @dynamic
+{
+	alignment = alignment_for_allocation(alignment);
+	usz total_len = self.data.len;
+	if (size > total_len) return AllocationFailure.CHUNK_TOO_LARGE?;
+	void* start_mem = self.data.ptr;
+	void* unaligned_pointer_to_offset = start_mem + self.used + ArenaAllocatorHeader.sizeof;
+	void* mem = mem::aligned_pointer(unaligned_pointer_to_offset, alignment);
+	usz end = (usz)(mem - self.data.ptr) + size;
+	if (end > total_len) return AllocationFailure.OUT_OF_MEMORY?;
+	self.used = end;
+	ArenaAllocatorHeader* header = mem - ArenaAllocatorHeader.sizeof;
+	header.size = size;
+	if (init_type == ZERO) mem::clear(mem, size, mem::DEFAULT_MEM_ALIGNMENT);
+	return mem;
+}
+
+<*
+ @require !alignment || math::is_power_of_2(alignment)
+ @require alignment <= mem::MAX_MEMORY_ALIGNMENT `alignment too big`
+ @require old_pointer != null
+ @require size > 0
+*>
+fn void*! ArenaAllocator.resize(&self, void *old_pointer, usz size, usz alignment) @dynamic
+{
+	alignment = alignment_for_allocation(alignment);
+	assert(old_pointer >= self.data.ptr, "Pointer originates from a different allocator.");
+	usz total_len = self.data.len;
+	if (size > total_len) return AllocationFailure.CHUNK_TOO_LARGE?;
+	ArenaAllocatorHeader* header = old_pointer - ArenaAllocatorHeader.sizeof;
+	usz old_size = header.size;
+	// Do last allocation and alignment match?
+	if (&self.data[self.used] == old_pointer + old_size && mem::ptr_is_aligned(old_pointer, alignment))
+	{
+		if (old_size >= size)
+		{
+			self.used -= old_size - size;
+		}
+		else
+		{
+			usz new_used = self.used + size - old_size;
+			if (new_used > total_len) return AllocationFailure.OUT_OF_MEMORY?;
+			self.used = new_used;
+		}
+		header.size = size;
+		return old_pointer;
+	}
+	// Otherwise just allocate new memory.
+	void* mem = self.acquire(size, NO_ZERO, alignment)!;
+	mem::copy(mem, old_pointer, old_size, mem::DEFAULT_MEM_ALIGNMENT, mem::DEFAULT_MEM_ALIGNMENT);
+	return mem;
+}

lib/std/core/allocators/dynamic_arena.c3

@@ -0,0 +1,209 @@
+// Copyright (c) 2021-2024 Christoffer Lerno. All rights reserved.
+// Use of this source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+module std::core::mem::allocator;
+import std::math;
+
+struct DynamicArenaAllocator (Allocator)
+{
+	Allocator backing_allocator;
+	DynamicArenaPage* page;
+	DynamicArenaPage* unused_page;
+	usz page_size;
+}
+
+<*
+ @param [&inout] allocator
+ @require page_size >= 128
+*>
+fn void DynamicArenaAllocator.init(&self, usz page_size, Allocator allocator)
+{
+	self.page = null;
+	self.unused_page = null;
+	self.page_size = page_size;
+	self.backing_allocator = allocator;
+}
+
+fn void DynamicArenaAllocator.free(&self)
+{
+	DynamicArenaPage* page = self.page;
+	while (page)
+	{
+		DynamicArenaPage* next_page = page.prev_arena;
+		allocator::free(self.backing_allocator, page.memory);
+		allocator::free(self.backing_allocator, page);
+		page = next_page;
+	}
+	page = self.unused_page;
+	while (page)
+	{
+		DynamicArenaPage* next_page = page.prev_arena;
+		allocator::free(self.backing_allocator, page.memory);
+		allocator::free(self.backing_allocator, page);
+		page = next_page;
+	}
+	self.page = null;
+	self.unused_page = null;
+}
+
+struct DynamicArenaPage @local
+{
+	void* memory;
+	void* prev_arena;
+	usz total;
+	usz used;
+	void* current_stack_ptr;
+}
+
+struct DynamicArenaChunk @local
+{
+	usz size;
+}
+
+<*
+ @require ptr != null
+ @require self.page != null  `tried to free pointer on invalid allocator`
+*>
+fn void DynamicArenaAllocator.release(&self, void* ptr, bool) @dynamic
+{
+	DynamicArenaPage* current_page = self.page;
+	if (ptr == current_page.current_stack_ptr)
+	{
+		current_page.used = (usz)((ptr - DEFAULT_SIZE_PREFIX) - current_page.memory);
+	}
+	current_page.current_stack_ptr = null;
+}
+
+<*
+ @require size > 0 `Resize doesn't support zeroing`
+ @require old_pointer != null `Resize doesn't handle null pointers`
+ @require self.page != null  `tried to realloc pointer on invalid allocator`
+*>
+fn void*! DynamicArenaAllocator.resize(&self, void* old_pointer, usz size, usz alignment) @dynamic
+{
+	DynamicArenaPage* current_page = self.page;
+	alignment = alignment_for_allocation(alignment);
+	usz* old_size_ptr = old_pointer - DEFAULT_SIZE_PREFIX;
+	usz old_size = *old_size_ptr;
+	// We have the old pointer and it's correctly aligned.
+	if (old_size >= size && mem::ptr_is_aligned(old_pointer, alignment))
+	{
+		*old_size_ptr = size;
+		if (current_page.current_stack_ptr == old_pointer)
+		{
+			current_page.used = (usz)((old_pointer - DEFAULT_SIZE_PREFIX) - current_page.memory);
+		}
+		return old_pointer;
+	}
+	if REUSE: (current_page.current_stack_ptr == old_pointer && mem::ptr_is_aligned(old_pointer, alignment))
+	{
+		assert(size > old_size);
+		usz add_size = size - old_size;
+		if (add_size + current_page.used > current_page.total) break REUSE;
+		*old_size_ptr = size;
+		current_page.used += add_size;
+		return old_pointer;
+	}
+	void* new_mem = self.acquire(size, NO_ZERO, alignment)!;
+	mem::copy(new_mem, old_pointer, old_size, mem::DEFAULT_MEM_ALIGNMENT);
+	return new_mem;
+}
+
+fn void DynamicArenaAllocator.reset(&self, usz mark = 0) @dynamic
+{
+	assert(mark == 0, "Unexpectedly reset dynamic arena allocator with mark %d", mark);
+	DynamicArenaPage* page = self.page;
+	DynamicArenaPage** unused_page_ptr = &self.unused_page;
+	while (page)
+	{
+		DynamicArenaPage* next_page = page.prev_arena;
+		page.used = 0;
+		DynamicArenaPage* prev_unused = *unused_page_ptr;
+		*unused_page_ptr = page;
+		page.prev_arena = prev_unused;
+		page = next_page;
+	}
+	self.page = page;
+}
+
+<*
+ @require math::is_power_of_2(alignment)
+ @require size > 0
+*>
+fn void*! DynamicArenaAllocator._alloc_new(&self, usz size, usz alignment) @local
+{
+	// First, make sure that we can align it, extending the page size if needed.
+	usz page_size = max(self.page_size, mem::aligned_offset(size + DynamicArenaChunk.sizeof + alignment, alignment));
+	assert(page_size > size + DynamicArenaChunk.sizeof);
+	// Grab the page without alignment (we do it ourselves)
+	void* mem = allocator::malloc_try(self.backing_allocator, page_size)!;
+	DynamicArenaPage*! page = allocator::new_try(self.backing_allocator, DynamicArenaPage);
+	if (catch err = page)
+	{
+		allocator::free(self.backing_allocator, mem);
+		return err?;
+	}
+	page.memory = mem;
+	void* mem_start = mem::aligned_pointer(mem + DynamicArenaChunk.sizeof, alignment);
+	assert(mem_start + size < mem + page_size);
+	DynamicArenaChunk* chunk = (DynamicArenaChunk*)mem_start - 1;
+	chunk.size = size;
+	page.prev_arena = self.page;
+	page.total = page_size;
+	page.used = mem_start + size - page.memory;
+	self.page = page;
+	page.current_stack_ptr = mem_start;
+	return mem_start;
+}
+
+<*
+ @require size > 0 `acquire expects size > 0`
+ @require !alignment || math::is_power_of_2(alignment)
+*>
+fn void*! DynamicArenaAllocator.acquire(&self, usz size, AllocInitType init_type, usz alignment) @dynamic
+{
+	alignment = alignment_for_allocation(alignment);
+	DynamicArenaPage* page = self.page;
+
+	void* ptr @noinit;
+	do SET_DONE:
+	{
+		if (!page && self.unused_page)
+		{
+			self.page = page = self.unused_page;
+			self.unused_page = page.prev_arena;
+			page.prev_arena = null;
+		}
+		if (!page)
+		{
+			ptr = self._alloc_new(size, alignment)!;
+			break SET_DONE;
+		}
+		void* start = mem::aligned_pointer(page.memory + page.used + DynamicArenaChunk.sizeof, alignment);
+		usz new_used = start - page.memory + size;
+		if ALLOCATE_NEW: (new_used > page.total)
+		{
+			if ((page = self.unused_page))
+			{
+				start = mem::aligned_pointer(page.memory + page.used + DynamicArenaChunk.sizeof, alignment);
+				new_used = start + size - page.memory;
+				if (page.total >= new_used)
+				{
+					self.unused_page = page.prev_arena;
+					page.prev_arena = self.page;
+					self.page = page;
+					break ALLOCATE_NEW;
+				}
+			}
+			ptr = self._alloc_new(size, alignment)!;
+			break SET_DONE;
+		}
+		page.used = new_used;
+		assert(start + size == page.memory + page.used);
+		ptr = start;
+		DynamicArenaChunk* chunk = (DynamicArenaChunk*)ptr - 1;
+		chunk.size = size;
+	};
+	if (init_type == ZERO) mem::clear(ptr, size, mem::DEFAULT_MEM_ALIGNMENT);
+	return ptr;
+}

lib/std/core/allocators/heap_allocator.c3

@@ -0,0 +1,209 @@
+// Copyright (c) 2021-2024 Christoffer Lerno. All rights reserved.
+// Use of this source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+
+module std::core::mem::allocator;
+import std::math;
+
+struct SimpleHeapAllocator (Allocator)
+{
+	MemoryAllocFn alloc_fn;
+	Header* free_list;
+}
+
+<*
+ @require allocator != null  "An underlying memory provider must be given"
+ @require !self.free_list "The allocator may not be already initialized"
+*>
+fn void SimpleHeapAllocator.init(&self, MemoryAllocFn allocator)
+{
+	self.alloc_fn = allocator;
+	self.free_list = null;
+}
+
+fn void*! SimpleHeapAllocator.acquire(&self, usz size, AllocInitType init_type, usz alignment) @dynamic
+{
+	if (init_type == ZERO)
+	{
+		return alignment > 0 ? @aligned_alloc(self._calloc, size, alignment) : self._calloc(size);
+	}
+	return alignment > 0 ? @aligned_alloc(self._alloc, size, alignment) : self._alloc(size);
+}
+
+fn void*! SimpleHeapAllocator.resize(&self, void* old_pointer, usz size, usz alignment) @dynamic
+{
+	return alignment > 0
+		? @aligned_realloc(self._calloc, self._free, old_pointer, size, alignment)
+		:  self._realloc(old_pointer, size);
+}
+
+fn void SimpleHeapAllocator.release(&self, void* old_pointer, bool aligned) @dynamic
+{
+	if (aligned)
+	{
+		@aligned_free(self._free, old_pointer)!!;
+	}
+	else
+	{
+		self._free(old_pointer);
+	}
+}
+
+<*
+ @require old_pointer && bytes > 0
+*>
+fn void*! SimpleHeapAllocator._realloc(&self, void* old_pointer, usz bytes) @local
+{
+	// Find the block header.
+	Header* block = (Header*)old_pointer - 1;
+	if (block.size >= bytes) return old_pointer;
+	void* new = self._alloc(bytes)!;
+	usz max_to_copy = math::min(block.size, bytes);
+	mem::copy(new, old_pointer, max_to_copy);
+	self._free(old_pointer);
+	return new;
+}
+
+fn void*! SimpleHeapAllocator._calloc(&self, usz bytes) @local
+{
+	void* data = self._alloc(bytes)!;
+	mem::clear(data, bytes, mem::DEFAULT_MEM_ALIGNMENT);
+	return data;
+}
+
+fn void*! SimpleHeapAllocator._alloc(&self, usz bytes) @local
+{
+	usz aligned_bytes = mem::aligned_offset(bytes, mem::DEFAULT_MEM_ALIGNMENT);
+	if (!self.free_list)
+	{
+		self.add_block(aligned_bytes)!;
+	}
+
+	Header* current = self.free_list;
+	Header* previous = current;
+	while (current)
+	{
+		switch
+		{
+			case current.size >= aligned_bytes && current.size <= aligned_bytes + Header.sizeof + 64:
+				if (current == previous)
+				{
+					self.free_list = current.next;
+				}
+				else
+				{
+					previous.next = current.next;
+				}
+				current.next = null;
+				return current + 1;
+			case current.size > aligned_bytes:
+				Header* unallocated = (Header*)((char*)current + aligned_bytes + Header.sizeof);
+				unallocated.size = current.size - aligned_bytes - Header.sizeof;
+				unallocated.next = current.next;
+				if (current == self.free_list)
+				{
+					self.free_list = unallocated;
+				}
+				else
+				{
+					previous.next = unallocated;
+				}
+				current.size = aligned_bytes;
+				current.next = null;
+				return current + 1;
+			default:
+				previous = current;
+				current = current.next;
+		}
+	}
+	self.add_block(aligned_bytes)!;
+	return self._alloc(aligned_bytes);
+}
+
+fn void! SimpleHeapAllocator.add_block(&self, usz aligned_bytes) @local
+{
+	assert(mem::aligned_offset(aligned_bytes, mem::DEFAULT_MEM_ALIGNMENT) == aligned_bytes);
+	char[] result = self.alloc_fn(aligned_bytes + Header.sizeof)!;
+	Header* new_block = (Header*)result.ptr;
+	new_block.size = result.len - Header.sizeof;
+	new_block.next = null;
+	self._free(new_block + 1);
+}
+
+
+fn void SimpleHeapAllocator._free(&self, void* ptr) @local
+{
+	// Empty ptr -> do nothing.
+	if (!ptr) return;
+
+	// Find the block header.
+	Header* block = (Header*)ptr - 1;
+
+	// No free list? Then just return self.
+	if (!self.free_list)
+	{
+		self.free_list = block;
+		return;
+	}
+
+	// Find where in the list it should be inserted.
+	Header* current = self.free_list;
+	Header* prev = current;
+	while (current)
+	{
+		if (block < current)
+		{
+			// Between prev and current
+			if (block > prev) break;
+			// Before current
+			if (current == prev) break;
+		}
+		prev = current;
+		current = prev.next;
+	}
+	if (current)
+	{
+		// Insert after the current block.
+		// Are the blocks adjacent?
+		if (current == (Header*)((char*)(block + 1) + block.size))
+		{
+			// Merge
+			block.size += current.size + Header.sizeof;
+			block.next = current.next;
+		}
+		else
+		{
+			// Chain to current
+			block.next = current;
+		}
+	}
+	if (prev == current)
+	{
+		// Swap new start of free list
+		self.free_list = block;
+	}
+	else
+	{
+		// Prev adjacent?
+		if (block == (Header*)((char*)(prev + 1) + prev.size))
+		{
+			prev.size += block.size + Header.sizeof;
+			prev.next = block.next;
+		}
+		else
+		{
+			// Link prev to block
+			prev.next = block;
+		}
+	}
+}
+
+union Header @local
+{
+	struct
+	{
+		Header* next;
+		usz size;
+	}
+	usz align;
+}

lib/std/core/allocators/libc_allocator.c3

@@ -0,0 +1,161 @@
+// Copyright (c) 2021-2024 Christoffer Lerno. All rights reserved.
+// Use of this source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+
+module std::core::mem::allocator @if(env::LIBC);
+import std::io;
+import libc;
+
+const LibcAllocator LIBC_ALLOCATOR = {};
+distinct LibcAllocator (Allocator, Printable) = uptr;
+
+fn String LibcAllocator.to_string(&self, Allocator allocator) @dynamic => "Libc allocator".copy(allocator);
+fn usz! LibcAllocator.to_format(&self, Formatter *format) @dynamic => format.print("Libc allocator");
+
+module std::core::mem::allocator @if(env::POSIX);
+import std::os;
+import libc;
+
+
+fn void*! LibcAllocator.acquire(&self, usz bytes, AllocInitType init_type, usz alignment) @dynamic
+{
+	if (init_type == ZERO)
+	{
+		void* data @noinit;
+		if (alignment > mem::DEFAULT_MEM_ALIGNMENT)
+		{
+			if (posix::posix_memalign(&data, alignment, bytes)) return AllocationFailure.OUT_OF_MEMORY?;
+			mem::clear(data, bytes, mem::DEFAULT_MEM_ALIGNMENT);
+			return data;
+		}
+		return libc::calloc(1, bytes) ?: AllocationFailure.OUT_OF_MEMORY?;
+	}
+	else
+	{
+		void* data @noinit;
+		if (alignment > mem::DEFAULT_MEM_ALIGNMENT)
+		{
+			if (posix::posix_memalign(&data, alignment, bytes)) return AllocationFailure.OUT_OF_MEMORY?;
+		}
+		else
+		{
+			if (!(data = libc::malloc(bytes))) return AllocationFailure.OUT_OF_MEMORY?;
+		}
+		$if env::TESTING:
+			for (usz i = 0; i < bytes; i++) ((char*)data)[i] = 0xAA;
+		$endif
+		return data;
+	}
+}
+
+fn void*! LibcAllocator.resize(&self, void* old_ptr, usz new_bytes, usz alignment) @dynamic
+{
+	if (alignment <= mem::DEFAULT_MEM_ALIGNMENT) return libc::realloc(old_ptr, new_bytes) ?: AllocationFailure.OUT_OF_MEMORY?;
+	void* new_ptr;
+	if (posix::posix_memalign(&new_ptr, alignment, new_bytes)) return AllocationFailure.OUT_OF_MEMORY?;
+
+	$switch
+	$case env::DARWIN:
+		usz old_usable_size = darwin::malloc_size(old_ptr);
+	$case env::LINUX:
+		usz old_usable_size = linux::malloc_usable_size(old_ptr);
+	$default:
+		usz old_usable_size = new_bytes;
+	$endswitch
+
+	usz copy_size = new_bytes < old_usable_size ? new_bytes : old_usable_size;
+	mem::copy(new_ptr, old_ptr, copy_size, mem::DEFAULT_MEM_ALIGNMENT, mem::DEFAULT_MEM_ALIGNMENT);
+	libc::free(old_ptr);
+	return new_ptr;
+}
+
+fn void LibcAllocator.release(&self, void* old_ptr, bool aligned) @dynamic
+{
+	libc::free(old_ptr);
+}
+
+module std::core::mem::allocator @if(env::WIN32);
+import std::os::win32;
+import libc;
+
+fn void*! LibcAllocator.acquire(&self, usz bytes, AllocInitType init_type, usz alignment) @dynamic
+{
+	if (init_type == ZERO)
+	{
+		if (alignment > 0)
+		{
+			return win32::_aligned_recalloc(null, 1, bytes, alignment) ?: AllocationFailure.OUT_OF_MEMORY?;
+		}
+		return libc::calloc(1, bytes) ?: AllocationFailure.OUT_OF_MEMORY?;
+	}
+	void* data = alignment > 0 ? win32::_aligned_malloc(bytes, alignment) : libc::malloc(bytes);
+	if (!data) return AllocationFailure.OUT_OF_MEMORY?;
+	$if env::TESTING:
+		for (usz i = 0; i < bytes; i++) ((char*)data)[i] = 0xAA;
+	$endif
+	return data;
+}
+
+fn void*! LibcAllocator.resize(&self, void* old_ptr, usz new_bytes, usz alignment) @dynamic
+{
+	if (alignment)
+	{
+		return win32::_aligned_realloc(old_ptr, new_bytes, alignment) ?: AllocationFailure.OUT_OF_MEMORY?;
+	}
+	return libc::realloc(old_ptr, new_bytes) ?: AllocationFailure.OUT_OF_MEMORY?;
+}
+
+fn void LibcAllocator.release(&self, void* old_ptr, bool aligned) @dynamic
+{
+	if (aligned)
+	{
+		win32::_aligned_free(old_ptr);
+		return;
+	}
+	libc::free(old_ptr);
+}
+
+module std::core::mem::allocator @if(!env::WIN32 && !env::POSIX && env::LIBC);
+import libc;
+
+fn void*! LibcAllocator.acquire(&self, usz bytes, AllocInitType init_type, usz alignment) @dynamic
+{
+	if (init_type == ZERO)
+	{
+		void* data = alignment ? @aligned_alloc(fn void*(usz bytes) => libc::calloc(bytes, 1), bytes, alignment)!! : libc::calloc(bytes, 1);
+		return data ?: AllocationFailure.OUT_OF_MEMORY?;
+	}
+	else
+	{
+		void* data = alignment ? @aligned_alloc(libc::malloc, bytes, alignment)!! : libc::malloc(bytes);
+		if (!data) return AllocationFailure.OUT_OF_MEMORY?;
+		$if env::TESTING:
+			for (usz i = 0; i < bytes; i++) ((char*)data)[i] = 0xAA;
+		$endif
+		return data;
+	}
+}
+
+
+fn void*! LibcAllocator.resize(&self, void* old_ptr, usz new_bytes, usz alignment) @dynamic
+{
+	if (alignment)
+	{
+		void* data = @aligned_realloc(fn void*(usz bytes) => libc::malloc(bytes), libc::free, old_ptr, new_bytes, alignment)!!;
+		return data ?: AllocationFailure.OUT_OF_MEMORY?;
+	}
+	return libc::realloc(old_ptr, new_bytes) ?: AllocationFailure.OUT_OF_MEMORY?;
+}
+
+
+fn void LibcAllocator.release(&self, void* old_ptr, bool aligned) @dynamic
+{
+	if (aligned)
+	{
+		@aligned_free(libc::free, old_ptr)!!;
+	}
+	else
+	{
+		libc::free(old_ptr);
+	}
+}

lib/std/core/allocators/on_stack_allocator.c3

@@ -0,0 +1,150 @@
+module std::core::mem::allocator;
+
+struct OnStackAllocator (Allocator)
+{
+	Allocator backing_allocator;
+	char[] data;
+	usz used;
+	OnStackAllocatorExtraChunk* chunk;
+}
+
+
+struct OnStackAllocatorExtraChunk @local
+{
+	bool is_aligned;
+	OnStackAllocatorExtraChunk* prev;
+	void* data;
+}
+
+<*
+ Initialize a memory arena for use using the provided bytes.
+
+ @param [&inout] allocator
+*>
+fn void OnStackAllocator.init(&self, char[] data, Allocator allocator)
+{
+	self.data = data;
+	self.backing_allocator = allocator;
+	self.used = 0;
+}
+
+fn void OnStackAllocator.free(&self)
+{
+	OnStackAllocatorExtraChunk* chunk = self.chunk;
+	while (chunk)
+	{
+		if (chunk.is_aligned)
+		{
+			allocator::free_aligned(self.backing_allocator, chunk.data);
+		}
+		else
+		{
+			allocator::free(self.backing_allocator, chunk.data);
+		}
+		void* old = chunk;
+		chunk = chunk.prev;
+		allocator::free(self.backing_allocator, old);
+	}
+	self.chunk = null;
+	self.used = 0;
+}
+
+struct OnStackAllocatorHeader
+{
+	usz size;
+	char[?] data;
+}
+
+<*
+ @require old_pointer != null
+*>
+fn void OnStackAllocator.release(&self, void* old_pointer, bool aligned) @dynamic
+{
+	if (allocation_in_stack_mem(self, old_pointer)) return;
+	on_stack_allocator_remove_chunk(self, old_pointer);
+	self.backing_allocator.release(old_pointer, aligned);
+}
+
+fn bool allocation_in_stack_mem(OnStackAllocator* a, void* ptr) @local
+{
+	return ptr >= a.data.ptr && ptr <= &a.data[^1];
+}
+
+fn void on_stack_allocator_remove_chunk(OnStackAllocator* a, void* ptr) @local
+{
+	OnStackAllocatorExtraChunk* chunk = a.chunk;
+	OnStackAllocatorExtraChunk** addr = &a.chunk;
+	while (chunk)
+	{
+		if (chunk.data == ptr)
+		{
+			*addr = chunk.prev;
+			allocator::free(a.backing_allocator, chunk);
+			return;
+		}
+		addr = &chunk.prev;
+		chunk = *addr;
+	}
+	unreachable("Missing chunk");
+}
+
+fn OnStackAllocatorExtraChunk* on_stack_allocator_find_chunk(OnStackAllocator* a, void* ptr) @local
+{
+	OnStackAllocatorExtraChunk* chunk = a.chunk;
+	while (chunk)
+	{
+		if (chunk.data == ptr) return chunk;
+		chunk = chunk.prev;
+	}
+	return null;
+}
+
+<*
+ @require size > 0
+ @require old_pointer != null
+ @require alignment <= mem::MAX_MEMORY_ALIGNMENT `alignment too big`
+*>
+fn void*! OnStackAllocator.resize(&self, void* old_pointer, usz size, usz alignment) @dynamic
+{
+	if (!allocation_in_stack_mem(self, old_pointer))
+	{
+		OnStackAllocatorExtraChunk* chunk = on_stack_allocator_find_chunk(self, old_pointer);
+		assert(chunk, "Tried to realloc pointer not belonging to the allocator");
+		return chunk.data = self.backing_allocator.resize(old_pointer, size, alignment)!;
+	}
+
+	OnStackAllocatorHeader* header = old_pointer - OnStackAllocatorHeader.sizeof;
+	usz old_size = header.size;
+	void* mem = self.acquire(size, NO_ZERO, alignment)!;
+	mem::copy(mem, old_pointer, old_size, mem::DEFAULT_MEM_ALIGNMENT, mem::DEFAULT_MEM_ALIGNMENT);
+	return mem;
+}
+
+<*
+ @require alignment <= mem::MAX_MEMORY_ALIGNMENT `alignment too big`
+ @require size > 0
+*>
+fn void*! OnStackAllocator.acquire(&self, usz size, AllocInitType init_type, usz alignment) @dynamic
+{
+	bool aligned = alignment > 0;
+	alignment = alignment_for_allocation(alignment);
+	usz total_len = self.data.len;
+	void* start_mem = self.data.ptr;
+	void* unaligned_pointer_to_offset = start_mem + self.used + OnStackAllocatorHeader.sizeof ;
+	void* mem = mem::aligned_pointer(unaligned_pointer_to_offset, alignment);
+	usz end = (usz)(mem - self.data.ptr) + size;
+	Allocator backing_allocator = self.backing_allocator;
+
+	if (end > total_len)
+	{
+		OnStackAllocatorExtraChunk* chunk = allocator::alloc_try(backing_allocator, OnStackAllocatorExtraChunk)!;
+		defer catch allocator::free(backing_allocator, chunk);
+		defer try self.chunk = chunk;
+		*chunk = { .prev = self.chunk, .is_aligned = aligned };
+		return chunk.data = backing_allocator.acquire(size, init_type, aligned ? alignment : 0)!;
+	}
+	self.used = end;
+	OnStackAllocatorHeader* header = mem - OnStackAllocatorHeader.sizeof;
+	header.size = size;
+	return mem;
+}

lib/std/core/allocators/temp_allocator.c3

@@ -0,0 +1,230 @@
+module std::core::mem::allocator;
+import std::io, std::math;
+
+struct TempAllocatorChunk @local
+{
+	usz size;
+	char[?] data;
+}
+
+struct TempAllocator (Allocator)
+{
+	Allocator backing_allocator;
+	TempAllocatorPage* last_page;
+	usz used;
+	usz capacity;
+	char[?] data;
+}
+
+const usz PAGE_IS_ALIGNED @private = (usz)isz.max + 1u;
+
+
+struct TempAllocatorPage
+{
+	TempAllocatorPage* prev_page;
+	void* start;
+	usz mark;
+	usz size;
+	usz ident;
+	char[?] data;
+}
+
+macro usz TempAllocatorPage.pagesize(&self) => self.size & ~PAGE_IS_ALIGNED;
+macro bool TempAllocatorPage.is_aligned(&self) => self.size & PAGE_IS_ALIGNED == PAGE_IS_ALIGNED;
+
+<*
+ @require size >= 16
+*>
+fn TempAllocator*! new_temp_allocator(usz size, Allocator allocator)
+{
+	TempAllocator* temp = allocator::alloc_with_padding(allocator, TempAllocator, size)!;
+	temp.last_page = null;
+	temp.backing_allocator = allocator;
+	temp.used = 0;
+	temp.capacity = size;
+	return temp;
+}
+
+fn void TempAllocator.destroy(&self)
+{
+	self.reset(0);
+	if (self.last_page) (void)self._free_page(self.last_page);
+	allocator::free(self.backing_allocator, self);
+}
+
+fn usz TempAllocator.mark(&self) @dynamic => self.used;
+
+fn void TempAllocator.release(&self, void* old_pointer, bool) @dynamic
+{
+	usz old_size = *(usz*)(old_pointer - DEFAULT_SIZE_PREFIX);
+	if (old_pointer + old_size == &self.data[self.used])
+	{
+		self.used -= old_size;
+		asan::poison_memory_region(&self.data[self.used], old_size);
+	}
+}
+fn void TempAllocator.reset(&self, usz mark) @dynamic
+{
+	TempAllocatorPage *last_page = self.last_page;
+	while (last_page && last_page.mark > mark)
+	{
+		self.used = last_page.mark;
+		TempAllocatorPage *to_free = last_page;
+		last_page = last_page.prev_page;
+		self._free_page(to_free)!!;
+	}
+	self.last_page = last_page;
+	$if env::COMPILER_SAFE_MODE || env::ADDRESS_SANITIZER:
+		if (!last_page)
+		{
+			usz cleaned = self.used - mark;
+			if (cleaned > 0)
+			{
+				$if env::COMPILER_SAFE_MODE && !env::ADDRESS_SANITIZER:
+					self.data[mark : cleaned] = 0xAA;
+				$endif
+				asan::poison_memory_region(&self.data[mark], cleaned);
+			}
+		}
+	$endif
+	self.used = mark;
+}
+
+fn void! TempAllocator._free_page(&self, TempAllocatorPage* page) @inline @local
+{
+	void* mem = page.start;
+	return self.backing_allocator.release(mem, page.is_aligned());
+}
+
+fn void*! TempAllocator._realloc_page(&self, TempAllocatorPage* page, usz size, usz alignment) @inline @local
+{
+	// Then the actual start pointer:
+	void* real_pointer = page.start;
+
+	// Walk backwards to find the pointer to this page.
+	TempAllocatorPage **pointer_to_prev = &self.last_page;
+	// Remove the page from the list
+	while (*pointer_to_prev != page)
+	{
+		pointer_to_prev = &((*pointer_to_prev).prev_page);
+	}
+	*pointer_to_prev = page.prev_page;
+	usz page_size = page.pagesize();
+	// Clear on size > original size.
+	void* data = self.acquire(size, NO_ZERO, alignment)!;
+	mem::copy(data, &page.data[0], page_size, mem::DEFAULT_MEM_ALIGNMENT, mem::DEFAULT_MEM_ALIGNMENT);
+	self.backing_allocator.release(real_pointer, page.is_aligned());
+	return data;
+}
+
+fn void*! TempAllocator.resize(&self, void* pointer, usz size, usz alignment) @dynamic
+{
+	TempAllocatorChunk *chunk = pointer - TempAllocatorChunk.sizeof;
+	if (chunk.size == (usz)-1)
+	{
+		assert(self.last_page, "Realloc of non temp pointer");
+		// First grab the page
+		TempAllocatorPage *page = pointer - TempAllocatorPage.sizeof;
+		return self._realloc_page(page, size, alignment);
+	}
+
+	TempAllocatorChunk* data = self.acquire(size, NO_ZERO, alignment)!;
+	mem::copy(data, pointer, chunk.size, mem::DEFAULT_MEM_ALIGNMENT, mem::DEFAULT_MEM_ALIGNMENT);
+
+	return data;
+}
+
+<*
+ @require size > 0
+ @require !alignment || math::is_power_of_2(alignment)
+ @require alignment <= mem::MAX_MEMORY_ALIGNMENT `alignment too big`
+*>
+fn void*! TempAllocator.acquire(&self, usz size, AllocInitType init_type, usz alignment) @dynamic
+{
+	alignment = alignment_for_allocation(alignment);
+	void* start_mem = &self.data;
+	void* starting_ptr = start_mem + self.used;
+	void* aligned_header_start = mem::aligned_pointer(starting_ptr, TempAllocatorChunk.alignof);
+	void* mem = aligned_header_start + TempAllocatorChunk.sizeof;
+	if (alignment > TempAllocatorChunk.alignof)
+	{
+		mem = mem::aligned_pointer(mem, alignment);
+	}
+	usz new_usage = (usz)(mem - start_mem) + size;
+
+	// Arena allocation, simple!
+	if (new_usage <= self.capacity)
+	{
+		asan::unpoison_memory_region(starting_ptr, new_usage - self.used);
+		TempAllocatorChunk* chunk_start = mem - TempAllocatorChunk.sizeof;
+		chunk_start.size = size;
+		self.used = new_usage;
+		if (init_type == ZERO) mem::clear(mem, size, mem::DEFAULT_MEM_ALIGNMENT);
+		return mem;
+	}
+
+	// Fallback to backing allocator
+	TempAllocatorPage* page;
+
+	// We have something we need to align.
+	if (alignment > mem::DEFAULT_MEM_ALIGNMENT)
+	{
+		// This is actually simpler, since it will create the offset for us.
+		usz total_alloc_size = mem::aligned_offset(TempAllocatorPage.sizeof + size, alignment);
+		if (init_type == ZERO)
+		{
+			mem = allocator::calloc_aligned(self.backing_allocator, total_alloc_size, alignment)!;
+		}
+		else
+		{
+			mem = allocator::malloc_aligned(self.backing_allocator, total_alloc_size, alignment)!;
+		}
+		void* start = mem;
+		mem += mem::aligned_offset(TempAllocatorPage.sizeof, alignment);
+		page = (TempAllocatorPage*)mem - 1;
+		page.start = start;
+		page.size = size | PAGE_IS_ALIGNED;
+	}
+	else
+	{
+		// Here we might need to pad
+		usz padded_header_size = mem::aligned_offset(TempAllocatorPage.sizeof, mem::DEFAULT_MEM_ALIGNMENT);
+		usz total_alloc_size = padded_header_size + size;
+		void* alloc = self.backing_allocator.acquire(total_alloc_size, init_type, 0)!;
+
+		// Find the page.
+		page = alloc + padded_header_size - TempAllocatorPage.sizeof;
+		assert(mem::ptr_is_aligned(page, TempAllocator.alignof));
+		assert(mem::ptr_is_aligned(&page.data[0], mem::DEFAULT_MEM_ALIGNMENT));
+		page.start = alloc;
+		page.size = size;
+	}
+
+	// Mark it as a page
+	page.ident = ~(usz)0;
+	// Store when it was created
+	page.mark = ++self.used;
+	// Hook up the page.
+	page.prev_page = self.last_page;
+	self.last_page = page;
+	return &page.data[0];
+}
+
+fn void! TempAllocator.print_pages(&self, File* f)
+{
+	TempAllocatorPage *last_page = self.last_page;
+	if (!last_page)
+	{
+		io::fprintf(f, "No pages.\n")!;
+		return;
+	}
+	io::fprintf(f, "---Pages----\n")!;
+	uint index = 0;
+	while (last_page)
+	{
+		bool is_not_aligned = !(last_page.size & (1u64 << 63));
+		io::fprintf(f, "%d. Alloc: %d %d at %p%s\n", ++index,
+				last_page.size & ~(1u64 << 63), last_page.mark, &last_page.data[0], is_not_aligned ? "" : " [aligned]")!;
+		last_page = last_page.prev_page;
+	}
+}

lib/std/core/allocators/tracking_allocator.c3

@@ -0,0 +1,215 @@
+// Copyright (c) 2021-2024 Christoffer Lerno. All rights reserved.
+// Use of this source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+
+module std::core::mem::allocator;
+import std::collections, std::io, std::os::backtrace;
+
+const MAX_BACKTRACE = 16;
+struct Allocation
+{
+	void* ptr;
+	usz size;
+	void*[MAX_BACKTRACE] backtrace;
+}
+
+def AllocMap = HashMap(<uptr, Allocation>);
+
+// A simple tracking allocator.
+// It tracks allocations using a hash map but
+// is not compatible with allocators that uses mark()
+struct TrackingAllocator (Allocator)
+{
+	Allocator inner_allocator;
+	AllocMap map;
+	usz mem_total;
+	usz allocs_total;
+}
+
+<*
+ Initialize a tracking allocator to wrap (and track) another allocator.
+
+ @param [&inout] allocator "The allocator to track"
+*>
+fn void TrackingAllocator.init(&self, Allocator allocator)
+{
+	*self = { .inner_allocator = allocator };
+	self.map.init(allocator);
+}
+
+<*
+ Free this tracking allocator.
+*>
+fn void TrackingAllocator.free(&self)
+{
+	self.map.free();
+	*self = {};
+}
+
+<*
+ @return "the total allocated memory not yet freed."
+*>
+fn usz TrackingAllocator.allocated(&self) => @pool()
+{
+	usz allocated = 0;
+	foreach (&allocation : self.map.value_tlist()) allocated += allocation.size;
+	return allocated;
+}
+
+<*
+ @return "the total memory allocated (freed or not)."
+*>
+fn usz TrackingAllocator.total_allocated(&self) => self.mem_total;
+
+<*
+ @return "the total number of allocations (freed or not)."
+*>
+fn usz TrackingAllocator.total_allocation_count(&self) => self.allocs_total;
+
+fn Allocation[] TrackingAllocator.allocations_tlist(&self, Allocator allocator)
+{
+	return self.map.value_tlist();
+}
+
+<*
+ @return "the number of non-freed allocations."
+*>
+fn usz TrackingAllocator.allocation_count(&self) => self.map.count;
+
+fn void*! TrackingAllocator.acquire(&self, usz size, AllocInitType init_type, usz alignment) @dynamic
+{
+	void* data = self.inner_allocator.acquire(size, init_type, alignment)!;
+	self.allocs_total++;
+	void*[MAX_BACKTRACE] bt;
+	backtrace::capture_current(&bt);
+	self.map.set((uptr)data, { data, size, bt });
+	self.mem_total += size;
+	return data;
+}
+
+fn void*! TrackingAllocator.resize(&self, void* old_pointer, usz size, usz alignment) @dynamic
+{
+	void* data = self.inner_allocator.resize(old_pointer, size, alignment)!;
+	self.map.remove((uptr)old_pointer);
+	void*[MAX_BACKTRACE] bt;
+	backtrace::capture_current(&bt);
+	self.map.set((uptr)data, { data, size, bt });
+	self.mem_total += size;
+	self.allocs_total++;
+	return data;
+}
+
+fn void TrackingAllocator.release(&self, void* old_pointer, bool is_aligned) @dynamic
+{
+	if (catch self.map.remove((uptr)old_pointer))
+	{
+		unreachable("Attempt to release untracked pointer %p, this is likely a bug.", old_pointer);
+	}
+	self.inner_allocator.release(old_pointer, is_aligned);
+}
+
+fn void TrackingAllocator.clear(&self)
+{
+	self.map.clear();
+}
+
+fn bool TrackingAllocator.has_leaks(&self)
+{
+	return self.map.len() > 0;
+}
+
+fn void TrackingAllocator.print_report(&self) => self.fprint_report(io::stdout())!!;
+
+
+fn void! TrackingAllocator.fprint_report(&self, OutStream out) => @pool()
+{
+	usz total = 0;
+	usz entries = 0;
+	bool leaks = false;
+
+	Allocation[] allocs = self.map.value_tlist();
+	if (allocs.len)
+	{
+		if (!allocs[0].backtrace[0])
+		{
+			io::fprintn(out, "======== Memory Report ========")!;
+			io::fprintn(out, "Size in bytes   Address")!;
+			foreach (i, &allocation : allocs)
+			{
+				entries++;
+				total += allocation.size;
+				io::fprintfn(out, "%13s   %p", allocation.size, allocation.ptr)!;
+			}
+			io::fprintn(out, "===============================")!;
+
+		}
+		else
+		{
+			io::fprintn(out, "================================== Memory Report ==================================")!;
+			io::fprintn(out, "Size in bytes   Address          Function                       ")!;
+			foreach (i, &allocation : allocs)
+			{
+				entries++;
+				total += allocation.size;
+				BacktraceList backtraces = {};
+				Backtrace trace = backtrace::BACKTRACE_UNKNOWN;
+				if (allocation.backtrace[3])
+				{
+					trace = backtrace::symbolize_backtrace(allocation.backtrace[3:1], allocator::temp()).get(0) ?? backtrace::BACKTRACE_UNKNOWN;
+				}
+				if (trace.function.len)	leaks = true;
+				io::fprintfn(out, "%13s   %p   %s:%d", allocation.size,
+					allocation.ptr, trace.function.len ? trace.function : "???",
+					trace.line ? trace.line : 0)!;
+			}
+			io::fprintn(out, "===================================================================================")!;
+		}
+	}
+	else
+	{
+		io::fprintn(out, "* NO ALLOCATIONS FOUND *")!;
+	}
+	io::fprintfn(out, "- Total currently allocated memory:            %d", total)!;
+	io::fprintfn(out, "- Total current allocations:                   %d", entries)!;
+	io::fprintfn(out, "- Total allocations (freed and retained):      %d", self.allocs_total)!;
+	io::fprintfn(out, "- Total allocated memory (freed and retained): %d", self.mem_total)!;
+	if (leaks)
+	{
+		io::fprintn(out)!;
+		io::fprintn(out, "Full leak report:")!;
+		foreach (i, &allocation : allocs)
+		{
+			if (!allocation.backtrace[3])
+			{
+				io::fprintfn(out, "Allocation %d (%d bytes) - no backtrace available.", i + 1, allocation.size)!;
+				continue;
+			}
+			BacktraceList backtraces = {};
+			usz end = MAX_BACKTRACE;
+			foreach (j, val : allocation.backtrace)
+			{
+				if (!val)
+				{
+					end = j;
+					break;
+				}
+			}
+			BacktraceList list = backtrace::symbolize_backtrace(allocation.backtrace[3..(end - 1)], allocator::temp())!;
+			io::fprintfn(out, "Allocation %d (%d bytes): ", i + 1, allocation.size)!;
+			foreach (trace : list)
+			{
+				if (trace.has_file())
+				{
+					io::fprintfn(out, "   %s (in %s:%d)", trace.function, trace.file, trace.line);
+					continue;
+				}
+				if (trace.is_unknown())
+				{
+					io::fprintfn(out, "   ??? (in unknown)");
+					continue;
+				}
+				io::fprintfn(out, "   %s (source unavailable)", trace.function);
+			}
+		}
+	}
+}

lib/std/core/array.c3

@@ -0,0 +1,193 @@
+module std::core::array;
+import std::core::array::slice;
+
+<*
+ @param [in] array
+ @param [in] element
+ @return "the first index of the element"
+ @return! SearchResult.MISSING
+*>
+macro index_of(array, element)
+{
+	foreach (i, &e : array)
+	{
+		if (*e == element) return i;
+	}
+	return SearchResult.MISSING?;
+}
+
+<*
+ @require @typekind(array_ptr) == POINTER
+ @require @typekind(*array_ptr) == VECTOR || @typekind(*array_ptr) == ARRAY
+ @require @typekind((*array_ptr)[0]) == VECTOR || @typekind((*array_ptr)[0]) == ARRAY
+*>
+macro slice2d(array_ptr, x = 0, xlen = 0, y = 0, ylen = 0)
+{
+	if (xlen < 1) xlen = $typeof((*array_ptr)[0]).len + xlen;
+	if (ylen < 1) ylen = $typeof((*array_ptr)).len + ylen;
+	var $ElementType = $typeof((*array_ptr)[0][0]);
+	return Slice2d(<$ElementType>) { ($ElementType*)array_ptr, $typeof((*array_ptr)[0]).len, y, ylen, x, xlen };
+}
+
+
+<*
+ @param [in] array
+ @param [in] element
+ @return "the last index of the element"
+ @return! SearchResult.MISSING
+*>
+macro rindex_of(array, element)
+{
+	foreach_r (i, &e : array)
+	{
+		if (*e == element) return i;
+	}
+	return SearchResult.MISSING?;
+}
+
+<*
+ Concatenate two arrays or slices, returning a slice containing the concatenation of them.
+
+ @param [in] arr1
+ @param [in] arr2
+ @param [&inout] allocator "The allocator to use, default is the heap allocator"
+ @require @typekind(arr1) == SLICE || @typekind(arr1) == ARRAY
+ @require @typekind(arr2) == SLICE || @typekind(arr2) == ARRAY
+ @require @typeis(arr1[0], $typeof(arr2[0])) "Arrays must have the same type"
+ @ensure result.len == arr1.len + arr2.len
+*>
+macro concat(arr1, arr2, Allocator allocator) @nodiscard
+{
+	var $Type = $typeof(arr1[0]);
+	$Type[] result = allocator::alloc_array(allocator, $Type, arr1.len + arr2.len);
+	if (arr1.len > 0)
+	{
+		mem::copy(result.ptr, &arr1[0], arr1.len * $Type.sizeof, $Type.alignof, $Type.alignof);
+	}
+	if (arr2.len > 0)
+	{
+		mem::copy(&result[arr1.len], &arr2[0], arr2.len * $Type.sizeof, $Type.alignof, $Type.alignof);
+	}
+	return result;
+}
+<*
+ Concatenate two arrays or slices, returning a slice containing the concatenation of them.
+
+ @param [in] arr1
+ @param [in] arr2
+ @param [&inout] allocator "The allocator to use, default is the heap allocator"
+ @require @typekind(arr1) == SLICE || @typekind(arr1) == ARRAY
+ @require @typekind(arr2) == SLICE || @typekind(arr2) == ARRAY
+ @require @typeis(arr1[0], $typeof(arr2[0])) "Arrays must have the same type"
+ @ensure return.len == arr1.len + arr2.len
+*>
+macro concat_new(arr1, arr2, Allocator allocator = allocator::heap()) @nodiscard
+{
+	return concat(arr1, arr2, allocator);
+}
+
+<*
+ Concatenate two arrays or slices, returning a slice containing the concatenation of them,
+ allocated using the temp allocator.
+
+ @param [in] arr1
+ @param [in] arr2
+ @require @typekind(arr1) == SLICE || @typekind(arr1) == ARRAY
+ @require @typekind(arr2) == SLICE || @typekind(arr2) == ARRAY
+ @require @typeis(arr1[0], $typeof(arr2[0])) "Arrays must have the same type"
+ @ensure return.len == arr1.len + arr2.len
+*>
+macro tconcat(arr1, arr2) @nodiscard => concat(arr1, arr2, allocator::temp());
+
+module std::core::array::slice(<Type>);
+
+struct Slice2d
+{
+	Type* ptr;
+	usz inner_len;
+	usz ystart;
+	usz ylen;
+	usz xstart;
+	usz xlen;
+}
+
+fn usz Slice2d.len(&self) @operator(len)
+{
+	return self.ylen;
+}
+
+fn usz Slice2d.count(&self)
+{
+	return self.ylen * self.xlen;
+}
+
+macro void Slice2d.@each(&self; @body(usz[<2>], Type))
+{
+	foreach (y, line : *self)
+	{
+		foreach (x, val : line)
+		{
+			@body({ x, y }, val);
+		}
+	}
+}
+
+macro void Slice2d.@each_ref(&self; @body(usz[<2>], Type*))
+{
+	foreach (y, line : *self)
+	{
+		foreach (x, &val : line)
+		{
+			@body({ x, y }, val);
+		}
+	}
+}
+
+<*
+ @require idy >= 0 && idy < self.ylen
+*>
+macro Type[] Slice2d.get_row(self, usz idy) @operator([])
+{
+	return (self.ptr + self.inner_len * (idy + self.ystart))[self.xstart:self.xlen];
+}
+
+macro Type Slice2d.get_coord(self, usz[<2>] coord)
+{
+	return *self.get_coord_ref(coord);
+}
+
+macro Type Slice2d.get_xy(self, x, y)
+{
+	return *self.get_xy_ref(x, y);
+}
+
+macro Type* Slice2d.get_xy_ref(self, x, y)
+{
+	return self.ptr + self.inner_len * (y + self.ystart) + self.xstart + x;
+}
+
+macro Type* Slice2d.get_coord_ref(self, usz[<2>] coord)
+{
+	return self.get_xy_ref(coord.x, coord.y);
+}
+
+macro void Slice2d.set_coord(self, usz[<2>] coord, Type value)
+{
+	*self.get_coord_ref(coord) = value;
+}
+
+macro void Slice2d.set_xy(self, x, y, Type value)
+{
+	*self.get_xy_ref(x, y) = value;
+}
+
+<*
+ @require y >= 0 && y < self.ylen
+ @require x >= 0 && x < self.xlen
+*>
+fn Slice2d Slice2d.slice(&self, isz x = 0, isz xlen = 0, isz y = 0, isz ylen = 0)
+{
+	if (xlen < 1) xlen = self.xlen + xlen;
+	if (ylen < 1) ylen = self.ylen + ylen;
+	return { self.ptr, self.inner_len, y + self.ystart, ylen, x + self.xstart, xlen };
+}

lib/std/core/bitorder.c3

@@ -0,0 +1,180 @@
+// Copyright (c) 2023 Christoffer Lerno and contributors. All rights reserved.
+// Use of this source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+module std::core::bitorder;
+
+// This module contains types of different endianness.
+// *BE types represent big-endian types
+// *LE types represent little-endian types.
+
+bitstruct ShortBE : short @bigendian
+{
+	short val : 0..15;
+}
+
+bitstruct UShortBE : ushort @bigendian
+{
+	ushort val : 0..15;
+}
+
+bitstruct IntBE : int @bigendian
+{
+	int val : 0..31;
+}
+
+bitstruct UIntBE : int @bigendian
+{
+	uint val : 0..31;
+}
+
+bitstruct LongBE : long @bigendian
+{
+	long val : 0..63;
+}
+
+bitstruct ULongBE : ulong @bigendian
+{
+	ulong val : 0..63;
+}
+
+bitstruct Int128BE : int128 @bigendian
+{
+	int128 val : 0..127;
+}
+
+bitstruct UInt128BE : uint128 @bigendian
+{
+	uint128 val : 0..127;
+}
+
+bitstruct ShortLE : short @littleendian
+{
+	short val : 0..15;
+}
+
+bitstruct UShortLE : ushort @littleendian
+{
+	ushort val : 0..15;
+}
+
+bitstruct IntLE : int @littleendian
+{
+	int val : 0..31;
+}
+
+bitstruct UIntLE : int @littleendian
+{
+	uint val : 0..31;
+}
+
+bitstruct LongLE : long @littleendian
+{
+	long val : 0..63;
+}
+
+bitstruct ULongLE : ulong @littleendian
+{
+	ulong val : 0..63;
+}
+
+bitstruct Int128LE : int128 @littleendian
+{
+	int128 val : 0..127;
+}
+
+bitstruct UInt128LE : uint128 @littleendian
+{
+	uint128 val : 0..127;
+}
+
+<*
+ @require is_array_or_slice_of_char(bytes) "argument must be an array, a pointer to an array or a slice of char"
+ @require is_bitorder($Type) "type must be a bitorder integer"
+*>
+macro read(bytes, $Type)
+{
+	char[] s;
+	$switch (@typekind(bytes))
+	$case POINTER:
+		s = (*bytes)[:$Type.sizeof];
+	$default:
+		s = bytes[:$Type.sizeof];
+	$endswitch
+	return bitcast(*(char[$Type.sizeof]*)s.ptr, $Type).val;
+}
+
+<*
+ @require is_arrayptr_or_slice_of_char(bytes) "argument must be a pointer to an array or a slice of char"
+ @require is_bitorder($Type) "type must be a bitorder integer"
+*>
+macro write(x, bytes, $Type)
+{
+	char[] s;
+	$switch (@typekind(bytes))
+	$case POINTER:
+		s = (*bytes)[:$Type.sizeof];
+	$default:
+		s = bytes[:$Type.sizeof];
+	$endswitch
+	*($typeof(x)*)s.ptr = bitcast(x, $Type).val;
+}
+
+macro is_bitorder($Type)
+{
+	$switch ($Type)
+	$case UShortLE:
+	$case ShortLE:
+	$case UIntLE:
+	$case IntLE:
+	$case ULongLE:
+	$case LongLE:
+	$case UInt128LE:
+	$case Int128LE:
+	$case UShortBE:
+	$case ShortBE:
+	$case UIntBE:
+	$case IntBE:
+	$case ULongBE:
+	$case LongBE:
+	$case UInt128BE:
+	$case Int128BE:
+		return true;
+	$default:
+		return false;
+	$endswitch
+}
+
+macro bool is_array_or_slice_of_char(bytes)
+{
+	$switch (@typekind(bytes))
+	$case POINTER:
+		var $Inner = $typefrom($typeof(bytes).inner);
+		$if $Inner.kindof == ARRAY:
+			var $Inner2 = $typefrom($Inner.inner);
+			return $Inner2.typeid == char.typeid;
+		$endif
+	$case ARRAY:
+	$case SLICE:
+		var $Inner = $typefrom($typeof(bytes).inner);
+		return $Inner.typeid == char.typeid;
+	$default:
+		return false;
+	$endswitch
+}
+
+macro bool is_arrayptr_or_slice_of_char(bytes)
+{
+	$switch (@typekind(bytes))
+	$case POINTER:
+		var $Inner = $typefrom($typeof(bytes).inner);
+		$if $Inner.kindof == ARRAY:
+			var $Inner2 = $typefrom($Inner.inner);
+			return $Inner2.typeid == char.typeid;
+		$endif
+	$case SLICE:
+		var $Inner = $typefrom($typeof(bytes).inner);
+		return $Inner.typeid == char.typeid;
+	$default:
+		return false;
+	$endswitch
+}

lib/std/core/builtin.c3

@@ -0,0 +1,758 @@
+// Copyright (c) 2021-2024 Christoffer Lerno and contributors. All rights reserved.
+// Use of this source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+module std::core::builtin;
+import libc, std::hash, std::io, std::os::backtrace;
+
+/*
+ Use `IteratorResult` when reading the end of an iterator, or accessing a result out of bounds.
+*/
+fault IteratorResult { NO_MORE_ELEMENT }
+
+/*
+ Use `SearchResult` when trying to return a value from some collection but the element is missing.
+*/
+fault SearchResult { MISSING }
+
+/*
+ Use `CastResult` when an attempt at conversion fails.
+*/
+fault CastResult { TYPE_MISMATCH }
+
+
+def VoidFn = fn void();
+
+<*
+ Stores a variable on the stack, then restores it at the end of the
+ macro scope.
+
+ @param #variable `the variable to store and restore`
+ @require values::@is_lvalue(#variable)
+*>
+macro void @scope(#variable; @body) @builtin
+{
+	var temp = #variable;
+	defer #variable = temp;
+	@body();
+}
+
+<*
+ Swap two variables
+ @require $defined(#a = #b, #b = #a) `The values must be mutually assignable`
+*>
+macro void @swap(#a, #b) @builtin
+{
+	var temp = #a;
+	#a = #b;
+	#b = temp;
+}
+
+<*
+ Convert an `any` type to a type, returning an failure if there is a type mismatch.
+
+ @param v `the any to convert to the given type.`
+ @param $Type `the type to convert to`
+ @return `The any.ptr converted to its type.`
+ @ensure @typeis(return, $Type*)
+ @return! CastResult.TYPE_MISMATCH
+*>
+macro anycast(any v, $Type) @builtin
+{
+	if (v.type != $Type.typeid) return CastResult.TYPE_MISMATCH?;
+	return ($Type*)v.ptr;
+}
+
+fn bool print_backtrace(String message, int backtraces_to_ignore) @if(env::NATIVE_STACKTRACE) => @pool()
+{
+	void*[256] buffer;
+	void*[] backtraces = backtrace::capture_current(&buffer);
+	backtraces_to_ignore++;
+	BacktraceList! backtrace = backtrace::symbolize_backtrace(backtraces, allocator::temp());
+	if (catch backtrace) return false;
+	if (backtrace.len() <= backtraces_to_ignore) return false;
+	io::eprint("\nERROR: '");
+	io::eprint(message);
+	io::eprintn("'");
+	foreach (i, &trace : backtrace)
+	{
+		if (i < backtraces_to_ignore) continue;
+		String inline_suffix = trace.is_inline ? " [inline]" : "";
+		if (trace.is_unknown())
+		{
+			io::eprintfn("  in ???%s", inline_suffix);
+			continue;
+		}
+		if (trace.has_file())
+		{
+			io::eprintfn("  in %s (%s:%d) [%s]%s", trace.function, trace.file, trace.line, trace.object_file, inline_suffix);
+			continue;
+		}
+		io::eprintfn("  in %s (source unavailable) [%s]%s", trace.function, trace.object_file, inline_suffix);
+	}
+	return true;
+
+}
+
+fn void default_panic(String message, String file, String function, uint line) @if(env::NATIVE_STACKTRACE)
+{
+	$if $defined(io::stderr):
+	if (!print_backtrace(message, 2))
+	{
+		io::eprintfn("\nERROR: '%s', in %s (%s:%d)", message, function, file, line);
+	}
+	$endif
+	$$trap();
+
+}
+
+macro void abort(String string = "Unrecoverable error reached", ...) @builtin @noreturn
+{
+	panicf(string, $$FILE, $$FUNC, $$LINE, $vasplat);
+	$$trap();
+}
+
+bool in_panic @local = false;
+
+fn void default_panic(String message, String file, String function, uint line) @if(!env::NATIVE_STACKTRACE)
+{
+	if (in_panic)
+	{
+		io::eprintn("Panic inside of panic.");
+		return;
+	}
+	in_panic = true;
+	$if $defined(io::stderr):
+	io::eprint("\nERROR: '");
+	io::eprint(message);
+	io::eprintfn("', in %s (%s:%d)", function, file, line);
+	$endif
+	in_panic = false;
+	$$trap();
+}
+
+def PanicFn = fn void(String message, String file, String function, uint line);
+
+PanicFn panic = &default_panic;
+
+fn void panicf(String fmt, String file, String function, uint line, args...)
+{
+	if (in_panic)
+	{
+		io::eprint("Panic inside of panic: ");
+		io::eprintn(fmt);
+		return;
+	}
+	in_panic = true;
+	@stack_mem(512; Allocator allocator)
+	{
+		DString s;
+		s.init(allocator);
+		s.appendf(fmt, ...args);
+		in_panic = false;
+		panic(s.str_view(), file, function, line);
+	};
+}
+
+<*
+ Marks the path as unreachable. This will panic in safe mode, and in fast will simply be assumed
+ never happens.
+ @param [in] string "The panic message or format string"
+*>
+macro void unreachable(String string = "Unreachable statement reached.", ...) @builtin @noreturn
+{
+	$if env::COMPILER_SAFE_MODE:
+	panicf(string, $$FILE, $$FUNC, $$LINE, $vasplat);
+	$endif;
+	$$unreachable();
+}
+
+<*
+ Marks the path as unsupported, this is similar to unreachable.
+ @param [in] string "The error message"
+*>
+macro void unsupported(String string = "Unsupported function invoked") @builtin @noreturn
+{
+	panicf(string, $$FILE, $$FUNC, $$LINE, $vasplat);
+	$$unreachable();
+}
+
+<*
+ Unconditionally break into an attached debugger when reached.
+*>
+macro void breakpoint() @builtin
+{
+	$$breakpoint();
+}
+
+macro any_make(void* ptr, typeid type) @builtin
+{
+	return $$any_make(ptr, type);
+}
+
+macro any.retype_to(&self, typeid type)
+{
+	return $$any_make(self.ptr, type);
+}
+
+macro any.as_inner(&self)
+{
+	return $$any_make(self.ptr, self.type.inner);
+}
+
+<*
+ @param expr "the expression to cast"
+ @param $Type "the type to cast to"
+
+ @require $sizeof(expr) == $Type.sizeof  "Cannot bitcast between types of different size."
+ @ensure @typeis(return, $Type)
+*>
+macro bitcast(expr, $Type) @builtin
+{
+	$if $Type.alignof <= $alignof(expr):
+		return *($Type*)&expr;
+	$else
+		$Type x @noinit;
+		$$memcpy(&x, &expr, $sizeof(expr), false, $Type.alignof, $alignof(expr));
+		return x;
+	$endif
+}
+
+<*
+ @param $Type `The type of the enum`
+ @param [in] enum_name `The name of the enum to search for`
+ @require $Type.kindof == ENUM `Only enums may be used`
+ @ensure @typeis(return, $Type)
+ @return! SearchResult.MISSING
+*>
+macro enum_by_name($Type, String enum_name) @builtin
+{
+	typeid x = $Type.typeid;
+	foreach (i, name : x.names)
+	{
+		if (name == enum_name) return $Type.from_ordinal(i);
+	}
+	return SearchResult.MISSING?;
+}
+
+<*
+ @param $Type `The type of the enum`
+ @require $Type.kindof == ENUM `Only enums may be used`
+ @require $defined($Type.#value) `Expected '#value' to match an enum associated value`
+ @require $assignable(value, $typeof(($Type){}.#value)) `Expected the value to match the type of the associated value`
+ @ensure @typeis(return, $Type)
+ @return! SearchResult.MISSING
+*>
+macro @enum_from_value($Type, #value, value) @builtin
+{
+	usz elements = $Type.elements;
+	foreach (e : $Type.values)
+	{
+		if (e.#value == value) return e;
+	}
+	return SearchResult.MISSING?;
+}
+
+<*
+ Mark an expression as likely to be true
+
+ @param #value "expression to be marked likely"
+ @param $probability "in the range 0 - 1"
+ @require $probability >= 0 && $probability <= 1.0
+*>
+macro bool @likely(bool #value, $probability = 1.0) @builtin
+{
+	$switch
+	$case env::BUILTIN_EXPECT_IS_DISABLED:
+	return #value;
+	$case $probability == 1.0:
+	return $$expect(#value, true);
+	$default:
+	return $$expect_with_probability(#value, true, $probability);
+	$endswitch
+}
+
+<*
+ Mark an expression as unlikely to be true
+
+ @param #value "expression to be marked unlikely"
+ @param $probability "in the range 0 - 1"
+ @require $probability >= 0 && $probability <= 1.0
+*>
+macro bool @unlikely(bool #value, $probability = 1.0) @builtin
+{
+	$switch
+	$case env::BUILTIN_EXPECT_IS_DISABLED:
+	return #value;
+	$case $probability == 1.0:
+	return $$expect(#value, false);
+	$default:
+	return $$expect_with_probability(#value, false, $probability);
+	$endswitch
+}
+
+<*
+ @require values::@is_int(#value) || values::@is_bool(#value)
+ @require $assignable(expected, $typeof(#value))
+ @require $probability >= 0 && $probability <= 1.0
+*>
+macro @expect(#value, expected, $probability = 1.0) @builtin
+{
+	$switch
+	$case env::BUILTIN_EXPECT_IS_DISABLED:
+	return #value == expected;
+	$case $probability == 1.0:
+	return $$expect(#value, ($typeof(#value))expected);
+	$default:
+	return $$expect_with_probability(#value, expected, $probability);
+	$endswitch
+}
+
+<*
+ Locality for prefetch, levels 0 - 3, corresponding
+ to "extremely local" to "no locality"
+*>
+enum PrefetchLocality
+{
+	NO_LOCALITY,
+	FAR,
+	NEAR,
+	VERY_NEAR,
+}
+
+<*
+ Prefetch a pointer.
+
+ @param [in] ptr `Pointer to prefetch`
+ @param $locality `Locality ranging from none to extremely local`
+ @param $write `Prefetch for write, otherwise prefetch for read.`
+*>
+macro @prefetch(void* ptr, PrefetchLocality $locality = VERY_NEAR, bool $write = false) @builtin
+{
+	$if !env::BUILTIN_PREFETCH_IS_DISABLED:
+		$$prefetch(ptr, $write ? 1 : 0, $locality.ordinal);
+	$endif
+}
+
+macro swizzle(v, ...) @builtin
+{
+	return $$swizzle(v, $vasplat);
+}
+
+macro swizzle2(v, v2, ...) @builtin
+{
+	return $$swizzle2(v, v2, $vasplat);
+}
+<*
+ Return the excuse in the Optional if it is Empty, otherwise
+ return a null fault.
+
+ @require @typekind(#expr) == OPTIONAL : `@catch expects an Optional value`
+*>
+macro anyfault @catch(#expr) @builtin
+{
+	if (catch f = #expr) return f;
+	return {};
+}
+
+<*
+ Check if an Optional expression holds a value or is empty, returning true
+ if it has a value.
+
+ @require @typekind(#expr) == OPTIONAL : `@ok expects an Optional value`
+*>
+macro bool @ok(#expr) @builtin
+{
+	if (catch #expr) return false;
+	return true;
+}
+
+<*
+ @require $defined(&#value, (char*)&#value) "This must be a value that can be viewed as a char array"
+ *>
+macro char[] @as_char_view(#value) @builtin
+{
+	return ((char*)&#value)[:$sizeof(#value)];
+}
+
+macro isz @str_find(String $string, String $needle) @builtin => $$str_find($string, $needle);
+macro String @str_upper(String $str) @builtin => $$str_upper($str);
+macro String @str_lower(String $str) @builtin => $$str_lower($str);
+macro uint @str_hash(String $str) @builtin => $$str_hash($str);
+
+macro @generic_hash_core(h, value)
+{
+	h ^= (uint)value; // insert lowest 32 bits
+	h *= 0x96f59e5b; // diffuse them up
+	h ^= h >> 16; // diffuse them down
+	return h;
+}
+
+macro @generic_hash(value)
+{
+	uint h = @generic_hash_core((uint)0x3efd4391, value);
+	$for (var $cnt = 4; $cnt < $sizeof(value); $cnt += 4)
+		value >>= 32; // reduce value
+		h = @generic_hash_core(h, value);
+	$endfor
+	return h;
+}
+
+macro uint int.hash(int i) => @generic_hash(i);
+macro uint uint.hash(uint i) => @generic_hash(i);
+macro uint short.hash(short s) => @generic_hash(s);
+macro uint ushort.hash(ushort s) => @generic_hash(s);
+macro uint char.hash(char c) => @generic_hash(c);
+macro uint ichar.hash(ichar c) => @generic_hash(c);
+macro uint long.hash(long i) => @generic_hash(i);
+macro uint ulong.hash(ulong i) => @generic_hash(i);
+macro uint int128.hash(int128 i) => @generic_hash(i);
+macro uint uint128.hash(uint128 i) => @generic_hash(i);
+macro uint bool.hash(bool b) => @generic_hash(b);
+macro uint typeid.hash(typeid t) => @generic_hash(((ulong)(uptr)t));
+macro uint String.hash(String c) => (uint)fnv32a::encode(c);
+macro uint char[].hash(char[] c) => (uint)fnv32a::encode(c);
+macro uint void*.hash(void* ptr) => @generic_hash(((ulong)(uptr)ptr));
+
+distinct EmptySlot = void*;
+const EmptySlot EMPTY_MACRO_SLOT @builtin = null;
+macro @is_empty_macro_slot(#arg) @builtin => @typeis(#arg, EmptySlot);
+macro @is_valid_macro_slot(#arg) @builtin => !@typeis(#arg, EmptySlot);
+
+const MAX_FRAMEADDRESS = 128;
+<*
+ @require n >= 0
+*>
+macro void* get_frameaddress(int n)
+{
+	if (n > MAX_FRAMEADDRESS) return null;
+	switch (n)
+	{
+		case 0: return $$frameaddress(0);
+		case 1: return $$frameaddress(1);
+		case 2: return $$frameaddress(2);
+		case 3: return $$frameaddress(3);
+		case 4: return $$frameaddress(4);
+		case 5: return $$frameaddress(5);
+		case 6: return $$frameaddress(6);
+		case 7: return $$frameaddress(7);
+		case 8: return $$frameaddress(8);
+		case 9: return $$frameaddress(9);
+		case 10: return $$frameaddress(10);
+		case 11: return $$frameaddress(11);
+		case 12: return $$frameaddress(12);
+		case 13: return $$frameaddress(13);
+		case 14: return $$frameaddress(14);
+		case 15: return $$frameaddress(15);
+		case 16: return $$frameaddress(16);
+		case 17: return $$frameaddress(17);
+		case 18: return $$frameaddress(18);
+		case 19: return $$frameaddress(19);
+		case 20: return $$frameaddress(20);
+		case 21: return $$frameaddress(21);
+		case 22: return $$frameaddress(22);
+		case 23: return $$frameaddress(23);
+		case 24: return $$frameaddress(24);
+		case 25: return $$frameaddress(25);
+		case 26: return $$frameaddress(26);
+		case 27: return $$frameaddress(27);
+		case 28: return $$frameaddress(28);
+		case 29: return $$frameaddress(29);
+		case 30: return $$frameaddress(30);
+		case 31: return $$frameaddress(31);
+		case 32: return $$frameaddress(32);
+		case 33: return $$frameaddress(33);
+		case 34: return $$frameaddress(34);
+		case 35: return $$frameaddress(35);
+		case 36: return $$frameaddress(36);
+		case 37: return $$frameaddress(37);
+		case 38: return $$frameaddress(38);
+		case 39: return $$frameaddress(39);
+		case 40: return $$frameaddress(40);
+		case 41: return $$frameaddress(41);
+		case 42: return $$frameaddress(42);
+		case 43: return $$frameaddress(43);
+		case 44: return $$frameaddress(44);
+		case 45: return $$frameaddress(45);
+		case 46: return $$frameaddress(46);
+		case 47: return $$frameaddress(47);
+		case 48: return $$frameaddress(48);
+		case 49: return $$frameaddress(49);
+		case 50: return $$frameaddress(50);
+		case 51: return $$frameaddress(51);
+		case 52: return $$frameaddress(52);
+		case 53: return $$frameaddress(53);
+		case 54: return $$frameaddress(54);
+		case 55: return $$frameaddress(55);
+		case 56: return $$frameaddress(56);
+		case 57: return $$frameaddress(57);
+		case 58: return $$frameaddress(58);
+		case 59: return $$frameaddress(59);
+		case 60: return $$frameaddress(60);
+		case 61: return $$frameaddress(61);
+		case 62: return $$frameaddress(62);
+		case 63: return $$frameaddress(63);
+		case 64: return $$frameaddress(64);
+		case 65: return $$frameaddress(65);
+		case 66: return $$frameaddress(66);
+		case 67: return $$frameaddress(67);
+		case 68: return $$frameaddress(68);
+		case 69: return $$frameaddress(69);
+		case 70: return $$frameaddress(70);
+		case 71: return $$frameaddress(71);
+		case 72: return $$frameaddress(72);
+		case 73: return $$frameaddress(73);
+		case 74: return $$frameaddress(74);
+		case 75: return $$frameaddress(75);
+		case 76: return $$frameaddress(76);
+		case 77: return $$frameaddress(77);
+		case 78: return $$frameaddress(78);
+		case 79: return $$frameaddress(79);
+		case 80: return $$frameaddress(80);
+		case 81: return $$frameaddress(81);
+		case 82: return $$frameaddress(82);
+		case 83: return $$frameaddress(83);
+		case 84: return $$frameaddress(84);
+		case 85: return $$frameaddress(85);
+		case 86: return $$frameaddress(86);
+		case 87: return $$frameaddress(87);
+		case 88: return $$frameaddress(88);
+		case 89: return $$frameaddress(89);
+		case 90: return $$frameaddress(90);
+		case 91: return $$frameaddress(91);
+		case 92: return $$frameaddress(92);
+		case 93: return $$frameaddress(93);
+		case 94: return $$frameaddress(94);
+		case 95: return $$frameaddress(95);
+		case 96: return $$frameaddress(96);
+		case 97: return $$frameaddress(97);
+		case 98: return $$frameaddress(98);
+		case 99: return $$frameaddress(99);
+		case 100: return $$frameaddress(100);
+		case 101: return $$frameaddress(101);
+		case 102: return $$frameaddress(102);
+		case 103: return $$frameaddress(103);
+		case 104: return $$frameaddress(104);
+		case 105: return $$frameaddress(105);
+		case 106: return $$frameaddress(106);
+		case 107: return $$frameaddress(107);
+		case 108: return $$frameaddress(108);
+		case 109: return $$frameaddress(109);
+		case 110: return $$frameaddress(110);
+		case 111: return $$frameaddress(111);
+		case 112: return $$frameaddress(112);
+		case 113: return $$frameaddress(113);
+		case 114: return $$frameaddress(114);
+		case 115: return $$frameaddress(115);
+		case 116: return $$frameaddress(116);
+		case 117: return $$frameaddress(117);
+		case 118: return $$frameaddress(118);
+		case 119: return $$frameaddress(119);
+		case 120: return $$frameaddress(120);
+		case 121: return $$frameaddress(121);
+		case 122: return $$frameaddress(122);
+		case 123: return $$frameaddress(123);
+		case 124: return $$frameaddress(124);
+		case 125: return $$frameaddress(125);
+		case 126: return $$frameaddress(126);
+		case 127: return $$frameaddress(127);
+		case 128: return $$frameaddress(128);
+		default: unreachable();
+	}
+}
+
+<*
+ @require n >= 0
+*>
+macro void* get_returnaddress(int n)
+{
+	if (n > MAX_FRAMEADDRESS) return null;
+	switch (n)
+	{
+		case 0: return $$returnaddress(0);
+		case 1: return $$returnaddress(1);
+		case 2: return $$returnaddress(2);
+		case 3: return $$returnaddress(3);
+		case 4: return $$returnaddress(4);
+		case 5: return $$returnaddress(5);
+		case 6: return $$returnaddress(6);
+		case 7: return $$returnaddress(7);
+		case 8: return $$returnaddress(8);
+		case 9: return $$returnaddress(9);
+		case 10: return $$returnaddress(10);
+		case 11: return $$returnaddress(11);
+		case 12: return $$returnaddress(12);
+		case 13: return $$returnaddress(13);
+		case 14: return $$returnaddress(14);
+		case 15: return $$returnaddress(15);
+		case 16: return $$returnaddress(16);
+		case 17: return $$returnaddress(17);
+		case 18: return $$returnaddress(18);
+		case 19: return $$returnaddress(19);
+		case 20: return $$returnaddress(20);
+		case 21: return $$returnaddress(21);
+		case 22: return $$returnaddress(22);
+		case 23: return $$returnaddress(23);
+		case 24: return $$returnaddress(24);
+		case 25: return $$returnaddress(25);
+		case 26: return $$returnaddress(26);
+		case 27: return $$returnaddress(27);
+		case 28: return $$returnaddress(28);
+		case 29: return $$returnaddress(29);
+		case 30: return $$returnaddress(30);
+		case 31: return $$returnaddress(31);
+		case 32: return $$returnaddress(32);
+		case 33: return $$returnaddress(33);
+		case 34: return $$returnaddress(34);
+		case 35: return $$returnaddress(35);
+		case 36: return $$returnaddress(36);
+		case 37: return $$returnaddress(37);
+		case 38: return $$returnaddress(38);
+		case 39: return $$returnaddress(39);
+		case 40: return $$returnaddress(40);
+		case 41: return $$returnaddress(41);
+		case 42: return $$returnaddress(42);
+		case 43: return $$returnaddress(43);
+		case 44: return $$returnaddress(44);
+		case 45: return $$returnaddress(45);
+		case 46: return $$returnaddress(46);
+		case 47: return $$returnaddress(47);
+		case 48: return $$returnaddress(48);
+		case 49: return $$returnaddress(49);
+		case 50: return $$returnaddress(50);
+		case 51: return $$returnaddress(51);
+		case 52: return $$returnaddress(52);
+		case 53: return $$returnaddress(53);
+		case 54: return $$returnaddress(54);
+		case 55: return $$returnaddress(55);
+		case 56: return $$returnaddress(56);
+		case 57: return $$returnaddress(57);
+		case 58: return $$returnaddress(58);
+		case 59: return $$returnaddress(59);
+		case 60: return $$returnaddress(60);
+		case 61: return $$returnaddress(61);
+		case 62: return $$returnaddress(62);
+		case 63: return $$returnaddress(63);
+		case 64: return $$returnaddress(64);
+		case 65: return $$returnaddress(65);
+		case 66: return $$returnaddress(66);
+		case 67: return $$returnaddress(67);
+		case 68: return $$returnaddress(68);
+		case 69: return $$returnaddress(69);
+		case 70: return $$returnaddress(70);
+		case 71: return $$returnaddress(71);
+		case 72: return $$returnaddress(72);
+		case 73: return $$returnaddress(73);
+		case 74: return $$returnaddress(74);
+		case 75: return $$returnaddress(75);
+		case 76: return $$returnaddress(76);
+		case 77: return $$returnaddress(77);
+		case 78: return $$returnaddress(78);
+		case 79: return $$returnaddress(79);
+		case 80: return $$returnaddress(80);
+		case 81: return $$returnaddress(81);
+		case 82: return $$returnaddress(82);
+		case 83: return $$returnaddress(83);
+		case 84: return $$returnaddress(84);
+		case 85: return $$returnaddress(85);
+		case 86: return $$returnaddress(86);
+		case 87: return $$returnaddress(87);
+		case 88: return $$returnaddress(88);
+		case 89: return $$returnaddress(89);
+		case 90: return $$returnaddress(90);
+		case 91: return $$returnaddress(91);
+		case 92: return $$returnaddress(92);
+		case 93: return $$returnaddress(93);
+		case 94: return $$returnaddress(94);
+		case 95: return $$returnaddress(95);
+		case 96: return $$returnaddress(96);
+		case 97: return $$returnaddress(97);
+		case 98: return $$returnaddress(98);
+		case 99: return $$returnaddress(99);
+		case 100: return $$returnaddress(100);
+		case 101: return $$returnaddress(101);
+		case 102: return $$returnaddress(102);
+		case 103: return $$returnaddress(103);
+		case 104: return $$returnaddress(104);
+		case 105: return $$returnaddress(105);
+		case 106: return $$returnaddress(106);
+		case 107: return $$returnaddress(107);
+		case 108: return $$returnaddress(108);
+		case 109: return $$returnaddress(109);
+		case 110: return $$returnaddress(110);
+		case 111: return $$returnaddress(111);
+		case 112: return $$returnaddress(112);
+		case 113: return $$returnaddress(113);
+		case 114: return $$returnaddress(114);
+		case 115: return $$returnaddress(115);
+		case 116: return $$returnaddress(116);
+		case 117: return $$returnaddress(117);
+		case 118: return $$returnaddress(118);
+		case 119: return $$returnaddress(119);
+		case 120: return $$returnaddress(120);
+		case 121: return $$returnaddress(121);
+		case 122: return $$returnaddress(122);
+		case 123: return $$returnaddress(123);
+		case 124: return $$returnaddress(124);
+		case 125: return $$returnaddress(125);
+		case 126: return $$returnaddress(126);
+		case 127: return $$returnaddress(127);
+		case 128: return $$returnaddress(128);
+		default: unreachable();
+	}
+}
+
+module std::core::builtin @if((env::LINUX || env::DARWIN) && env::COMPILER_SAFE_MODE && env::DEBUG_SYMBOLS);
+import libc, std::io;
+
+fn void sig_panic(String message)
+{
+	default_panic(message, "???", "???", 0);
+}
+
+SignalFunction old_bus_error;
+SignalFunction old_segmentation_fault;
+
+fn void sig_bus_error(CInt i)
+{
+	$if !env::NATIVE_STACKTRACE:
+		sig_panic("Illegal memory access.");
+	$else
+		$if $defined(io::stderr):
+			if (!print_backtrace("Illegal memory access.", 1))
+			{
+				io::eprintn("\nERROR: 'Illegal memory access'.");
+			}
+		$endif
+	$endif
+	$$trap();
+}
+
+fn void sig_segmentation_fault(CInt i)
+{
+	$if !env::NATIVE_STACKTRACE:
+		sig_panic("Out of bounds memory access.");
+	$else
+		$if $defined(io::stderr):
+			if (!print_backtrace("Out of bounds memory access.", 1))
+			{
+				io::eprintn("\nERROR: Memory error without backtrace, possible stack overflow.");
+			}
+		$endif
+	$endif
+	$$trap();
+}
+
+fn void install_signal_handler(CInt signal, SignalFunction func) @local
+{
+	SignalFunction old = libc::signal(signal, func);
+	// Restore
+	if ((iptr)old > 1024) libc::signal(signal, old);
+}
+
+// Clean this up
+fn void install_signal_handlers() @init(101) @local @if(env::BACKTRACE)
+{
+	install_signal_handler(libc::SIGBUS, &sig_bus_error);
+	install_signal_handler(libc::SIGSEGV, &sig_segmentation_fault);
+}
+

lib/std/core/builtin_comparison.c3

@@ -0,0 +1,128 @@
+// Copyright (c) 2021-2024 Christoffer Lerno and contributors. All rights reserved.
+// Use of this source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+module std::core::builtin;
+
+<*
+ @require types::@comparable_value(a) && types::@comparable_value(b)
+*>
+macro less(a, b) @builtin
+{
+	$switch
+	$case $defined(a.less):
+		return a.less(b);
+	$case $defined(a.compare_to):
+		return a.compare_to(b) < 0;
+	$default:
+		return a < b;
+	$endswitch
+}
+
+<*
+ @require types::@comparable_value(a) && types::@comparable_value(b)
+*>
+macro less_eq(a, b) @builtin
+{
+	$switch
+	$case $defined(a.less):
+		return !b.less(a);
+	$case $defined(a.compare_to):
+		return a.compare_to(b) <= 0;
+	$default:
+		return a <= b;
+	$endswitch
+}
+
+<*
+ @require types::@comparable_value(a) && types::@comparable_value(b)
+*>
+macro greater(a, b) @builtin
+{
+	$switch
+	$case $defined(a.less):
+		return b.less(a);
+	$case $defined(a.compare_to):
+		return a.compare_to(b) > 0;
+	$default:
+		return a > b;
+	$endswitch
+}
+
+<*
+ @require types::@comparable_value(a) && types::@comparable_value(b)
+*>
+macro int compare_to(a, b) @builtin
+{
+	$switch
+	$case $defined(a.compare_to):
+		return a.compare_to(b);
+	$case $defined(a.less):
+		return (int)b.less(a) - (int)a.less(b);
+	$default:
+		return (int)(a > b) - (int)(a < b);
+	$endswitch
+}
+<*
+ @require types::@comparable_value(a) && types::@comparable_value(b)
+*>
+macro greater_eq(a, b) @builtin
+{
+	$switch
+	$case $defined(a.less):
+		return !a.less(b);
+	$case $defined(a.compare_to):
+		return a.compare_to(b) >= 0;
+	$default:
+		return a >= b;
+	$endswitch
+}
+
+<*
+ @require types::@equatable_value(a) && types::@equatable_value(b) `values must be equatable`
+*>
+macro bool equals(a, b) @builtin
+{
+	$switch
+	$case $defined(a.equals, a.equals(b)):
+		return a.equals(b);
+	$case $defined(a.compare_to, a.compare_to(b)):
+		return a.compare_to(b) == 0;
+	$case $defined(a.less):
+		return !a.less(b) && !b.less(a);
+	$default:
+		return a == b;
+	$endswitch
+}
+
+macro min(x, ...) @builtin
+{
+	$if $vacount == 1:
+		return less(x, $vaarg[0]) ? x : $vaarg[0];
+	$else
+		var result = x;
+		$for (var $i = 0; $i < $vacount; $i++)
+			if (less($vaarg[$i], result))
+			{
+				 result = $vaarg[$i];
+			}
+		$endfor
+		return result;
+	$endif
+}
+
+macro max(x, ...) @builtin
+{
+	$if $vacount == 1:
+		return greater(x, $vaarg[0]) ? x : $vaarg[0];
+	$else
+		var result = x;
+		$for (var $i = 0; $i < $vacount; $i++)
+			if (greater($vaarg[$i], result))
+			{
+				 result = $vaarg[$i];
+			}
+		$endfor
+		return result;
+	$endif
+}
+

lib/std/core/cinterop.c3

@@ -0,0 +1,61 @@
+// Copyright (c) 2021-2024 Christoffer Lerno. All rights reserved.
+// Use of this source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+module std::core::cinterop;
+
+const C_INT_SIZE = $$C_INT_SIZE;
+const C_LONG_SIZE = $$C_LONG_SIZE;
+const C_SHORT_SIZE = $$C_SHORT_SIZE;
+const C_LONG_LONG_SIZE = $$C_LONG_LONG_SIZE;
+
+$assert C_SHORT_SIZE < 32;
+$assert C_INT_SIZE < 128;
+$assert C_LONG_SIZE < 128;
+$assert C_LONG_LONG_SIZE <= 128;
+$assert C_SHORT_SIZE <= C_INT_SIZE;
+$assert C_INT_SIZE <= C_LONG_SIZE;
+$assert C_LONG_SIZE <= C_LONG_LONG_SIZE;
+
+def CShort = $typefrom(signed_int_from_bitsize($$C_SHORT_SIZE));
+def CUShort = $typefrom(unsigned_int_from_bitsize($$C_SHORT_SIZE));
+def CInt = $typefrom(signed_int_from_bitsize($$C_INT_SIZE));
+def CUInt = $typefrom(unsigned_int_from_bitsize($$C_INT_SIZE));
+def CLong = $typefrom(signed_int_from_bitsize($$C_LONG_SIZE));
+def CULong = $typefrom(unsigned_int_from_bitsize($$C_LONG_SIZE));
+def CLongLong = $typefrom(signed_int_from_bitsize($$C_LONG_LONG_SIZE));
+def CULongLong = $typefrom(unsigned_int_from_bitsize($$C_LONG_LONG_SIZE));
+def CSChar = ichar;
+def CUChar = char;
+
+def CChar = $typefrom($$C_CHAR_IS_SIGNED ? ichar.typeid : char.typeid);
+
+enum CBool : char
+{
+	FALSE,
+	TRUE
+}
+
+// Helper macros
+macro typeid signed_int_from_bitsize(usz $bitsize) @private
+{
+	$switch ($bitsize)
+		$case 128: return int128.typeid;
+		$case 64: return long.typeid;
+		$case 32: return int.typeid;
+		$case 16: return short.typeid;
+		$case 8: return ichar.typeid;
+		$default: $error("Invalid bitsize");
+	$endswitch
+}
+
+macro typeid unsigned_int_from_bitsize(usz $bitsize) @private
+{
+	$switch ($bitsize)
+		$case 128: return uint128.typeid;
+		$case 64: return ulong.typeid;
+		$case 32: return uint.typeid;
+		$case 16: return ushort.typeid;
+		$case 8: return char.typeid;
+		$default: $error("Invalid bitsize");
+	$endswitch
+}

lib/std/core/conv.c3

@@ -0,0 +1,416 @@
+module std::core::string::conv;
+
+const uint UTF16_SURROGATE_OFFSET @private = 0x10000;
+const uint UTF16_SURROGATE_GENERIC_MASK @private = 0xF800;
+const uint UTF16_SURROGATE_GENERIC_VALUE @private = 0xD800;
+const uint UTF16_SURROGATE_MASK @private = 0xFC00;
+const uint UTF16_SURROGATE_CODEPOINT_MASK @private = 0x03FF;
+const uint UTF16_SURROGATE_BITS @private = 10;
+const uint UTF16_SURROGATE_LOW_VALUE @private = 0xDC00;
+const uint UTF16_SURROGATE_HIGH_VALUE @private = 0xD800;
+
+<*
+ @param c `The utf32 codepoint to convert`
+ @param [out] output `the resulting buffer`
+*>
+fn usz! char32_to_utf8(Char32 c, char[] output)
+{
+	if (!output.len) return UnicodeResult.CONVERSION_FAILED?;
+	switch (true)
+	{
+		case c <= 0x7f:
+			output[0] = (char)c;
+			return 1;
+		case c <= 0x7ff:
+			if (output.len < 2) return UnicodeResult.CONVERSION_FAILED?;
+			output[0] = (char)(0xC0 | c >> 6);
+			output[1] = (char)(0x80 | (c & 0x3F));
+			return 2;
+		case c <= 0xffff:
+			if (output.len < 3) return UnicodeResult.CONVERSION_FAILED?;
+			output[0] = (char)(0xE0 | c >> 12);
+			output[1] = (char)(0x80 | (c >> 6 & 0x3F));
+			output[2] = (char)(0x80 | (c & 0x3F));
+			return 3;
+		case c <= 0x10ffff:
+			if (output.len < 4) return UnicodeResult.CONVERSION_FAILED?;
+			output[0] = (char)(0xF0 | c >> 18);
+			output[1] = (char)(0x80 | (c >> 12 & 0x3F));
+			output[2] = (char)(0x80 | (c >> 6 & 0x3F));
+			output[3] = (char)(0x80 | (c & 0x3F));
+			return 4;
+		default:
+			// 0x10FFFF and above is not defined.
+			return UnicodeResult.CONVERSION_FAILED?;
+	}
+}
+
+<*
+ Convert a code pointer into 1-2 UTF16 characters.
+
+ @param c `The character to convert.`
+ @param [inout] output `the resulting UTF16 buffer to write to.`
+*>
+fn void char32_to_utf16_unsafe(Char32 c, Char16** output)
+{
+	if (c < UTF16_SURROGATE_OFFSET)
+	{
+		(*output)++[0] = (Char16)c;
+		return;
+	}
+	c -= UTF16_SURROGATE_OFFSET;
+	Char16 low = (Char16)(UTF16_SURROGATE_LOW_VALUE | (c & UTF16_SURROGATE_CODEPOINT_MASK));
+	c >>= UTF16_SURROGATE_BITS;
+	Char16 high = (Char16)(UTF16_SURROGATE_HIGH_VALUE | (c & UTF16_SURROGATE_CODEPOINT_MASK));
+	(*output)++[0] = (Char16)high;
+	(*output)++[0] = (Char16)low;
+}
+
+<*
+ Convert 1-2 UTF16 data points into UTF8.
+
+ @param [in] ptr `The UTF16 data to convert.`
+ @param [inout] available `amount of UTF16 data available.`
+ @param [inout] output `the resulting utf8 buffer to write to.`
+*>
+fn void! char16_to_utf8_unsafe(Char16 *ptr, usz *available, char** output)
+{
+	Char16 high = *ptr;
+	if (high & UTF16_SURROGATE_GENERIC_MASK != UTF16_SURROGATE_GENERIC_VALUE)
+	{
+		char32_to_utf8_unsafe(high, output);
+		*available = 1;
+		return;
+	}
+	// Low surrogate first is an error
+	if (high & UTF16_SURROGATE_MASK != UTF16_SURROGATE_HIGH_VALUE) return UnicodeResult.INVALID_UTF16?;
+
+	// Unmatched high surrogate is an error
+	if (*available == 1) return UnicodeResult.INVALID_UTF16?;
+
+	Char16 low = ptr[1];
+
+	// Unmatched high surrogate, invalid
+	if (low & UTF16_SURROGATE_MASK != UTF16_SURROGATE_LOW_VALUE) return UnicodeResult.INVALID_UTF16?;
+
+	// The high bits of the codepoint are the value bits of the high surrogate
+	// The low bits of the codepoint are the value bits of the low surrogate
+	Char32 uc = (high & UTF16_SURROGATE_CODEPOINT_MASK) << UTF16_SURROGATE_BITS
+				| (low & UTF16_SURROGATE_CODEPOINT_MASK) + UTF16_SURROGATE_OFFSET;
+	char32_to_utf8_unsafe(uc, output);
+	*available = 2;
+}
+<*
+ @param c `The utf32 codepoint to convert`
+ @param [inout] output `the resulting buffer`
+*>
+fn usz char32_to_utf8_unsafe(Char32 c, char** output)
+{
+	switch
+	{
+		case c <= 0x7f:
+			(*output)++[0] = (char)c;
+			return 1;
+		case c <= 0x7ff:
+			(*output)++[0] = (char)(0xC0 | c >> 6);
+			(*output)++[0] = (char)(0x80 | (c & 0x3F));
+			return 2;
+		case c <= 0xffff:
+			(*output)++[0] = (char)(0xE0 | c >> 12);
+			(*output)++[0] = (char)(0x80 | (c >> 6 & 0x3F));
+			(*output)++[0] = (char)(0x80 | (c & 0x3F));
+			return 3;
+		default:
+			(*output)++[0] = (char)(0xF0 | c >> 18);
+			(*output)++[0] = (char)(0x80 | (c >> 12 & 0x3F));
+			(*output)++[0] = (char)(0x80 | (c >> 6 & 0x3F));
+			(*output)++[0] = (char)(0x80 | (c & 0x3F));
+			return 4;
+	}
+}
+
+<*
+ @param [in] ptr `pointer to the first character to parse`
+ @param [inout] size `Set to max characters to read, set to characters read`
+ @return `the parsed 32 bit codepoint`
+*>
+fn Char32! utf8_to_char32(char* ptr, usz* size)
+{
+	usz max_size = *size;
+	if (max_size < 1) return UnicodeResult.INVALID_UTF8?;
+	char c = (ptr++)[0];
+
+	if ((c & 0x80) == 0)
+	{
+		*size = 1;
+		return c;
+	}
+	if ((c & 0xE0) == 0xC0)
+	{
+		if (max_size < 2) return UnicodeResult.INVALID_UTF8?;
+		*size = 2;
+		Char32 uc = (c & 0x1F) << 6;
+		c = *ptr;
+		// Overlong sequence or invalid second.
+		if (!uc || c & 0xC0 != 0x80) return UnicodeResult.INVALID_UTF8?;
+		return uc + c & 0x3F;
+	}
+	if ((c & 0xF0) == 0xE0)
+	{
+		if (max_size < 3) return UnicodeResult.INVALID_UTF8?;
+		*size = 3;
+		Char32 uc = (c & 0x0F) << 12;
+		c = ptr++[0];
+		if (c & 0xC0 != 0x80) return UnicodeResult.INVALID_UTF8?;
+		uc += (c & 0x3F) << 6;
+		c = ptr++[0];
+		// Overlong sequence or invalid last
+		if (!uc || c & 0xC0 != 0x80) return UnicodeResult.INVALID_UTF8?;
+		return uc + c & 0x3F;
+	}
+	if (max_size < 4) return UnicodeResult.INVALID_UTF8?;
+	if ((c & 0xF8) != 0xF0) return UnicodeResult.INVALID_UTF8?;
+	*size = 4;
+	Char32 uc = (c & 0x07) << 18;
+	c = ptr++[0];
+	if (c & 0xC0 != 0x80) return UnicodeResult.INVALID_UTF8?;
+	uc += (c & 0x3F) << 12;
+	c = ptr++[0];
+	if (c & 0xC0 != 0x80) return UnicodeResult.INVALID_UTF8?;
+	uc += (c & 0x3F) << 6;
+	c = ptr++[0];
+	// Overlong sequence or invalid last
+	if (!uc || c & 0xC0 != 0x80) return UnicodeResult.INVALID_UTF8?;
+	return uc + c & 0x3F;
+}
+
+<*
+ @param utf8 `An UTF-8 encoded slice of bytes`
+ @return `the number of encoded code points`
+*>
+fn usz utf8_codepoints(String utf8)
+{
+	usz len = 0;
+	foreach (char c : utf8)
+	{
+		if (c & 0xC0 != 0x80) len++;
+	}
+	return len;
+}
+
+<*
+ Calculate the UTF8 length required to encode an UTF32 array.
+ @param [in] utf32 `the utf32 data to calculate from`
+ @return `the length of the resulting UTF8 array`
+*>
+fn usz utf8len_for_utf32(Char32[] utf32)
+{
+	usz len = 0;
+	foreach (Char32 uc : utf32)
+	{
+		switch (true)
+		{
+			case uc <= 0x7f:
+				len++;
+			case uc <= 0x7ff:
+				len += 2;
+			case uc <= 0xffff:
+				len += 3;
+			default:
+				len += 4;
+		}
+	}
+	return len;
+}
+
+<*
+ Calculate the UTF8 length required to encode an UTF16 array.
+ @param [in] utf16 `the utf16 data to calculate from`
+ @return `the length of the resulting UTF8 array`
+*>
+fn usz utf8len_for_utf16(Char16[] utf16)
+{
+	usz len = 0;
+	usz len16 = utf16.len;
+	for (usz i = 0; i < len16; i++)
+	{
+		Char16 c = utf16[i];
+		if (c & UTF16_SURROGATE_GENERIC_MASK != UTF16_SURROGATE_GENERIC_VALUE)
+		{
+			if (c <= 0x7f)
+			{
+				len++;
+				continue;
+			}
+			if (c <= 0x7ff)
+			{
+				len += 2;
+				continue;
+			}
+			len += 3;
+			continue;
+		}
+		len += 4;
+	}
+	return len;
+}
+
+<*
+ Calculate the UTF16 length required to encode a UTF8 array.
+ @param utf8 `the utf8 data to calculate from`
+ @return `the length of the resulting UTF16 array`
+*>
+fn usz utf16len_for_utf8(String utf8)
+{
+	usz len = utf8.len;
+	usz len16 = 0;
+	for (usz i = 0; i < len; i++)
+	{
+		len16++;
+		char c = utf8[i];
+		if (c & 0x80 == 0) continue;
+		i++;
+		if (c & 0xE0 == 0xC0) continue;
+		i++;
+		if (c & 0xF0 == 0xE0) continue;
+		i++;
+		len16++;
+	}
+	return len16;
+}
+
+<*
+ @param [in] utf32 `the UTF32 array to check the length for`
+ @return `the required length of an UTF16 array to hold the UTF32 data.`
+*>
+fn usz utf16len_for_utf32(Char32[] utf32)
+{
+	usz len = utf32.len;
+	foreach (Char32 uc : utf32)
+	{
+		if (uc >= UTF16_SURROGATE_OFFSET) len++;
+	}
+	return len;
+}
+
+<*
+ Convert an UTF32 array to an UTF8 array.
+
+ @param [in] utf32
+ @param [out] utf8_buffer
+ @return `the number of bytes written.`
+*>
+fn usz! utf32to8(Char32[] utf32, char[] utf8_buffer)
+{
+	char[] buffer = utf8_buffer;
+	foreach (uc : utf32)
+	{
+		usz used = char32_to_utf8(uc, buffer) @inline!;
+		buffer = buffer[used..];
+	}
+	// Zero terminate if there is space.
+	if (buffer.len > 0) buffer[0] = 0;
+	return utf8_buffer.len - buffer.len;
+}
+
+<*
+ Convert an UTF8 array to an UTF32 array.
+
+ @param [in] utf8
+ @param [out] utf32_buffer
+ @return `the number of Char32s written.`
+*>
+fn usz! utf8to32(String utf8, Char32[] utf32_buffer)
+{
+	usz len = utf8.len;
+	Char32* ptr = utf32_buffer.ptr;
+	usz len32 = 0;
+	usz buf_len = utf32_buffer.len;
+	for (usz i = 0; i < len;)
+	{
+		if (len32 == buf_len) return UnicodeResult.CONVERSION_FAILED?;
+		usz width = len - i;
+		Char32 uc = utf8_to_char32(&utf8[i], &width) @inline!;
+		i += width;
+		ptr[len32++] = uc;
+	}
+	// Zero terminate if possible
+	if (len32 + 1 < buf_len) ptr[len32] = 0;
+	return len32;
+}
+
+<*
+ Copy an array of UTF16 data into an UTF8 buffer without bounds
+ checking. This will assume the buffer is sufficiently large to hold
+ the converted data.
+
+ @param [in] utf16 `The UTF16 array containing the data to convert.`
+ @param [out] utf8_buffer `the (sufficiently large) buffer to hold the UTF16 data.`
+*>
+fn void! utf16to8_unsafe(Char16[] utf16, char* utf8_buffer)
+{
+	usz len16 = utf16.len;
+	for (usz i = 0; i < len16;)
+	{
+		usz available = len16 - i;
+		char16_to_utf8_unsafe(&utf16[i], &available, &utf8_buffer) @inline!;
+		i += available;
+	}
+}
+
+<*
+ Copy an array of UTF8 data into an UTF32 buffer without bounds
+ checking. This will assume the buffer is sufficiently large to hold
+ the converted data.
+
+ @param [in] utf8 `The UTF8 buffer containing the data to convert.`
+ @param [out] utf32_buffer `the (sufficiently large) buffer to hold the UTF8 data.`
+*>
+fn void! utf8to32_unsafe(String utf8, Char32* utf32_buffer)
+{
+	usz len = utf8.len;
+	for (usz i = 0; i < len;)
+	{
+		usz width = len - i;
+		Char32 uc = utf8_to_char32(&utf8[i], &width) @inline!;
+		i += width;
+		(utf32_buffer++)[0] = uc;
+	}
+}
+
+<*
+ Copy an array of UTF8 data into an UTF16 buffer without bounds
+ checking. This will assume the buffer is sufficiently large to hold
+ the converted data.
+
+ @param [in] utf8 `The UTF8 buffer containing the data to convert.`
+ @param [out] utf16_buffer `the (sufficiently large) buffer to hold the UTF8 data.`
+*>
+fn void! utf8to16_unsafe(String utf8, Char16* utf16_buffer)
+{
+	usz len = utf8.len;
+	for (usz i = 0; i < len;)
+	{
+		usz width = len - i;
+		Char32 uc = utf8_to_char32(&utf8[i], &width) @inline!;
+		char32_to_utf16_unsafe(uc, &utf16_buffer) @inline;
+		i += width;
+	}
+}
+
+<*
+ Copy an array of UTF32 code points into an UTF8 buffer without bounds
+ checking. This will assume the buffer is sufficiently large to hold
+ the converted data.
+
+ @param [in] utf32 `The UTF32 buffer containing the data to convert.`
+ @param [out] utf8_buffer `the (sufficiently large) buffer to hold the UTF8 data.`
+*>
+fn void utf32to8_unsafe(Char32[] utf32, char* utf8_buffer)
+{
+	char* start = utf8_buffer;
+	foreach (Char32 uc : utf32)
+	{
+		char32_to_utf8_unsafe(uc, &utf8_buffer) @inline;
+	}
+}

lib/std/core/dstring.c3

@@ -0,0 +1,695 @@
+module std::core::dstring;
+import std::io;
+
+distinct DString (OutStream) = DStringOpaque*;
+distinct DStringOpaque = void;
+
+const usz MIN_CAPACITY @private = 16;
+
+<*
+ @require !self.data() "String already initialized"
+*>
+fn DString DString.new_init(&self, usz capacity = MIN_CAPACITY, Allocator allocator = allocator::heap()) @deprecated("Use init(mem)")
+{
+	if (capacity < MIN_CAPACITY) capacity = MIN_CAPACITY;
+	StringData* data = allocator::alloc_with_padding(allocator, StringData, capacity)!!;
+	data.allocator = allocator;
+	data.len = 0;
+	data.capacity = capacity;
+	return *self = (DString)data;
+}
+
+<*
+ @require !self.data() "String already initialized"
+*>
+fn DString DString.init(&self, Allocator allocator, usz capacity = MIN_CAPACITY)
+{
+	if (capacity < MIN_CAPACITY) capacity = MIN_CAPACITY;
+	StringData* data = allocator::alloc_with_padding(allocator, StringData, capacity)!!;
+	data.allocator = allocator;
+	data.len = 0;
+	data.capacity = capacity;
+	return *self = (DString)data;
+}
+
+<*
+ @require !self.data() "String already initialized"
+*>
+fn DString DString.temp_init(&self, usz capacity = MIN_CAPACITY) @deprecated("Use tinit()")
+{
+	self.init(allocator::temp(), capacity) @inline;
+	return *self;
+}
+
+<*
+ @require !self.data() "String already initialized"
+*>
+fn DString DString.tinit(&self, usz capacity = MIN_CAPACITY)
+{
+	self.init(allocator::temp(), capacity) @inline;
+	return *self;
+}
+
+fn DString new_with_capacity(usz capacity, Allocator allocator = allocator::heap())
+{
+	return (DString){}.init(allocator, capacity);
+}
+
+fn DString temp_with_capacity(usz capacity) => new_with_capacity(capacity, allocator::temp()) @inline;
+
+fn DString new(String c = "", Allocator allocator = allocator::heap())
+{
+	usz len = c.len;
+	StringData* data = (StringData*)new_with_capacity(len, allocator);
+	if (len)
+	{
+		data.len = len;
+		mem::copy(&data.chars, c.ptr, len);
+	}
+	return (DString)data;
+}
+
+fn DString temp_new(String s = "") => new(s, allocator::temp()) @inline;
+
+
+fn void DString.replace_char(self, char ch, char replacement)
+{
+	StringData* data = self.data();
+	foreach (&c : data.chars[:data.len])
+	{
+		if (*c == ch) *c = replacement;
+	}
+}
+
+fn void DString.replace(&self, String needle, String replacement)
+{
+	StringData* data = self.data();
+	usz needle_len = needle.len;
+	if (!data || data.len < needle_len) return;
+	usz replace_len = replacement.len;
+	if (needle_len == 1 && replace_len == 1)
+	{
+		self.replace_char(needle[0], replacement[0]);
+		return;
+	}
+	@pool(data.allocator) {
+		String str = self.tcopy_str();
+		self.clear();
+		usz len = str.len;
+		usz match = 0;
+		foreach (i, c : str)
+		{
+			if (c == needle[match])
+			{
+				match++;
+				if (match == needle_len)
+				{
+					self.append_chars(replacement);
+					match = 0;
+					continue;
+				}
+				continue;
+			}
+			if (match > 0)
+			{
+				self.append_chars(str[i - match:match]);
+				match = 0;
+			}
+			self.append_char(c);
+		}
+		if (match > 0) self.append_chars(str[^match:match]);
+	};
+}
+
+fn DString DString.new_concat(self, DString b, Allocator allocator = allocator::heap()) @deprecated("Use concat(mem)")
+{
+	DString string;
+	string.init(allocator, self.len() + b.len());
+	string.append(self);
+	string.append(b);
+	return string;
+}
+
+fn DString DString.concat(self, Allocator allocator, DString b)
+{
+	DString string;
+	string.init(allocator, self.len() + b.len());
+	string.append(self);
+	string.append(b);
+	return string;
+}
+
+fn DString DString.temp_concat(self, DString b) => self.concat(allocator::temp(), b);
+
+fn ZString DString.zstr_view(&self)
+{
+	StringData* data = self.data();
+	if (!data) return "";
+	if (data.capacity == data.len)
+	{
+		self.reserve(1);
+		data = self.data();
+		data.chars[data.len] = 0;
+	}
+	else if (data.chars[data.len] != 0)
+	{
+		data.chars[data.len] = 0;
+	}
+	return (ZString)&data.chars[0];
+}
+
+fn usz DString.capacity(self)
+{
+	if (!self) return 0;
+	return self.data().capacity;
+}
+
+fn usz DString.len(&self) @dynamic @operator(len)
+{
+	if (!*self) return 0;
+	return self.data().len;
+}
+
+<*
+ @require new_size <= self.len()
+*>
+fn void DString.chop(self, usz new_size)
+{
+	if (!self) return;
+	self.data().len = new_size;
+}
+
+fn String DString.str_view(self)
+{
+	StringData* data = self.data();
+	if (!data) return "";
+	return (String)data.chars[:data.len];
+}
+
+<*
+ @require index < self.len()
+ @require self.data() != null "Empty string"
+*>
+fn char DString.char_at(self, usz index) @operator([])
+{
+	return self.data().chars[index];
+}
+
+<*
+ @require index < self.len()
+ @require self.data() != null  "Empty string"
+*>
+fn char* DString.char_ref(&self, usz index) @operator(&[])
+{
+	return &self.data().chars[index];
+}
+
+fn usz DString.append_utf32(&self, Char32[] chars)
+{
+	self.reserve(chars.len);
+	usz end = self.len();
+	foreach (Char32 c : chars)
+	{
+		self.append_char32(c);
+	}
+	return self.data().len - end;
+}
+
+<*
+ @require index < self.len()
+*>
+fn void DString.set(self, usz index, char c) @operator([]=)
+{
+	self.data().chars[index] = c;
+}
+
+fn void DString.append_repeat(&self, char c, usz times)
+{
+	if (times == 0) return;
+	self.reserve(times);
+	StringData* data = self.data();
+	for (usz i = 0; i < times; i++)
+	{
+		data.chars[data.len++] = c;
+	}
+}
+
+<*
+ @require c <= 0x10ffff
+*>
+fn usz DString.append_char32(&self, Char32 c)
+{
+	char[4] buffer @noinit;
+	char* p = &buffer;
+	usz n = conv::char32_to_utf8_unsafe(c, &p);
+	self.reserve(n);
+	StringData* data = self.data();
+	data.chars[data.len:n] = buffer[:n];
+	data.len += n;
+	return n;
+}
+
+fn DString DString.tcopy(&self) => self.copy(allocator::temp());
+
+fn DString DString.copy(self, Allocator allocator = null)
+{
+	if (!self)
+	{
+		if (allocator) return new_with_capacity(0, allocator);
+		return (DString)null;
+	}
+	StringData* data = self.data();
+	if (!allocator) allocator = allocator::heap();
+	DString new_string = new_with_capacity(data.capacity, allocator);
+	mem::copy((char*)new_string.data(), (char*)data, StringData.sizeof + data.len);
+	return new_string;
+}
+
+fn ZString DString.copy_zstr(self, Allocator allocator = allocator::heap())
+{
+	usz str_len = self.len();
+	if (!str_len)
+	{
+		return (ZString)allocator::calloc(allocator, 1);
+	}
+	char* zstr = allocator::malloc(allocator, str_len + 1);
+	StringData* data = self.data();
+	mem::copy(zstr, &data.chars, str_len);
+	zstr[str_len] = 0;
+	return (ZString)zstr;
+}
+
+fn String DString.copy_str(self, Allocator allocator = allocator::heap())
+{
+	return (String)self.copy_zstr(allocator)[:self.len()];
+}
+
+fn String DString.tcopy_str(self) => self.copy_str(allocator::temp()) @inline;
+
+fn bool DString.equals(self, DString other_string)
+{
+	StringData *str1 = self.data();
+	StringData *str2 = other_string.data();
+	if (str1 == str2) return true;
+	if (!str1) return str2.len == 0;
+	if (!str2) return str1.len == 0;
+	usz str1_len = str1.len;
+	if (str1_len != str2.len) return false;
+	for (int i = 0; i < str1_len; i++)
+	{
+		if (str1.chars[i] != str2.chars[i]) return false;
+	}
+	return true;
+}
+
+fn void DString.free(&self)
+{
+	if (!*self) return;
+	StringData* data = self.data();
+	if (!data) return;
+	allocator::free(data.allocator, data);
+	*self = (DString)null;
+}
+
+fn bool DString.less(self, DString other_string)
+{
+	StringData* str1 = self.data();
+	StringData* str2 = other_string.data();
+	if (str1 == str2) return false;
+	if (!str1) return str2.len != 0;
+	if (!str2) return str1.len == 0;
+	usz str1_len = str1.len;
+	usz str2_len = str2.len;
+	if (str1_len != str2_len) return str1_len < str2_len;
+	for (int i = 0; i < str1_len; i++)
+	{
+		if (str1.chars[i] >= str2.chars[i]) return false;
+	}
+	return true;
+}
+
+fn void DString.append_chars(&self, String str)
+{
+	usz other_len = str.len;
+	if (!other_len) return;
+	if (!*self)
+	{
+		*self = new(str);
+		return;
+	}
+	self.reserve(other_len);
+	StringData* data = self.data();
+	mem::copy(&data.chars[data.len], str.ptr, other_len);
+	data.len += other_len;
+}
+
+fn Char32[] DString.copy_utf32(&self, Allocator allocator = allocator::heap())
+{
+	return self.str_view().to_utf32(allocator) @inline!!;
+}
+
+fn void DString.append_string(&self, DString str)
+{
+	StringData* other = str.data();
+	if (!other) return;
+	self.append(str.str_view());
+}
+
+fn void DString.clear(self)
+{
+	if (!self) return;
+	self.data().len = 0;
+}
+
+fn usz! DString.write(&self, char[] buffer) @dynamic
+{
+	self.append_chars((String)buffer);
+	return buffer.len;
+}
+
+fn void! DString.write_byte(&self, char c) @dynamic
+{
+	self.append_char(c);
+}
+
+fn void DString.append_char(&self, char c)
+{
+	if (!*self)
+	{
+		*self = new_with_capacity(MIN_CAPACITY);
+	}
+	self.reserve(1);
+	StringData* data = self.data();
+	data.chars[data.len++] = c;
+}
+
+<*
+ @require start < self.len()
+ @require end < self.len()
+ @require end >= start "End must be same or equal to the start"
+*>
+fn void DString.delete_range(&self, usz start, usz end)
+{
+	self.delete(start, end - start + 1);
+}
+
+<*
+ @require start < self.len()
+ @require start + len <= self.len()
+*>
+fn void DString.delete(&self, usz start, usz len = 1)
+{
+	if (!len) return;
+	StringData* data = self.data();
+	usz new_len = data.len - len;
+	if (new_len == 0)
+	{
+		data.len = 0;
+		return;
+	}
+	usz len_after = data.len - start - len;
+	if (len_after > 0)
+	{
+		data.chars[start:len_after] = data.chars[start + len:len_after];
+	}
+	data.len = new_len;
+}
+
+macro void DString.append(&self, value)
+{
+	var $Type = $typeof(value);
+	$switch ($Type)
+		$case char:
+		$case ichar:
+			self.append_char(value);
+		$case DString:
+			self.append_string(value);
+		$case String:
+			self.append_chars(value);
+		$case Char32:
+			self.append_char32(value);
+		$default:
+			$switch
+				$case $defined((Char32)value):
+					self.append_char32((Char32)value);
+				$case $defined((String)value):
+					self.append_chars((String)value);
+				$default:
+					$error "Unsupported type for append – use appendf instead.";
+			$endswitch
+	$endswitch
+}
+
+<*
+ @require index <= self.len()
+*>
+fn void DString.insert_chars_at(&self, usz index, String s)
+{
+	if (s.len == 0) return;
+	self.reserve(s.len);
+	StringData* data = self.data();
+	usz len = self.len();
+	if (data.chars[:len].ptr == s.ptr)
+	{
+		// Source and destination are the same: nothing to do.
+		return;
+	}
+	index = min(index, len);
+	data.len += s.len;
+
+	char* start = data.chars[index:s.len].ptr; // area to insert into
+	mem::move(start + s.len, start, len - index); // move existing data
+	switch
+	{
+		case s.ptr <= start && start < s.ptr + s.len:
+			// Overlapping areas.
+			foreach_r (i, c : s)
+			{
+				data.chars[index + i] = c;
+			}
+		case start <= s.ptr && s.ptr < start + len:
+			// Source has moved.
+			mem::move(start, s.ptr + s.len, s.len);
+		default:
+			mem::move(start, s, s.len);
+	}
+}
+
+<*
+ @require index <= self.len()
+*>
+fn void DString.insert_string_at(&self, usz index, DString str)
+{
+	StringData* other = str.data();
+	if (!other) return;
+	self.insert_at(index, str.str_view());
+}
+
+<*
+ @require index <= self.len()
+*>
+fn void DString.insert_char_at(&self, usz index, char c)
+{
+	self.reserve(1);
+	StringData* data = self.data();
+
+	char* start = &data.chars[index];
+	mem::move(start + 1, start, self.len() - index);
+	data.chars[index] = c;
+	data.len++;
+}
+
+<*
+ @require index <= self.len()
+*>
+fn usz DString.insert_char32_at(&self, usz index, Char32 c)
+{
+	char[4] buffer @noinit;
+	char* p = &buffer;
+	usz n = conv::char32_to_utf8_unsafe(c, &p);
+
+	self.reserve(n);
+	StringData* data = self.data();
+
+	char* start = &data.chars[index];
+	mem::move(start + n, start, self.len() - index);
+	data.chars[index:n] = buffer[:n];
+	data.len += n;
+
+	return n;
+}
+
+<*
+ @require index <= self.len()
+*>
+fn usz DString.insert_utf32_at(&self, usz index, Char32[] chars)
+{
+	usz n = conv::utf8len_for_utf32(chars);
+	
+	self.reserve(n);
+	StringData* data = self.data();
+
+	char* start = &data.chars[index];
+	mem::move(start + n, start, self.len() - index);
+
+	char[4] buffer @noinit;
+
+	foreach(c : chars)
+	{
+		char* p = &buffer;
+		usz m = conv::char32_to_utf8_unsafe(c, &p);
+		data.chars[index:m] = buffer[:m];
+		index += m;
+	}
+
+	data.len += n;
+
+	return n;
+}
+
+macro void DString.insert_at(&self, usz index, value)
+{
+	var $Type = $typeof(value);
+	$switch ($Type)
+		$case char:
+		$case ichar:
+			self.insert_char_at(index, value);
+		$case DString:
+			self.insert_string_at(index, value);
+		$case String:
+			self.insert_chars_at(index, value);
+		$case Char32:
+			self.insert_char32_at(index, value);
+		$default:
+			$switch
+				$case $defined((Char32)value):
+					self.insert_char32_at(index, (Char32)value);
+				$case $defined((String)value):
+					self.insert_chars_at(index, (String)value);
+				$default:
+					$error "Unsupported type for insert";
+			$endswitch
+	$endswitch
+}
+
+fn usz! DString.appendf(&self, String format, args...) @maydiscard
+{
+	if (!self.data()) self.init(mem, format.len + 20);
+	@pool(self.data().allocator)
+	{
+		Formatter formatter;
+		formatter.init(&out_string_append_fn, self);
+		return formatter.vprintf(format, args);
+	};
+}
+
+fn usz! DString.appendfn(&self, String format, args...) @maydiscard
+{
+	if (!self.data()) self.init(mem, format.len + 20);
+	@pool(self.data().allocator)
+	{
+		Formatter formatter;
+		formatter.init(&out_string_append_fn, self);
+		usz len = formatter.vprintf(format, args)!;
+		self.append('\n');
+		return len + 1;
+	};
+}
+
+fn DString new_join(String[] s, String joiner, Allocator allocator = allocator::heap())
+{
+	if (!s.len) return (DString)null;
+	usz total_size = joiner.len * s.len;
+	foreach (String* &str : s)
+	{
+		total_size += str.len;
+	}
+	DString res = new_with_capacity(total_size, allocator);
+	res.append(s[0]);
+	foreach (String* &str : s[1..])
+	{
+		res.append(joiner);
+		res.append(*str);
+	}
+	return res;
+}
+
+fn void! out_string_append_fn(void* data, char c) @private
+{
+	DString* s = data;
+	s.append_char(c);
+}
+
+fn void DString.reverse(self)
+{
+	StringData *data = self.data();
+	if (!data) return;
+	isz mid = data.len / 2;
+	for (isz i = 0; i < mid; i++)
+	{
+		char temp = data.chars[i];
+		isz reverse_index = data.len - 1 - i;
+		data.chars[i] = data.chars[reverse_index];
+		data.chars[reverse_index] = temp;
+	}
+}
+
+fn StringData* DString.data(self) @inline @private
+{
+	return (StringData*)self;
+}
+
+fn void DString.reserve(&self, usz addition)
+{
+	StringData* data = self.data();
+	if (!data)
+	{
+		*self = dstring::new_with_capacity(addition);
+		return;
+	}
+	usz len = data.len + addition;
+	if (data.capacity >= len) return;
+	usz new_capacity = data.capacity * 2;
+	if (new_capacity < MIN_CAPACITY) new_capacity = MIN_CAPACITY;
+	while (new_capacity < len) new_capacity *= 2;
+	data.capacity = new_capacity;
+	*self = (DString)allocator::realloc(data.allocator, data, StringData.sizeof + new_capacity);
+}
+
+fn usz! DString.read_from_stream(&self, InStream reader)
+{
+	if (&reader.available)
+	{
+		usz total_read = 0;
+		while (usz available = reader.available()!)
+		{
+			self.reserve(available);
+			StringData* data = self.data();
+			usz len = reader.read(data.chars[data.len..(data.capacity - 1)])!;
+			total_read += len;
+			data.len += len;
+		}
+		return total_read;
+	}
+	usz total_read = 0;
+	while (true)
+	{
+		// Reserve at least 16 bytes
+		self.reserve(16);
+		StringData* data = self.data();
+		// Read into the rest of the buffer
+		usz read = reader.read(data.chars[data.len..(data.capacity - 1)])!;
+		data.len += read;
+		// Ok, we reached the end.
+		if (read < 16) return total_read;
+		// Otherwise go another round
+	}
+}
+
+struct StringData @private
+{
+	Allocator allocator;
+	usz len;
+	usz capacity;
+	char[?] chars;
+}

lib/std/core/env.c3

@@ -0,0 +1,197 @@
+// Copyright (c) 2021-2024 Christoffer Lerno. All rights reserved.
+// Use of this source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+module std::core::env;
+
+enum CompilerOptLevel
+{
+	O0,
+	O1,
+	O2,
+	O3
+}
+
+enum MemoryEnvironment
+{
+	NORMAL,
+	SMALL,
+	TINY,
+	NONE
+}
+
+enum OsType
+{
+	UNKNOWN,
+	NONE,
+	ANANAS,
+	CLOUD_ABI,
+	DRAGON_FLY,
+	FREEBSD,
+	FUCHSIA,
+	IOS,
+	KFREEBSD,
+	LINUX,
+	PS3,
+	MACOS,
+	NETBSD,
+	OPENBSD,
+	SOLARIS,
+	WIN32,
+	HAIKU,
+	MINIX,
+	RTEMS,
+	NACL,       // Native Client
+	CNK,        // BG/P Compute-Node Kernel
+	AIX,
+	CUDA,
+	NVOPENCL,
+	AMDHSA,
+	PS4,
+	ELFIAMCU,
+	TVOS,
+	WATCHOS,
+	MESA3D,
+	CONTIKI,
+	AMDPAL,
+	HERMITCORE,
+	HURD,
+	WASI,
+	EMSCRIPTEN,
+}
+
+enum ArchType
+{
+	UNKNOWN,
+	ARM,          // ARM (little endian): arm, armv.*, xscale
+	ARMB,         // ARM (big endian): armeb
+	AARCH64,      // AArch64 (little endian): aarch64
+	AARCH64_BE,   // AArch64 (big endian): aarch64_be
+	AARCH64_32,   // AArch64 (little endian) ILP32: aarch64_32
+	ARC,          // ARC: Synopsys ARC
+	AVR,          // AVR: Atmel AVR microcontroller
+	BPFEL,        // eBPF or extended BPF or 64-bit BPF (little endian)
+	BPFEB,        // eBPF or extended BPF or 64-bit BPF (big endian)
+	HEXAGON,      // Hexagon: hexagon
+	MIPS,         // MIPS: mips, mipsallegrex, mipsr6
+	MIPSEL,       // MIPSEL: mipsel, mipsallegrexe, mipsr6el
+	MIPS64,       // MIPS64: mips64, mips64r6, mipsn32, mipsn32r6
+	MIPS64EL,     // MIPS64EL: mips64el, mips64r6el, mipsn32el, mipsn32r6el
+	MSP430,       // MSP430: msp430
+	PPC,          // PPC: powerpc
+	PPC64,        // PPC64: powerpc64, ppu
+	PPC64LE,      // PPC64LE: powerpc64le
+	R600,         // R600: AMD GPUs HD2XXX - HD6XXX
+	AMDGCN,       // AMDGCN: AMD GCN GPUs
+	RISCV32,      // RISC-V (32-bit): riscv32
+	RISCV64,      // RISC-V (64-bit): riscv64
+	SPARC,        // Sparc: sparc
+	SPARCV9,      // Sparcv9: Sparcv9
+	SPARCEL,      // Sparc: (endianness = little). NB: 'Sparcle' is a CPU variant
+	SYSTEMZ,      // SystemZ: s390x
+	TCE,          // TCE (http://tce.cs.tut.fi/): tce
+	TCELE,        // TCE little endian (http://tce.cs.tut.fi/): tcele
+	THUMB,        // Thumb (little endian): thumb, thumbv.*
+	THUMBEB,      // Thumb (big endian): thumbeb
+	X86,          // X86: i[3-9]86
+	X86_64,       // X86-64: amd64, x86_64
+	XCORE,        // XCore: xcore
+	NVPTX,        // NVPTX: 32-bit
+	NVPTX64,      // NVPTX: 64-bit
+	LE32,         // le32: generic little-endian 32-bit CPU (PNaCl)
+	LE64,         // le64: generic little-endian 64-bit CPU (PNaCl)
+	AMDIL,        // AMDIL
+	AMDIL64,      // AMDIL with 64-bit pointers
+	HSAIL,        // AMD HSAIL
+	HSAIL64,      // AMD HSAIL with 64-bit pointers
+	SPIR,         // SPIR: standard portable IR for OpenCL 32-bit version
+	SPIR64,       // SPIR: standard portable IR for OpenCL 64-bit version
+	KALIMBA,      // Kalimba: generic kalimba
+	SHAVE,        // SHAVE: Movidius vector VLIW processors
+	LANAI,        // Lanai: Lanai 32-bit
+	WASM32,       // WebAssembly with 32-bit pointers
+	WASM64,       // WebAssembly with 64-bit pointers
+	RSCRIPT32,    // 32-bit RenderScript
+	RSCRIPT64,    // 64-bit RenderScript
+	XTENSA,       // Xtensa
+}
+
+const String COMPILER_BUILD_HASH = $$BUILD_HASH;
+const String COMPILER_BUILD_DATE = $$BUILD_DATE;
+const OsType OS_TYPE = OsType.from_ordinal($$OS_TYPE);
+const ArchType ARCH_TYPE = ArchType.from_ordinal($$ARCH_TYPE);
+const bool ARCH_32_BIT = $$REGISTER_SIZE == 32;
+const bool ARCH_64_BIT = $$REGISTER_SIZE == 64;
+const bool LIBC = $$COMPILER_LIBC_AVAILABLE;
+const bool NO_LIBC = !$$COMPILER_LIBC_AVAILABLE;
+const CompilerOptLevel COMPILER_OPT_LEVEL = CompilerOptLevel.from_ordinal($$COMPILER_OPT_LEVEL);
+const bool BIG_ENDIAN = $$PLATFORM_BIG_ENDIAN;
+const bool I128_NATIVE_SUPPORT = $$PLATFORM_I128_SUPPORTED;
+const bool F16_SUPPORT = $$PLATFORM_F16_SUPPORTED;
+const bool F128_SUPPORT = $$PLATFORM_F128_SUPPORTED;
+const REGISTER_SIZE = $$REGISTER_SIZE;
+const bool COMPILER_SAFE_MODE = $$COMPILER_SAFE_MODE;
+const bool DEBUG_SYMBOLS = $$DEBUG_SYMBOLS;
+const bool BACKTRACE = $$BACKTRACE;
+const usz LLVM_VERSION = $$LLVM_VERSION;
+const bool BENCHMARKING = $$BENCHMARKING;
+const bool TESTING = $$TESTING;
+const MemoryEnvironment MEMORY_ENV = MemoryEnvironment.from_ordinal($$MEMORY_ENVIRONMENT);
+const bool TRACK_MEMORY = DEBUG_SYMBOLS && (COMPILER_SAFE_MODE || TESTING);
+const bool X86_64 = ARCH_TYPE == X86_64;
+const bool X86 = ARCH_TYPE == X86;
+const bool AARCH64 = ARCH_TYPE == AARCH64;
+const bool NATIVE_STACKTRACE = LINUX || DARWIN || WIN32;
+const bool LINUX = LIBC && OS_TYPE == LINUX;
+const bool DARWIN = LIBC && os_is_darwin();
+const bool WIN32 = LIBC && OS_TYPE == WIN32;
+const bool POSIX = LIBC && os_is_posix();
+const bool OPENBSD = LIBC && OS_TYPE == OPENBSD;
+const bool FREEBSD = LIBC && OS_TYPE == FREEBSD;
+const bool NETBSD = LIBC && OS_TYPE == NETBSD;
+const bool BSD_FAMILY = env::FREEBSD || env::OPENBSD || env::NETBSD;
+const bool WASI = LIBC && OS_TYPE == WASI;
+const bool WASM_NOLIBC @builtin = !LIBC && ARCH_TYPE == ArchType.WASM32 || ARCH_TYPE == ArchType.WASM64;
+const bool ADDRESS_SANITIZER = $$ADDRESS_SANITIZER;
+const bool MEMORY_SANITIZER = $$MEMORY_SANITIZER;
+const bool THREAD_SANITIZER = $$THREAD_SANITIZER;
+const bool ANY_SANITIZER = ADDRESS_SANITIZER || MEMORY_SANITIZER || THREAD_SANITIZER;
+
+macro bool os_is_darwin() @const
+{
+	$switch (OS_TYPE)
+	$case IOS:
+	$case MACOS:
+	$case TVOS:
+	$case WATCHOS:
+		return true;
+	$default:
+		return false;
+	$endswitch
+}
+
+macro bool os_is_posix() @const
+{
+	$switch (OS_TYPE)
+	$case IOS:
+	$case MACOS:
+	$case NETBSD:
+	$case LINUX:
+	$case KFREEBSD:
+	$case FREEBSD:
+	$case OPENBSD:
+	$case SOLARIS:
+	$case TVOS:
+	$case WATCHOS:
+		return true;
+	$case WIN32:
+	$case WASI:
+	$case EMSCRIPTEN:
+		return false;
+	$default:
+		$echo("Assuming non-Posix environment");
+		return false;
+	$endswitch
+}
+
+const BUILTIN_EXPECT_IS_DISABLED = $feature(DISABLE_BUILTIN_EXPECT);
+const BUILTIN_PREFETCH_IS_DISABLED = $feature(DISABLE_BUILTIN_PREFETCH);

lib/std/core/mem.c3

@@ -0,0 +1,911 @@
+// Copyright (c) 2021-2023 Christoffer Lerno. All rights reserved.
+// Use of this source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+module std::core::mem;
+import std::core::mem::allocator @public;
+import std::math;
+
+const MAX_MEMORY_ALIGNMENT = 0x1000_0000;
+const DEFAULT_MEM_ALIGNMENT = (void*.alignof) * 2;
+
+macro bool @constant_is_power_of_2($x) @const @private
+{
+	return $x != 0 && ($x & ($x - 1)) == 0;
+}
+
+<*
+ Load a vector from memory according to a mask assuming default alignment.
+
+ @param ptr "The pointer address to load from."
+ @param mask "The mask for the load"
+ @param passthru "The value to use for non masked values"
+ @require $assignable(&&passthru, $typeof(ptr)) : "Pointer and passthru must match"
+ @require @typekind(passthru) == VECTOR : "Expected passthru to be a vector"
+ @require passthru.len == mask.len : "Mask and passthru must have the same length"
+
+ @return "A vector with the loaded values where the mask is true, passthru where the mask is false"
+*>
+macro masked_load(ptr, bool[<?>] mask, passthru)
+{
+	return $$masked_load(ptr, mask, passthru, 0);
+}
+
+<*
+ Load a vector from memory according to a mask.
+
+ @param ptr "The pointer address to load from."
+ @param mask "The mask for the load"
+ @param passthru "The value to use for non masked values"
+ @param $alignment "The alignment to assume for the pointer"
+
+ @require $assignable(&&passthru, $typeof(ptr)) : "Pointer and passthru must match"
+ @require @typekind(passthru) == VECTOR : "Expected passthru to be a vector"
+ @require passthru.len == mask.len : "Mask and passthru must have the same length"
+ @require @constant_is_power_of_2($alignment) : "The alignment must be a power of two"
+
+ @return "A vector with the loaded values where the mask is true, passthru where the mask is false"
+*>
+macro @masked_load_aligned(ptr, bool[<?>] mask, passthru, usz $alignment)
+{
+	return $$masked_load(ptr, mask, passthru, $alignment);
+}
+
+<*
+ Load values from a pointer vector, assuming default alignment.
+
+ @param ptrvec "The vector of pointers to load from."
+ @param mask "The mask for the load"
+ @param passthru "The value to use for non masked values"
+
+ @require @typekind(ptrvec) == VECTOR : "Expected ptrvec to be a vector"
+ @require @typekind(passthru) == VECTOR : "Expected passthru to be a vector"
+ @require $assignable(&&passthru[0], $typeof(ptrvec[0])) : "Pointer and passthru must match"
+ @require passthru.len == mask.len : "Mask and passthru must have the same length"
+ @require mask.len == ptrvec.len : "Mask and ptrvec must have the same length"
+
+ @return "A vector with the loaded values where the mask is true, passthru where the mask is false"
+*>
+macro gather(ptrvec, bool[<?>] mask, passthru)
+{
+	return $$gather(ptrvec, mask, passthru, 0);
+}
+
+
+<*
+ Load values from a pointer vector.
+
+ @param ptrvec "The vector of pointers to load from."
+ @param mask "The mask for the load"
+ @param passthru "The value to use for non masked values"
+ @param $alignment "The alignment to assume for the pointers"
+
+ @require @typekind(ptrvec) == VECTOR : "Expected ptrvec to be a vector"
+ @require @typekind(passthru) == VECTOR : "Expected passthru to be a vector"
+ @require $assignable(&&passthru[0], $typeof(ptrvec[0])) : "Pointer and passthru must match"
+ @require passthru.len == mask.len : "Mask and passthru must have the same length"
+ @require mask.len == ptrvec.len : "Mask and ptrvec must have the same length"
+ @require @constant_is_power_of_2($alignment) : "The alignment must be a power of two"
+
+ @return "A vector with the loaded values where the mask is true, passthru where the mask is false"
+*>
+macro @gather_aligned(ptrvec, bool[<?>] mask, passthru, usz $alignment)
+{
+	return $$gather(ptrvec, mask, passthru, $alignment);
+}
+
+
+<*
+ Store parts of a vector according to the mask, assuming default alignment.
+
+ @param ptr "The pointer address to store to."
+ @param value "The value to store masked"
+ @param mask "The mask for the store"
+
+ @require $assignable(&&value, $typeof(ptr)) : "Pointer and value must match"
+ @require @typekind(value) == VECTOR : "Expected value to be a vector"
+ @require value.len == mask.len : "Mask and value must have the same length"
+*>
+macro masked_store(ptr, value, bool[<?>] mask)
+{
+	return $$masked_store(ptr, value, mask, 0);
+}
+
+<*
+ @param ptr "The pointer address to store to."
+ @param value "The value to store masked"
+ @param mask "The mask for the store"
+ @param $alignment "The alignment of the pointer"
+
+ @require $assignable(&&value, $typeof(ptr)) : "Pointer and value must match"
+ @require @typekind(value) == VECTOR : "Expected value to be a vector"
+ @require value.len == mask.len : "Mask and value must have the same length"
+ @require @constant_is_power_of_2($alignment) : "The alignment must be a power of two"
+
+*>
+macro @masked_store_aligned(ptr, value, bool[<?>] mask, usz $alignment)
+{
+	return $$masked_store(ptr, value, mask, $alignment);
+}
+
+<*
+ @param ptrvec "The vector pointer containing the addresses to store to."
+ @param value "The value to store masked"
+ @param mask "The mask for the store"
+ @require @typekind(ptrvec) == VECTOR : "Expected ptrvec to be a vector"
+ @require @typekind(value) == VECTOR : "Expected value to be a vector"
+ @require $assignable(&&value[0], $typeof(ptrvec[0])) : "Pointer and value must match"
+ @require value.len == mask.len : "Mask and value must have the same length"
+ @require mask.len == ptrvec.len : "Mask and ptrvec must have the same length"
+
+*>
+macro scatter(ptrvec, value, bool[<?>] mask)
+{
+	return $$scatter(ptrvec, value, mask, 0);
+}
+
+<*
+ @param ptrvec "The vector pointer containing the addresses to store to."
+ @param value "The value to store masked"
+ @param mask "The mask for the store"
+ @param $alignment "The alignment of the load"
+
+ @require @typekind(ptrvec) == VECTOR : "Expected ptrvec to be a vector"
+ @require @typekind(value) == VECTOR : "Expected value to be a vector"
+ @require $assignable(&&value[0], $typeof(ptrvec[0])) : "Pointer and value must match"
+ @require value.len == mask.len : "Mask and value must have the same length"
+ @require mask.len == ptrvec.len : "Mask and ptrvec must have the same length"
+ @require @constant_is_power_of_2($alignment) : "The alignment must be a power of two"
+*>
+macro @scatter_aligned(ptrvec, value, bool[<?>] mask, usz $alignment)
+{
+	return $$scatter(ptrvec, value, mask, $alignment);
+}
+
+<*
+ @param #x "The variable or dereferenced pointer to load."
+ @param $alignment "The alignment to assume for the load"
+ @return "The value of the variable"
+
+ @require @constant_is_power_of_2($alignment) : "The alignment must be a power of two"
+ @require $defined(&#x) : "This must be a variable or dereferenced pointer"
+*>
+macro @unaligned_load(#x, usz $alignment) @builtin
+{
+	return $$unaligned_load(&#x, $alignment);
+}
+
+<*
+ @param #x "The variable or dereferenced pointer to store to."
+ @param value "The value to store."
+ @param $alignment "The alignment to assume for the store"
+ @return "The value stored"
+
+ @require $defined(&#x) : "This must be a variable or dereferenced pointer"
+ @require $defined(#x = value) : "The value doesn't match the variable"
+ @require @constant_is_power_of_2($alignment) : "The alignment must be a power of two"
+*>
+macro @unaligned_store(#x, value, usz $alignment) @builtin
+{
+	return $$unaligned_store(&#x, ($typeof(#x))value, $alignment);
+}
+
+<*
+ @param #x "The variable or dereferenced pointer to load."
+ @return "The value of the variable"
+
+ @require $defined(&#x) : "This must be a variable or dereferenced pointer"
+*>
+macro @volatile_load(#x) @builtin
+{
+	return $$volatile_load(&#x);
+}
+
+<*
+ @param #x "The variable or dereferenced pointer to store to."
+ @param value "The value to store."
+ @return "The value stored"
+
+ @require $defined(&#x) : "This must be a variable or dereferenced pointer"
+ @require $defined(#x = value) : "The value doesn't match the variable"
+*>
+macro @volatile_store(#x, value) @builtin
+{
+	return $$volatile_store(&#x, ($typeof(#x))value);
+}
+
+enum AtomicOrdering : int
+{
+	NOT_ATOMIC,         // Not atomic
+	UNORDERED,          // No lock
+	RELAXED,            // Consistent ordering
+	ACQUIRE,            // Barrier locking load/store
+	RELEASE,            // Barrier releasing load/store
+	ACQUIRE_RELEASE,    // Barrier fence to load/store
+	SEQ_CONSISTENT,     // Acquire semantics, ordered with other seq_consistent
+}
+
+<*
+ @param #x "the variable or dereferenced pointer to load."
+ @param $ordering "atomic ordering of the load, defaults to SEQ_CONSISTENT"
+ @param $volatile "whether the load should be volatile, defaults to 'false'"
+ @return "returns the value of x"
+
+ @require $defined(&#x) : "This must be a variable or dereferenced pointer"
+ @require $ordering != AtomicOrdering.RELEASE "Release ordering is not valid for load."
+ @require $ordering != AtomicOrdering.ACQUIRE_RELEASE "Acquire release is not valid for load."
+ @require types::may_load_atomic($typeof(#x)) "Only integer, float and pointers may be used."
+*>
+macro @atomic_load(#x, AtomicOrdering $ordering = SEQ_CONSISTENT, $volatile = false) @builtin
+{
+	return $$atomic_load(&#x, $volatile, $ordering.ordinal);
+}
+
+<*
+ @param #x "the variable or dereferenced pointer to store to."
+ @param value "the value to store."
+ @param $ordering "the atomic ordering of the store, defaults to SEQ_CONSISTENT"
+ @param $volatile "whether the store should be volatile, defaults to 'false'"
+
+ @require $ordering != AtomicOrdering.ACQUIRE "Acquire ordering is not valid for store."
+ @require $ordering != AtomicOrdering.ACQUIRE_RELEASE "Acquire release is not valid for store."
+ @require types::may_load_atomic($typeof(#x)) "Only integer, float and pointers may be used."
+ @require $defined(&#x) : "This must be a variable or dereferenced pointer"
+ @require $defined(#x = value) : "The value doesn't match the variable"
+*>
+macro void @atomic_store(#x, value, AtomicOrdering $ordering = SEQ_CONSISTENT, $volatile = false) @builtin
+{
+	$$atomic_store(&#x, value, $volatile, $ordering.ordinal);
+}
+
+<*
+ @require $success != AtomicOrdering.NOT_ATOMIC && $success != AtomicOrdering.UNORDERED "Acquire ordering is not valid."
+ @require $failure != AtomicOrdering.RELEASE && $failure != AtomicOrdering.ACQUIRE_RELEASE "Acquire release is not valid."
+*>
+macro compare_exchange(ptr, compare, value, AtomicOrdering $success = SEQ_CONSISTENT, AtomicOrdering $failure = SEQ_CONSISTENT, bool $volatile = true, bool $weak = false, usz $alignment = 0)
+{
+	return $$compare_exchange(ptr, compare, value, $volatile, $weak, $success.ordinal, $failure.ordinal, $alignment);
+}
+
+<*
+ @require $success != AtomicOrdering.NOT_ATOMIC && $success != AtomicOrdering.UNORDERED "Acquire ordering is not valid."
+ @require $failure != AtomicOrdering.RELEASE && $failure != AtomicOrdering.ACQUIRE_RELEASE "Acquire release is not valid."
+*>
+macro compare_exchange_volatile(ptr, compare, value, AtomicOrdering $success = SEQ_CONSISTENT, AtomicOrdering $failure = SEQ_CONSISTENT)
+{
+	return compare_exchange(ptr, compare, value, $success, $failure, true);
+}
+
+<*
+ @require math::is_power_of_2(alignment)
+*>
+fn usz aligned_offset(usz offset, usz alignment)
+{
+	return alignment * ((offset + alignment - 1) / alignment);
+}
+
+macro void* aligned_pointer(void* ptr, usz alignment)
+{
+	return (void*)(uptr)aligned_offset((uptr)ptr, alignment);
+}
+
+<*
+ @require math::is_power_of_2(alignment)
+*>
+fn bool ptr_is_aligned(void* ptr, usz alignment) @inline
+{
+	return (uptr)ptr & ((uptr)alignment - 1) == 0;
+}
+
+macro void zero_volatile(char[] data)
+{
+	$$memset(data.ptr, (char)0, data.len, true, (usz)1);
+}
+
+macro void clear(void* dst, usz len, usz $dst_align = 0, bool $is_volatile = false)
+{
+	$$memset(dst, (char)0, len, $is_volatile, $dst_align);
+}
+
+macro void clear_inline(void* dst, usz $len, usz $dst_align = 0, bool $is_volatile = false)
+{
+	$$memset_inline(dst, (char)0, $len, $is_volatile, $dst_align);
+}
+
+<*
+ Copy memory from src to dst efficiently, assuming the memory ranges do not overlap.
+
+ @param [&out] dst "The destination to copy to"
+ @param [&in] src "The source to copy from"
+ @param len "The number of bytes to copy"
+ @param $dst_align "the alignment of the destination if different from the default, 0 assumes the default"
+ @param $src_align "the alignment of the destination if different from the default, 0 assumes the default"
+ @param $is_volatile "True if this copy should be treated as volatile, i.e. it can't be optimized away."
+
+ @require len == 0 || dst + len <= src || src + len <= dst : "Ranges may not overlap"
+*>
+macro void copy(void* dst, void* src, usz len, usz $dst_align = 0, usz $src_align = 0, bool $is_volatile = false)
+{
+	$$memcpy(dst, src, len, $is_volatile, $dst_align, $src_align);
+}
+
+<*
+ Copy memory from src to dst efficiently, assuming the memory ranges do not overlap, it
+ will always be inlined and never call memcopy
+
+ @param [&out] dst "The destination to copy to"
+ @param [&in] src "The source to copy from"
+ @param $len "The number of bytes to copy"
+ @param $dst_align "the alignment of the destination if different from the default, 0 assumes the default"
+ @param $src_align "the alignment of the destination if different from the default, 0 assumes the default"
+ @param $is_volatile "True if this copy should be treated as volatile, i.e. it can't be optimized away."
+
+ @require $len == 0 || dst + $len <= src || src + $len <= dst : "Ranges may not overlap"
+*>
+macro void copy_inline(void* dst, void* src, usz $len, usz $dst_align = 0, usz $src_align = 0, bool $is_volatile = false)
+{
+	$$memcpy_inline(dst, src, $len, $is_volatile, $dst_align, $src_align);
+}
+
+<*
+ Copy memory from src to dst but correctly handle the possibility of overlapping ranges.
+
+ @param [&out] dst "The destination to copy to"
+ @param [&in] src "The source to copy from"
+ @param len "The number of bytes to copy"
+ @param $dst_align "the alignment of the destination if different from the default, 0 assumes the default"
+ @param $src_align "the alignment of the destination if different from the default, 0 assumes the default"
+ @param $is_volatile "True if this copy should be treated as volatile, i.e. it can't be optimized away."
+*>
+macro void move(void* dst, void* src, usz len, usz $dst_align = 0, usz $src_align = 0, bool $is_volatile = false)
+{
+	$$memmove(dst, src, len, $is_volatile, $dst_align, $src_align);
+}
+
+<*
+ Sets all memory in a region to that of the provided byte.
+
+ @param [&out] dst "The destination to copy to"
+ @param val "The value to copy into memory"
+ @param len "The number of bytes to copy"
+ @param $dst_align "the alignment of the destination if different from the default, 0 assumes the default"
+ @param $is_volatile "True if this copy should be treated as volatile, i.e. it can't be optimized away."
+
+ @ensure !len || (dst[0] == val && dst[len - 1] == val)
+*>
+macro void set(void* dst, char val, usz len, usz $dst_align = 0, bool $is_volatile = false)
+{
+	$$memset(dst, val, len, $is_volatile, $dst_align);
+}
+
+<*
+ Sets all memory in a region to that of the provided byte. Never calls OS memset.
+
+ @param [&out] dst "The destination to copy to"
+ @param val "The value to copy into memory"
+ @param $len "The number of bytes to copy"
+ @param $dst_align "the alignment of the destination if different from the default, 0 assumes the default"
+ @param $is_volatile "True if this copy should be treated as volatile, i.e. it can't be optimized away."
+
+ @ensure !$len || (dst[0] == val && dst[$len - 1] == val)
+*>
+macro void set_inline(void* dst, char val, usz $len, usz $dst_align = 0, bool $is_volatile = false)
+{
+	$$memset_inline(dst, val, $len, $is_volatile, $dst_align);
+}
+<*
+ Test if n elements are equal in a slice, pointed to by a pointer etc.
+
+ @require values::@inner_kind(a) == TypeKind.SLICE || values::@inner_kind(a) == TypeKind.POINTER
+ @require values::@inner_kind(b) == TypeKind.SLICE || values::@inner_kind(b) == TypeKind.POINTER
+ @require values::@inner_kind(a) != TypeKind.SLICE || len == -1
+ @require values::@inner_kind(a) != TypeKind.POINTER || len > -1
+ @require values::@assign_to(a, b) && values::@assign_to(b, a)
+*>
+macro bool equals(a, b, isz len = -1, usz $align = 0)
+{
+	$if !$align:
+		$align = $typeof(a[0]).alignof;
+	$endif
+	void* x @noinit;
+	void* y @noinit;
+	$if values::@inner_kind(a) == TypeKind.SLICE:
+		len = a.len;
+		if (len != b.len) return false;
+		x = a.ptr;
+		y = b.ptr;
+	$else
+		x = a;
+		y = b;
+		assert(len >= 0, "A zero or positive length must be given when comparing pointers.");
+	$endif
+
+	if (!len) return true;
+	var $Type;
+	$switch ($align)
+		$case 1:
+			$Type = char;
+		$case 2:
+			$Type = ushort;
+		$case 4:
+			$Type = uint;
+		$case 8:
+		$default:
+			$Type = ulong;
+	$endswitch
+	var $step = $Type.sizeof;
+	usz end = len / $step;
+	for (usz i = 0; i < end; i++)
+	{
+		if ((($Type*)x)[i] != (($Type*)y)[i]) return false;
+	}
+	usz last = len % $align;
+	for (usz i = len - last; i < len; i++)
+	{
+		if (((char*)x)[i] != ((char*)y)[i]) return false;
+	}
+	return true;
+}
+
+<*
+ Check if an allocation must be aligned given the type.
+
+ @return `true if the alignment of the type exceeds DEFAULT_MEM_ALIGNMENT.`
+*>
+macro bool type_alloc_must_be_aligned($Type)
+{
+	return $Type.alignof > DEFAULT_MEM_ALIGNMENT;
+}
+
+<*
+ Run with a specific allocator inside of the macro body.
+*>
+macro void @scoped(Allocator allocator; @body())
+{
+	Allocator old_allocator = allocator::thread_allocator;
+	allocator::thread_allocator = allocator;
+	defer allocator::thread_allocator = old_allocator;
+	@body();
+}
+
+<*
+ Run the tracking allocator in the scope, then
+ print out stats.
+
+ @param $enabled "Set to false to disable tracking"
+*>
+macro void @report_heap_allocs_in_scope($enabled = true; @body())
+{
+	$if $enabled:
+		TrackingAllocator tracker;
+		tracker.init(allocator::thread_allocator);
+		Allocator old_allocator = allocator::thread_allocator;
+		allocator::thread_allocator = &tracker;
+		defer
+		{
+			allocator::thread_allocator = old_allocator;
+			tracker.print_report();
+			tracker.free();
+		}
+	$endif
+	@body();
+}
+
+<*
+ Assert on memory leak in the scope of the macro body.
+
+ @param $report "Set to false to disable memory report"
+*>
+macro void @assert_leak($report = true; @body()) @builtin
+{
+	$if env::DEBUG_SYMBOLS || $feature(MEMORY_ASSERTS):
+		TrackingAllocator tracker;
+		tracker.init(allocator::thread_allocator);
+		Allocator old_allocator = allocator::thread_allocator;
+		allocator::thread_allocator = &tracker;
+		defer
+		{
+			allocator::thread_allocator = old_allocator;
+			defer tracker.free();
+			usz allocated = tracker.allocated();
+			if (allocated)
+			{
+				DString report = dstring::new();
+				defer report.free();
+				$if $report:
+					report.append_char('\n');
+					(void)tracker.fprint_report(&report);
+				$endif
+				assert(allocated == 0, "Memory leak detected"
+					" (%d bytes allocated).%s",
+					allocated, report.str_view());
+			}
+		}
+	$endif
+	@body();
+}
+
+<*
+ Allocate [size] bytes on the stack to use for allocation,
+ with the heap allocator as the backing allocator.
+
+ Release everything on scope exit.
+
+ @param $size `the size of the buffer`
+*>
+macro void @stack_mem(usz $size; @body(Allocator mem)) @builtin
+{
+	char[$size] buffer;
+	OnStackAllocator allocator;
+	allocator.init(&buffer, allocator::heap());
+	defer allocator.free();
+	@body(&allocator);
+}
+
+macro void @stack_pool(usz $size; @body) @builtin
+{
+	char[$size] buffer;
+	OnStackAllocator allocator;
+	allocator.init(&buffer, allocator::heap());
+	defer allocator.free();
+	mem::@scoped(&allocator)
+	{
+		@body();
+	};
+}
+
+struct TempState
+{
+	TempAllocator* old;
+	TempAllocator* current;
+	usz mark;
+}
+<*
+ Push the current temp allocator. A push must always be balanced with a pop using the current state.
+*>
+fn TempState temp_push(TempAllocator* other = null)
+{
+	TempAllocator* current = allocator::temp();
+	TempAllocator* old = current;
+	if (other == current)
+	{
+		current = allocator::temp_allocator_next();
+	}
+	return { old, current, current.used };
+}
+
+<*
+ Pop the current temp allocator. A pop must always be balanced with a push.
+*>
+fn void temp_pop(TempState old_state)
+{
+	assert(allocator::thread_temp_allocator == old_state.current, "Tried to pop temp allocators out of order.");
+	assert(old_state.current.used >= old_state.mark, "Tried to pop temp allocators out of order.");
+	old_state.current.reset(old_state.mark);
+	allocator::thread_temp_allocator = old_state.old;
+}
+
+<*
+ @require @is_empty_macro_slot(#other_temp) ||| $assignable(#other_temp, Allocator) "Must be an allocator"
+ *>
+macro void @pool(#other_temp = EMPTY_MACRO_SLOT; @body) @builtin
+{
+	TempAllocator* current = allocator::temp();
+	$if @is_valid_macro_slot(#other_temp):
+		TempAllocator* original = current;
+		if (current == #other_temp.ptr) current = allocator::temp_allocator_next();
+	$endif
+	usz mark = current.used;
+	defer
+	{
+		current.reset(mark);
+		$if @is_valid_macro_slot(#other_temp):
+			allocator::thread_temp_allocator = original;
+		$endif;
+	}
+	@body();
+}
+
+import libc;
+
+
+module std::core::mem @if(WASM_NOLIBC);
+import std::core::mem::allocator @public;
+SimpleHeapAllocator wasm_allocator @private;
+extern int __heap_base;
+
+fn void initialize_wasm_mem() @init(1024) @private
+{
+	allocator::wasm_memory.allocate_block(mem::DEFAULT_MEM_ALIGNMENT)!!; // Give us a valid null.
+	// Check if we need to move the heap.
+	uptr start = (uptr)&__heap_base;
+	if (start > mem::DEFAULT_MEM_ALIGNMENT) allocator::wasm_memory.use = start;
+	wasm_allocator.init(fn (x) => allocator::wasm_memory.allocate_block(x));
+	allocator::thread_allocator = &wasm_allocator;
+	allocator::temp_base_allocator = &wasm_allocator;
+	allocator::init_default_temp_allocators();
+}
+
+module std::core::mem;
+
+
+macro TrackingEnv* get_tracking_env()
+{
+	$if env::TRACK_MEMORY:
+		return &&TrackingEnv { $$FILE, $$FUNC, $$LINE };
+	$else
+		return null;
+	$endif
+}
+
+macro @clone(value) @builtin @nodiscard
+{
+	return allocator::clone(allocator::heap(), value);
+}
+
+macro @tclone(value) @builtin @nodiscard
+{
+	return temp_new($typeof(value), value);
+}
+
+fn void* malloc(usz size) @builtin @inline @nodiscard
+{
+	return allocator::malloc(allocator::heap(), size);
+}
+
+<*
+ Allocate using an aligned allocation. This is necessary for types with a default memory alignment
+ exceeding DEFAULT_MEM_ALIGNMENT. IMPORTANT! It must be freed using free_aligned.
+*>
+fn void* malloc_aligned(usz size, usz alignment) @builtin @inline @nodiscard
+{
+	return allocator::malloc_aligned(allocator::heap(), size, alignment)!!;
+}
+
+fn void* tmalloc(usz size, usz alignment = 0) @builtin @inline @nodiscard
+{
+	if (!size) return null;
+	return allocator::temp().acquire(size, NO_ZERO, alignment)!!;
+}
+
+<*
+ @require $vacount < 2 : "Too many arguments."
+ @require $vacount == 0 ||| $assignable($vaexpr[0], $Type) : "The second argument must be an initializer for the type"
+ @require $Type.alignof <= DEFAULT_MEM_ALIGNMENT : "Types with alignment exceeding the default must use 'alloc_aligned' instead"
+*>
+macro new($Type, ...) @nodiscard
+{
+	$if $vacount == 0:
+		return ($Type*)calloc($Type.sizeof);
+	$else
+		$Type* val = malloc($Type.sizeof);
+		*val = $vaexpr[0];
+		return val;
+	$endif
+}
+
+<*
+ @require $vacount < 2 : "Too many arguments."
+ @require $vacount == 0 ||| $assignable($vaexpr[0], $Type) : "The second argument must be an initializer for the type"
+ @require $Type.alignof <= DEFAULT_MEM_ALIGNMENT : "Types with alignment exceeding the default must use 'alloc_aligned' instead"
+*>
+macro new_with_padding($Type, usz padding, ...) @nodiscard
+{
+	$if $vacount == 0:
+		return ($Type*)calloc($Type.sizeof + padding);
+	$else
+		$Type* val = malloc($Type.sizeof + padding);
+		*val = $vaexpr[0];
+		return val;
+	$endif
+}
+
+<*
+ Allocate using an aligned allocation. This is necessary for types with a default memory alignment
+ exceeding DEFAULT_MEM_ALIGNMENT. IMPORTANT! It must be freed using free_aligned.
+ @require $vacount < 2 : "Too many arguments."
+ @require $vacount == 0 ||| $assignable($vaexpr[0], $Type) : "The second argument must be an initializer for the type"
+*>
+macro new_aligned($Type, ...) @nodiscard
+{
+	$if $vacount == 0:
+		return ($Type*)calloc_aligned($Type.sizeof, $Type.alignof);
+	$else
+		$Type* val = malloc_aligned($Type.sizeof, $Type.alignof);
+		*val = $vaexpr[0];
+		return val;
+	$endif
+}
+
+<*
+ @require $Type.alignof <= DEFAULT_MEM_ALIGNMENT : "Types with alignment exceeding the default must use 'alloc_aligned' instead"
+*>
+macro alloc($Type) @nodiscard
+{
+	return ($Type*)malloc($Type.sizeof);
+}
+
+<*
+ @require $Type.alignof <= DEFAULT_MEM_ALIGNMENT : "Types with alignment exceeding the default must use 'alloc_aligned' instead"
+*>
+macro alloc_with_padding($Type, usz padding) @nodiscard
+{
+	return ($Type*)malloc($Type.sizeof + padding);
+}
+
+<*
+ Allocate using an aligned allocation. This is necessary for types with a default memory alignment
+ exceeding DEFAULT_MEM_ALIGNMENT. IMPORTANT! It must be freed using free_aligned.
+*>
+macro alloc_aligned($Type) @nodiscard
+{
+	return ($Type*)malloc_aligned($Type.sizeof, $Type.alignof);
+}
+
+<*
+ @require $vacount < 2 : "Too many arguments."
+ @require $vacount == 0 ||| $assignable($vaexpr[0], $Type) : "The second argument must be an initializer for the type"
+*>
+macro temp_new($Type, ...) @nodiscard
+{
+	$if $vacount == 0:
+		return ($Type*)tcalloc($Type.sizeof) @inline;
+	$else
+		$Type* val = tmalloc($Type.sizeof) @inline;
+		*val = $vaexpr[0];
+		return val;
+	$endif
+}
+
+<*
+ @require $vacount < 2 : "Too many arguments."
+ @require $vacount == 0 ||| $assignable($vaexpr[0], $Type) : "The second argument must be an initializer for the type"
+*>
+macro temp_new_with_padding($Type, usz padding, ...) @nodiscard
+{
+	$if $vacount == 0:
+		return ($Type*)tcalloc($Type.sizeof + padding) @inline;
+	$else
+		$Type* val = tmalloc($Type.sizeof + padding) @inline;
+		*val = $vaexpr[0];
+		return val;
+	$endif
+}
+
+macro temp_alloc($Type) @nodiscard
+{
+	return tmalloc($Type.sizeof);
+}
+
+macro temp_alloc_with_padding($Type, usz padding) @nodiscard
+{
+	return tmalloc($Type.sizeof + padding);
+}
+
+<*
+ @require $Type.alignof <= DEFAULT_MEM_ALIGNMENT : "Types with alignment exceeding the default must use 'new_array_aligned' instead"
+*>
+macro new_array($Type, usz elements) @nodiscard
+{
+	return allocator::new_array(allocator::heap(), $Type, elements);
+}
+
+<*
+ Allocate using an aligned allocation. This is necessary for types with a default memory alignment
+ exceeding DEFAULT_MEM_ALIGNMENT. IMPORTANT! It must be freed using free_aligned.
+*>
+macro new_array_aligned($Type, usz elements) @nodiscard
+{
+	return allocator::new_array_aligned(allocator::heap(), $Type, elements);
+}
+
+<*
+ @require $Type.alignof <= DEFAULT_MEM_ALIGNMENT : "Types with alignment exceeding the default must use 'alloc_array_aligned' instead"
+*>
+macro alloc_array($Type, usz elements) @nodiscard
+{
+	return allocator::alloc_array(allocator::heap(), $Type, elements);
+}
+
+<*
+ Allocate using an aligned allocation. This is necessary for types with a default memory alignment
+ exceeding DEFAULT_MEM_ALIGNMENT. IMPORTANT! It must be freed using free_aligned.
+*>
+macro alloc_array_aligned($Type, usz elements) @nodiscard
+{
+	return allocator::alloc_array_aligned(allocator::heap(), $Type, elements);
+}
+
+macro temp_alloc_array($Type, usz elements) @nodiscard
+{
+	return (($Type*)tmalloc($Type.sizeof * elements, $Type.alignof))[:elements];
+}
+
+macro temp_new_array($Type, usz elements) @nodiscard
+{
+	return (($Type*)tcalloc($Type.sizeof * elements, $Type.alignof))[:elements];
+}
+
+fn void* calloc(usz size) @builtin @inline @nodiscard
+{
+	return allocator::calloc(allocator::heap(), size);
+}
+
+<*
+ Allocate using an aligned allocation. This is necessary for types with a default memory alignment
+ exceeding DEFAULT_MEM_ALIGNMENT. IMPORTANT! It must be freed using free_aligned.
+*>
+fn void* calloc_aligned(usz size, usz alignment) @builtin @inline @nodiscard
+{
+	return allocator::calloc_aligned(allocator::heap(), size, alignment)!!;
+}
+
+fn void* tcalloc(usz size, usz alignment = 0) @builtin @inline @nodiscard
+{
+	if (!size) return null;
+	return allocator::temp().acquire(size, ZERO, alignment)!!;
+}
+
+fn void* realloc(void *ptr, usz new_size) @builtin @inline @nodiscard
+{
+	return allocator::realloc(allocator::heap(), ptr, new_size);
+}
+
+fn void* realloc_aligned(void *ptr, usz new_size, usz alignment) @builtin @inline @nodiscard
+{
+	return allocator::realloc_aligned(allocator::heap(), ptr, new_size, alignment)!!;
+}
+
+fn void free(void* ptr) @builtin @inline
+{
+	return allocator::free(allocator::heap(), ptr);
+}
+
+fn void free_aligned(void* ptr) @builtin @inline
+{
+	return allocator::free_aligned(allocator::heap(), ptr);
+}
+
+fn void* trealloc(void* ptr, usz size, usz alignment = mem::DEFAULT_MEM_ALIGNMENT) @builtin @inline @nodiscard
+{
+	if (!size) return null;
+	if (!ptr) return tmalloc(size, alignment);
+	return allocator::temp().resize(ptr, size, alignment)!!;
+}
+
+module std::core::mem @if(env::NO_LIBC);
+
+fn CInt __memcmp(void* s1, void* s2, usz n) @weak @export("memcmp")
+{
+	char* p1 = s1;
+	char* p2 = s2;
+	for (usz i = 0; i < n; i++, p1++, p2++)
+	{
+		char c1 = *p1;
+		char c2 = *p2;
+		if (c1 < c2) return -1;
+		if (c1 > c2) return 1;
+	}
+	return 0;
+}
+
+fn void* __memset(void* str, CInt c, usz n) @weak @export("memset")
+{
+	char* p = str;
+	char cc = (char)c;
+	for (usz i = 0; i < n; i++, p++)
+	{
+		*p = cc;
+	}
+	return str;
+}
+
+fn void* __memcpy(void* dst, void* src, usz n) @weak @export("memcpy")
+{
+	char* d = dst;
+	char* s = src;
+	for (usz i = 0; i < n; i++, d++, s++)
+	{
+		*d = *s;
+	}
+	return dst;
+}

lib/std/core/mem_allocator.c3

@@ -0,0 +1,455 @@
+module std::core::mem::allocator;
+
+const DEFAULT_SIZE_PREFIX = usz.sizeof;
+const DEFAULT_SIZE_PREFIX_ALIGNMENT = usz.alignof;
+
+struct TrackingEnv
+{
+	String file;
+	String function;
+	uint line;
+}
+
+enum AllocInitType
+{
+	NO_ZERO,
+	ZERO
+}
+
+interface Allocator
+{
+	fn void reset(usz mark) @optional;
+	fn usz mark() @optional;
+	<*
+	 @require !alignment || math::is_power_of_2(alignment)
+	 @require alignment <= mem::MAX_MEMORY_ALIGNMENT `alignment too big`
+	 @require size > 0
+	*>
+	fn void*! acquire(usz size, AllocInitType init_type, usz alignment = 0);
+	<*
+	 @require !alignment || math::is_power_of_2(alignment)
+	 @require alignment <= mem::MAX_MEMORY_ALIGNMENT `alignment too big`
+	 @require ptr != null
+	 @require new_size > 0
+	*>
+	fn void*! resize(void* ptr, usz new_size, usz alignment = 0);
+	<*
+	 @require ptr != null
+	*>
+	fn void release(void* ptr, bool aligned);
+}
+
+def MemoryAllocFn = fn char[]!(usz);
+
+fault AllocationFailure
+{
+	OUT_OF_MEMORY,
+	CHUNK_TOO_LARGE,
+}
+
+fn usz alignment_for_allocation(usz alignment) @inline @private
+{
+	return alignment < mem::DEFAULT_MEM_ALIGNMENT ? mem::DEFAULT_MEM_ALIGNMENT : alignment;
+}
+
+macro void* malloc(Allocator allocator, usz size) @nodiscard
+{
+	return malloc_try(allocator, size)!!;
+}
+
+macro void*! malloc_try(Allocator allocator, usz size) @nodiscard
+{
+	if (!size) return null;
+	$if env::TESTING:
+		char* data = allocator.acquire(size, NO_ZERO)!;
+		mem::set(data, 0xAA, size, mem::DEFAULT_MEM_ALIGNMENT);
+		return data;
+	$else
+		return allocator.acquire(size, NO_ZERO);
+	$endif
+}
+
+macro void* calloc(Allocator allocator, usz size) @nodiscard
+{
+	return calloc_try(allocator, size)!!;
+}
+
+macro void*! calloc_try(Allocator allocator, usz size) @nodiscard
+{
+	if (!size) return null;
+	return allocator.acquire(size, ZERO);
+}
+
+macro void* realloc(Allocator allocator, void* ptr, usz new_size) @nodiscard
+{
+	return realloc_try(allocator, ptr, new_size)!!;
+}
+
+macro void*! realloc_try(Allocator allocator, void* ptr, usz new_size) @nodiscard
+{
+	if (!new_size)
+	{
+		free(allocator, ptr);
+		return null;
+	}
+	if (!ptr) return allocator.acquire(new_size, NO_ZERO);
+	return allocator.resize(ptr, new_size);
+}
+
+macro void free(Allocator allocator, void* ptr)
+{
+	if (!ptr) return;
+	$if env::TESTING:
+		((char*)ptr)[0] = 0xBA;
+	$endif
+	allocator.release(ptr, false);
+}
+
+macro void*! malloc_aligned(Allocator allocator, usz size, usz alignment) @nodiscard
+{
+	if (!size) return null;
+	$if env::TESTING:
+		char* data = allocator.acquire(size, NO_ZERO, alignment)!;
+		mem::set(data, 0xAA, size, mem::DEFAULT_MEM_ALIGNMENT);
+		return data;
+	$else
+		return allocator.acquire(size, NO_ZERO, alignment);
+	$endif
+}
+
+macro void*! calloc_aligned(Allocator allocator, usz size, usz alignment) @nodiscard
+{
+	if (!size) return null;
+	return allocator.acquire(size, ZERO, alignment);
+}
+
+macro void*! realloc_aligned(Allocator allocator, void* ptr, usz new_size, usz alignment) @nodiscard
+{
+	if (!new_size)
+	{
+		free_aligned(allocator, ptr);
+		return null;
+	}
+	if (!ptr)
+	{
+		return malloc_aligned(allocator, new_size, alignment);
+	}
+	return allocator.resize(ptr, new_size, alignment);
+}
+
+macro void free_aligned(Allocator allocator, void* ptr)
+{
+	if (!ptr) return;
+	$if env::TESTING:
+	((char*)ptr)[0] = 0xBA;
+	$endif
+	allocator.release(ptr, aligned: true);
+}
+
+<*
+ @require $Type.alignof <= mem::DEFAULT_MEM_ALIGNMENT : "Types with alignment exceeding the default must use 'new_aligned' instead"
+ @require $vacount < 2 : "Too many arguments."
+ @require $vacount == 0 ||| $assignable($vaexpr[0], $Type) : "The second argument must be an initializer for the type"
+*>
+macro new(Allocator allocator, $Type, ...) @nodiscard
+{
+	$if $vacount == 0:
+		return ($Type*)calloc(allocator, $Type.sizeof);
+	$else
+		$Type* val = malloc(allocator, $Type.sizeof);
+		*val = $vaexpr[0];
+		return val;
+	$endif
+}
+
+<*
+ @require $Type.alignof <= mem::DEFAULT_MEM_ALIGNMENT : "Types with alignment exceeding the default must use 'new_aligned' instead"
+ @require $vacount < 2 : "Too many arguments."
+ @require $vacount == 0 ||| $assignable($vaexpr[0], $Type) : "The second argument must be an initializer for the type"
+*>
+macro new_try(Allocator allocator, $Type, ...) @nodiscard
+{
+	$if $vacount == 0:
+		return ($Type*)calloc_try(allocator, $Type.sizeof);
+	$else
+		$Type* val = malloc_try(allocator, $Type.sizeof)!;
+		*val = $vaexpr[0];
+		return val;
+	$endif
+}
+
+<*
+ Allocate using an aligned allocation. This is necessary for types with a default memory alignment
+ exceeding DEFAULT_MEM_ALIGNMENT. IMPORTANT! It must be freed using free_aligned.
+ @require $vacount < 2 : "Too many arguments."
+ @require $vacount == 0 ||| $assignable($vaexpr[0], $Type) : "The second argument must be an initializer for the type"
+*>
+macro new_aligned(Allocator allocator, $Type, ...) @nodiscard
+{
+	$if $vacount == 0:
+		return ($Type*)calloc_aligned(allocator, $Type.sizeof, $Type.alignof);
+	$else
+		$Type* val = malloc_aligned(allocator, $Type.sizeof, $Type.alignof)!;
+		*val = $vaexpr[0];
+		return val;
+	$endif
+}
+
+<*
+ @require $Type.alignof <= mem::DEFAULT_MEM_ALIGNMENT
+*>
+macro new_with_padding(Allocator allocator, $Type, usz padding) @nodiscard
+{
+	return ($Type*)calloc_try(allocator, $Type.sizeof + padding);
+}
+
+<*
+ @require $Type.alignof <= mem::DEFAULT_MEM_ALIGNMENT : "Types with alignment exceeding the default must use 'alloc_aligned' instead"
+*>
+macro alloc(Allocator allocator, $Type) @nodiscard
+{
+	return ($Type*)malloc(allocator, $Type.sizeof);
+}
+
+<*
+ @require $Type.alignof <= mem::DEFAULT_MEM_ALIGNMENT : "Types with alignment exceeding the default must use 'alloc_aligned' instead"
+*>
+macro alloc_try(Allocator allocator, $Type) @nodiscard
+{
+	return ($Type*)malloc_try(allocator, $Type.sizeof);
+}
+
+<*
+ Allocate using an aligned allocation. This is necessary for types with a default memory alignment
+ exceeding DEFAULT_MEM_ALIGNMENT. IMPORTANT! It must be freed using free_aligned.
+*>
+macro alloc_aligned(Allocator allocator, $Type) @nodiscard
+{
+	return ($Type*)malloc_aligned(allocator, $Type.sizeof, $Type.alignof);
+}
+
+<*
+ @require $Type.alignof <= mem::DEFAULT_MEM_ALIGNMENT
+*>
+macro alloc_with_padding(Allocator allocator, $Type, usz padding) @nodiscard
+{
+	return ($Type*)malloc_try(allocator, $Type.sizeof + padding);
+}
+
+<*
+ @require $Type.alignof <= mem::DEFAULT_MEM_ALIGNMENT : "Types with alignment exceeding the default must use 'new_array_aligned' instead"
+*>
+macro new_array(Allocator allocator, $Type, usz elements) @nodiscard
+{
+	return new_array_try(allocator, $Type, elements)!!;
+}
+
+<*
+ @require $Type.alignof <= mem::DEFAULT_MEM_ALIGNMENT : "Types with alignment exceeding the default must use 'new_array_aligned' instead"
+*>
+macro new_array_try(Allocator allocator, $Type, usz elements) @nodiscard
+{
+	return (($Type*)calloc_try(allocator, $Type.sizeof * elements))[:elements];
+}
+
+<*
+ Allocate using an aligned allocation. This is necessary for types with a default memory alignment
+ exceeding DEFAULT_MEM_ALIGNMENT. IMPORTANT! It must be freed using free_aligned.
+*>
+macro new_array_aligned(Allocator allocator, $Type, usz elements) @nodiscard
+{
+	return (($Type*)calloc_aligned(allocator, $Type.sizeof * elements, $Type.alignof))[:elements]!!;
+}
+
+<*
+ @require $Type.alignof <= mem::DEFAULT_MEM_ALIGNMENT : "Types with alignment exceeding the default must use 'alloc_array_aligned' instead"
+*>
+macro alloc_array(Allocator allocator, $Type, usz elements) @nodiscard
+{
+	return alloc_array_try(allocator, $Type, elements)!!;
+}
+
+<*
+ Allocate using an aligned allocation. This is necessary for types with a default memory alignment
+ exceeding DEFAULT_MEM_ALIGNMENT. IMPORTANT! It must be freed using free_aligned.
+*>
+macro alloc_array_aligned(Allocator allocator, $Type, usz elements) @nodiscard
+{
+	return (($Type*)malloc_aligned(allocator, $Type.sizeof * elements, $Type.alignof))[:elements]!!;
+}
+
+<*
+ @require $Type.alignof <= mem::DEFAULT_MEM_ALIGNMENT : "Types with alignment exceeding the default must use 'alloc_array_aligned' instead"
+*>
+macro alloc_array_try(Allocator allocator, $Type, usz elements) @nodiscard
+{
+	return (($Type*)malloc_try(allocator, $Type.sizeof * elements))[:elements];
+}
+
+macro clone(Allocator allocator, value) @nodiscard
+{
+	return new(allocator, $typeof(value), value);
+}
+
+fn any clone_any(Allocator allocator, any value) @nodiscard
+{
+	usz size = value.type.sizeof;
+	void* data = malloc(allocator, size);
+	mem::copy(data, value.ptr, size);
+	return any_make(data, value.type);
+}
+
+
+<*
+ @require bytes > 0
+ @require alignment > 0
+ @require bytes <= isz.max
+*>
+macro void*! @aligned_alloc(#alloc_fn, usz bytes, usz alignment)
+{
+	if (alignment < void*.alignof) alignment = void*.alignof;
+	usz header = AlignedBlock.sizeof + alignment;
+	usz alignsize = bytes + header;
+	$if @typekind(#alloc_fn(bytes)) == OPTIONAL:
+		void* data = #alloc_fn(alignsize)!;
+	$else
+		void* data = #alloc_fn(alignsize);
+	$endif
+	void* mem = mem::aligned_pointer(data + AlignedBlock.sizeof, alignment);
+	AlignedBlock* desc = (AlignedBlock*)mem - 1;
+	assert(mem > data);
+	*desc = { bytes, data };
+	return mem;
+}
+
+struct AlignedBlock
+{
+	usz len;
+	void* start;
+}
+
+macro void! @aligned_free(#free_fn, void* old_pointer)
+{
+	AlignedBlock* desc = (AlignedBlock*)old_pointer - 1;
+	$if @typekind(#free_fn(desc.start)) == OPTIONAL:
+	#free_fn(desc.start)!;
+	$else
+	#free_fn(desc.start);
+	$endif
+}
+
+<*
+ @require bytes > 0
+ @require alignment > 0
+*>
+macro void*! @aligned_realloc(#calloc_fn, #free_fn, void* old_pointer, usz bytes, usz alignment)
+{
+	AlignedBlock* desc = (AlignedBlock*)old_pointer - 1;
+	void* data_start = desc.start;
+	void* new_data = @aligned_alloc(#calloc_fn, bytes, alignment)!;
+	mem::copy(new_data, old_pointer, desc.len < bytes ? desc.len : bytes, 1, 1);
+	$if @typekind(#free_fn(data_start)) == OPTIONAL:
+	#free_fn(data_start)!;
+	$else
+	#free_fn(data_start);
+	$endif
+	return new_data;
+}
+
+
+// All allocators
+
+
+def mem = thread_allocator @builtin;
+tlocal Allocator thread_allocator @private = base_allocator();
+Allocator temp_base_allocator @private = base_allocator();
+
+tlocal TempAllocator* thread_temp_allocator @private = null;
+tlocal TempAllocator*[2] temp_allocator_pair @private;
+
+macro Allocator base_allocator() @private
+{
+	$if env::LIBC:
+		return &allocator::LIBC_ALLOCATOR;
+	$else
+		return &allocator::NULL_ALLOCATOR;
+	$endif
+}
+
+macro TempAllocator* create_default_sized_temp_allocator(Allocator allocator) @local
+{
+	$switch (env::MEMORY_ENV)
+	$case NORMAL:
+		return new_temp_allocator(1024 * 256, allocator)!!;
+	$case SMALL:
+		return new_temp_allocator(1024 * 16, allocator)!!;
+	$case TINY:
+		return new_temp_allocator(1024 * 2, allocator)!!;
+	$case NONE:
+		unreachable("Temp allocator must explicitly created when memory-env is set to 'none'.");
+	$endswitch
+}
+
+macro Allocator heap() => thread_allocator;
+
+macro TempAllocator* temp()
+{
+	if (!thread_temp_allocator)
+	{
+		init_default_temp_allocators();
+	}
+	return thread_temp_allocator;
+}
+
+fn void init_default_temp_allocators() @private
+{
+	temp_allocator_pair[0] = create_default_sized_temp_allocator(temp_base_allocator);
+	temp_allocator_pair[1] = create_default_sized_temp_allocator(temp_base_allocator);
+	thread_temp_allocator = temp_allocator_pair[0];
+}
+
+fn void destroy_temp_allocators_after_exit() @finalizer(65535) @local @if(env::LIBC)
+{
+	destroy_temp_allocators();
+}
+
+<*
+ Call this to destroy any memory used by the temp allocators. This will invalidate all temp memory.
+*>
+fn void destroy_temp_allocators()
+{
+	if (!thread_temp_allocator) return;
+	temp_allocator_pair[0].destroy();
+	temp_allocator_pair[1].destroy();
+	temp_allocator_pair[..] = null;
+	thread_temp_allocator = null;
+}
+
+fn TempAllocator* temp_allocator_next() @private
+{
+	if (!thread_temp_allocator)
+	{
+		init_default_temp_allocators();
+		return thread_temp_allocator;
+	}
+	usz index = thread_temp_allocator == temp_allocator_pair[0] ? 1 : 0;
+	return thread_temp_allocator = temp_allocator_pair[index];
+}
+
+const NullAllocator NULL_ALLOCATOR = {};
+distinct NullAllocator (Allocator) = uptr;
+
+fn void*! NullAllocator.acquire(&self, usz bytes, AllocInitType init_type, usz alignment) @dynamic
+{
+	return AllocationFailure.OUT_OF_MEMORY?;
+}
+
+fn void*! NullAllocator.resize(&self, void* old_ptr, usz new_bytes, usz alignment) @dynamic
+{
+	return AllocationFailure.OUT_OF_MEMORY?;
+}
+
+fn void NullAllocator.release(&self, void* old_ptr, bool aligned) @dynamic
+{
+}
+

lib/std/core/os/wasm_memory.c3

@@ -0,0 +1,32 @@
+module std::core::mem::allocator;
+
+
+const usz WASM_BLOCK_SIZE = 65536;
+
+WasmMemory wasm_memory;
+
+struct WasmMemory
+{
+	usz allocation;
+	uptr use;
+}
+
+fn char[]! WasmMemory.allocate_block(&self, usz bytes)
+{
+	if (!self.allocation)
+	{
+		self.allocation = $$wasm_memory_size(0) * WASM_BLOCK_SIZE;
+	}
+	isz bytes_required = bytes + self.use - self.allocation;
+	if (bytes_required <= 0)
+	{
+		defer self.use += bytes;
+		return ((char*)self.use)[:bytes];
+	}
+
+	usz blocks_required = (bytes_required + WASM_BLOCK_SIZE + 1) / WASM_BLOCK_SIZE;
+	if ($$wasm_memory_grow(0, blocks_required) == -1) return AllocationFailure.OUT_OF_MEMORY?;
+	self.allocation = $$wasm_memory_size(0) * WASM_BLOCK_SIZE;
+	defer self.use += bytes;
+	return ((char*)self.use)[:bytes];
+}

lib/std/core/private/cpu_detect.c3

@@ -0,0 +1,268 @@
+module std::core::cpudetect @if(env::X86 || env::X86_64);
+
+struct CpuId
+{
+	uint eax, ebx, ecx, edx;
+}
+fn CpuId x86_cpuid(uint eax, uint ecx = 0)
+{
+	int edx;
+	int ebx;
+	asm
+	{
+		movl $eax, eax;
+		movl $ecx, ecx;
+		cpuid;
+		movl eax, $eax;
+		movl ebx, $ebx;
+		movl ecx, $ecx;
+		movl edx, $edx;
+	}
+	return { eax, ebx, ecx, edx };
+}
+
+enum X86Feature
+{
+	ADX,
+	AES,
+	AMX_AVX512,
+	AMX_FP8,
+	AMX_MOVRS,
+	AMX_TF32,
+	AMX_TRANSPOSE,
+	AMX_BF16,
+	AMX_COMPLEX,
+	AMX_FP16,
+	AMX_INT8,
+	AMX_TILE,
+	APXF,
+	AVX,
+	AVX10_1_256,
+	AVX10_1_512,
+	AVX10_2_256,
+	AVX10_2_512,
+	AVX2,
+	AVX5124FMAPS,
+	AVX5124VNNIW,
+	AVX512BF16,
+	AVX512BITALG,
+	AVX512BW,
+	AVX512CD,
+	AVX512DQ,
+	AVX512ER,
+	AVX512F,
+	AVX512FP16,
+	AVX512IFMA,
+	AVX512PF,
+	AVX512VBMI,
+	AVX512VBMI2,
+	AVX512VL,
+	AVX512VNNI,
+	AVX512VP2INTERSECT,
+	AVX512VPOPCNTDQ,
+	AVXIFMA,
+	AVXNECONVERT,
+	AVXVNNI,
+	AVXVNNIINT16,
+	AVXVNNIINT8,
+	BMI,
+	BMI2,
+	CLDEMOTE,
+	CLFLUSHOPT,
+	CLWB,
+	CLZERO,
+	CMOV,
+	CMPCCXADD,
+	CMPXCHG16B,
+	CX8,
+	ENQCMD,
+	F16C,
+	FMA,
+	FMA4,
+	FSGSBASE,
+	FXSR,
+	GFNI,
+	HRESET,
+	INVPCID,
+	KL,
+	LWP,
+	LZCNT,
+	MMX,
+	MOVBE,
+	MOVDIR64B,
+	MOVDIRI,
+	MOVRS,
+	MWAITX,
+	PCLMUL,
+	PCONFIG,
+	PKU,
+	POPCNT,
+	PREFETCHI,
+	PREFETCHWT1,
+	PRFCHW,
+	PTWRITE,
+	RAOINT,
+	RDPID,
+	RDPRU,
+	RDRND,
+	RDSEED,
+	RTM,
+	SAHF,
+	SERIALIZE,
+	SGX,
+	SHA,
+	SHA512,
+	SHSTK,
+	SM3,
+	SM4,
+	SSE,
+	SSE2,
+	SSE3,
+	SSE4_1,
+	SSE4_2,
+	SSE4_A,
+	SSSE3,
+	TBM,
+	TSXLDTRK,
+	UINTR,
+	USERMSR,
+	VAES,
+	VPCLMULQDQ,
+	WAITPKG,
+	WBNOINVD,
+	WIDEKL,
+	X87,
+	XOP,
+	XSAVE,
+	XSAVEC,
+	XSAVEOPT,
+	XSAVES,
+}
+
+uint128 x86_features;
+
+fn void add_feature_if_bit(X86Feature feature, uint register, int bit)
+{
+	if (register & 1U << bit) x86_features |= 1u128 << feature.ordinal;
+}
+
+fn void x86_initialize_cpu_features()
+{
+	uint max_level = x86_cpuid(0).eax;
+	CpuId feat = x86_cpuid(1);
+	CpuId leaf7 = max_level >= 8 ? x86_cpuid(7) : {};
+	CpuId leaf7s1 = leaf7.eax >= 1 ? x86_cpuid(7, 1) : {};
+	CpuId ext1 = x86_cpuid(0x80000000).eax >= 0x80000001 ? x86_cpuid(0x80000001) : {};
+	CpuId ext8 = x86_cpuid(0x80000000).eax >= 0x80000008 ? x86_cpuid(0x80000008) : {};
+	CpuId leaf_d = max_level >= 0xd ? x86_cpuid(0xd, 0x1) : {};
+	CpuId leaf_14 = max_level >= 0x14 ? x86_cpuid(0x14) : {};
+	CpuId leaf_19 = max_level >= 0x19 ? x86_cpuid(0x19) : {};
+	CpuId leaf_24 = max_level >= 0x24 ? x86_cpuid(0x24) : {};
+	add_feature_if_bit(ADX, leaf7.ebx, 19);
+	add_feature_if_bit(AES, feat.ecx, 25);
+	add_feature_if_bit(AMX_BF16, leaf7.edx, 22);
+	add_feature_if_bit(AMX_COMPLEX, leaf7s1.edx, 8);
+	add_feature_if_bit(AMX_FP16, leaf7s1.eax, 21);
+	add_feature_if_bit(AMX_INT8, leaf7.edx, 25);
+	add_feature_if_bit(AMX_TILE, leaf7.edx, 24);
+	add_feature_if_bit(APXF, leaf7s1.edx, 21);
+	add_feature_if_bit(AVX, feat.ecx, 28);
+	add_feature_if_bit(AVX10_1_256, leaf7s1.edx, 19);
+	add_feature_if_bit(AVX10_1_512, leaf_24.ebx, 18);
+	add_feature_if_bit(AVX2, leaf7.ebx, 5);
+	add_feature_if_bit(AVX5124FMAPS, leaf7.edx, 3);
+	add_feature_if_bit(AVX5124VNNIW, leaf7.edx, 2);
+	add_feature_if_bit(AVX512BF16, leaf7s1.eax, 5);
+	add_feature_if_bit(AVX512BITALG, leaf7.ecx, 12);
+	add_feature_if_bit(AVX512BW, leaf7.ebx, 30);
+	add_feature_if_bit(AVX512CD, leaf7.ebx, 28);
+	add_feature_if_bit(AVX512DQ, leaf7.ebx, 17);
+	add_feature_if_bit(AVX512ER, leaf7.ebx, 27);
+	add_feature_if_bit(AVX512F, leaf7.ebx, 16);
+	add_feature_if_bit(AVX512FP16, leaf7.edx, 23);
+	add_feature_if_bit(AVX512IFMA, leaf7.ebx, 21);
+	add_feature_if_bit(AVX512PF, leaf7.ebx, 26);
+	add_feature_if_bit(AVX512VBMI, leaf7.ecx, 1);
+	add_feature_if_bit(AVX512VBMI2, leaf7.ecx, 6);
+	add_feature_if_bit(AVX512VL, leaf7.ebx, 31);
+	add_feature_if_bit(AVX512VNNI, leaf7.ecx, 11);
+	add_feature_if_bit(AVX512VP2INTERSECT, leaf7.edx, 8);
+	add_feature_if_bit(AVX512VPOPCNTDQ, leaf7.ecx, 14);
+	add_feature_if_bit(AVXIFMA, leaf7s1.eax, 23);
+	add_feature_if_bit(AVXNECONVERT, leaf7s1.edx, 5);
+	add_feature_if_bit(AVXVNNI, leaf7s1.eax, 4);
+	add_feature_if_bit(AVXVNNIINT16, leaf7s1.edx, 10);
+	add_feature_if_bit(AVXVNNIINT8, leaf7s1.edx, 4);
+	add_feature_if_bit(BMI, leaf7.ebx, 3);
+	add_feature_if_bit(BMI2, leaf7.ebx, 8);
+	add_feature_if_bit(CLDEMOTE, leaf7.ecx, 25);
+	add_feature_if_bit(CLFLUSHOPT, leaf7.ebx, 23);
+	add_feature_if_bit(CLWB, leaf7.ebx, 24);
+	add_feature_if_bit(CLZERO, ext8.ecx, 0);
+	add_feature_if_bit(CMOV, feat.edx, 15);
+	add_feature_if_bit(CMPCCXADD, leaf7s1.eax, 7);
+	add_feature_if_bit(CMPXCHG16B, feat.ecx, 12);
+	add_feature_if_bit(CX8, feat.edx, 8);
+	add_feature_if_bit(ENQCMD, leaf7.ecx, 29);
+	add_feature_if_bit(F16C, feat.ecx, 29);
+	add_feature_if_bit(FMA, feat.ecx, 12);
+	add_feature_if_bit(FMA4, ext1.ecx, 16);
+	add_feature_if_bit(FSGSBASE, leaf7.ebx, 0);
+	add_feature_if_bit(FXSR, feat.edx, 24);
+	add_feature_if_bit(GFNI, leaf7.ecx, 8);
+	add_feature_if_bit(HRESET, leaf7s1.eax, 22);
+	add_feature_if_bit(INVPCID, leaf7.ebx, 10);
+	add_feature_if_bit(KL, leaf7.ecx, 23);
+	add_feature_if_bit(LWP, ext1.ecx, 15);
+	add_feature_if_bit(LZCNT, ext1.ecx, 5);
+	add_feature_if_bit(MMX, feat.edx, 23);
+	add_feature_if_bit(MOVBE, feat.ecx, 22);
+	add_feature_if_bit(MOVDIR64B, leaf7.ecx, 28);
+	add_feature_if_bit(MOVDIRI, leaf7.ecx, 27);
+	add_feature_if_bit(MWAITX, ext1.ecx, 29);
+	add_feature_if_bit(PCLMUL, feat.ecx, 1);
+	add_feature_if_bit(PCONFIG, leaf7.edx, 18);
+	add_feature_if_bit(PKU, leaf7.ecx, 4);
+	add_feature_if_bit(POPCNT, feat.ecx, 23);
+	add_feature_if_bit(PREFETCHI, leaf7s1.edx, 14);
+	add_feature_if_bit(PREFETCHWT1, leaf7.ecx, 0);
+	add_feature_if_bit(PRFCHW, ext1.ecx, 8);
+	add_feature_if_bit(PTWRITE, leaf_14.ebx, 4);
+	add_feature_if_bit(RAOINT, leaf7s1.eax, 3);
+	add_feature_if_bit(RDPID, leaf7.ecx, 22);
+	add_feature_if_bit(RDPRU, ext8.ecx, 4);
+	add_feature_if_bit(RDRND, feat.ecx, 30);
+	add_feature_if_bit(RDSEED, leaf7.ebx, 18);
+	add_feature_if_bit(RTM, leaf7.ebx, 11);
+	add_feature_if_bit(SAHF, ext1.ecx, 0);
+	add_feature_if_bit(SERIALIZE, leaf7.edx, 14);
+	add_feature_if_bit(SGX, leaf7.ebx, 2);
+	add_feature_if_bit(SHA, leaf7.ebx, 29);
+	add_feature_if_bit(SHA512, leaf7s1.eax, 0);
+	add_feature_if_bit(SHSTK, leaf7.ecx, 7);
+	add_feature_if_bit(SM3, leaf7s1.eax, 1);
+	add_feature_if_bit(SM4, leaf7s1.eax, 2);
+	add_feature_if_bit(SSE, feat.edx, 25);
+	add_feature_if_bit(SSE2, feat.edx, 26);
+	add_feature_if_bit(SSE3, feat.ecx, 0);
+	add_feature_if_bit(SSE4_1, feat.ecx, 19);
+	add_feature_if_bit(SSE4_2, feat.ecx, 20);
+	add_feature_if_bit(SSE4_A, ext1.ecx, 6);
+	add_feature_if_bit(SSSE3, feat.ecx, 9);
+	add_feature_if_bit(TBM, ext1.ecx, 21);
+	add_feature_if_bit(TSXLDTRK, leaf7.edx, 16);
+	add_feature_if_bit(UINTR, leaf7.edx, 5);
+	add_feature_if_bit(USERMSR, leaf7s1.edx, 15);
+	add_feature_if_bit(VAES, leaf7.ecx, 9);
+	add_feature_if_bit(VPCLMULQDQ, leaf7.ecx, 10);
+	add_feature_if_bit(WAITPKG, leaf7.ecx, 5);
+	add_feature_if_bit(WBNOINVD, ext8.ecx, 9);
+	add_feature_if_bit(WIDEKL, leaf_19.ebx, 2);
+	add_feature_if_bit(X87, feat.edx, 0);
+	add_feature_if_bit(XOP, ext1.ecx, 11);
+	add_feature_if_bit(XSAVE, feat.ecx, 26);
+	add_feature_if_bit(XSAVEC, leaf_d.eax, 1);
+	add_feature_if_bit(XSAVEOPT, leaf_d.eax, 0);
+	add_feature_if_bit(XSAVES, leaf_d.eax, 3);
+
+}

lib/std/core/private/macho_runtime.c3

@@ -0,0 +1,254 @@
+module std::core::machoruntime @if(env::DARWIN) @private;
+
+struct SegmentCommand64
+{
+	uint cmd;
+	uint cmdsize;
+	char[16] segname;
+	ulong vmaddr;
+	ulong vmsize;
+	ulong fileoff;
+	ulong filesize;
+	uint maxprot;
+	uint initprot;
+	uint nsects;
+	uint flags;
+}
+
+struct LoadCommand
+{
+	uint cmd;
+	uint cmdsize;
+}
+
+struct Section64
+{
+	char[16] sectname;
+	char[16] segname;
+	ulong addr;
+	ulong size;
+	uint offset;
+	uint align;
+	uint reloff;
+	uint nreloc;
+	uint flags;
+	uint reserved1;
+	uint reserved2;
+	uint reserved3;
+}
+
+struct MachHeader
+{
+	uint magic;
+	uint cputype;
+	uint cpusubtype;
+	uint filetype;
+	uint ncmds;
+	uint sizeofcmds;
+	uint flags;
+}
+
+struct MachHeader64
+{
+	inline MachHeader header;
+	uint reserved;
+}
+
+const LC_SEGMENT_64 = 0x19;
+
+fault MachoSearch
+{
+	NOT_FOUND
+}
+fn bool name_cmp(char* a, char[16]* b)
+{
+	for (usz i = 0; i < 16; i++)
+	{
+		if (a[i] != (*b)[i]) return false;
+		if (a[i] == '\0') return true;
+	}
+	return false;
+}
+
+fn SegmentCommand64*! find_segment(MachHeader* header, char* segname)
+{
+	LoadCommand* command = (void*)header + MachHeader64.sizeof;
+	for (uint i = 0; i < header.ncmds; i++)
+	{
+		if (command.cmd == LC_SEGMENT_64)
+		{
+			SegmentCommand64* segment = (SegmentCommand64*)command;
+			if (name_cmp(segname, &segment.segname)) return segment;
+		}
+		command = (void*)command + command.cmdsize;
+	}
+	return MachoSearch.NOT_FOUND?;
+}
+fn Section64*! find_section(SegmentCommand64* command, char* sectname)
+{
+	Section64* section = (void*)command + SegmentCommand64.sizeof;
+	for (uint i = 0; i < command.nsects; i++)
+	{
+		if (name_cmp(sectname, &section.sectname)) return section;
+		section++;
+	}
+	return MachoSearch.NOT_FOUND?;
+}
+
+macro find_segment_section_body(MachHeader* header, char* segname, char* sectname, $Type)
+{
+
+	Section64*! section = find_section(find_segment(header, segname), sectname);
+	if (catch section)
+	{
+		return ($Type[]){};
+	}
+	$Type* ptr = (void*)header + section.offset;
+	return ptr[:section.size / $Type.sizeof];
+}
+
+def DyldCallback = fn void (MachHeader* mh, isz vmaddr_slide);
+
+extern fn void _dyld_register_func_for_add_image(DyldCallback);
+
+
+struct DlInfo
+{
+	char* dli_fname;
+	void* dli_fbase;
+	char* dli_sname;
+	void* dli_saddr;
+}
+
+extern fn void printf(char*, ...);
+extern fn int dladdr(MachHeader* mh, DlInfo* dlinfo);
+extern fn void* realloc(void* ptr, usz size);
+extern fn void* malloc(usz size);
+extern fn void free(void* ptr);
+
+def CallbackFn = fn void();
+struct Callback
+{
+	uint priority;
+	CallbackFn xtor;
+	Callback* next;
+}
+struct DynamicMethod
+{
+	void* fn_ptr;
+	char* sel;
+	union
+	{
+		DynamicMethod* next;
+		TypeId* type;
+	}
+}
+
+enum StartupState
+{
+	NOT_STARTED,
+	INIT,
+	RUN_CTORS,
+	READ_DYLIB,
+	RUN_DYLIB_CTORS,
+	RUN_DTORS,
+	SHUTDOWN
+}
+
+StartupState runtime_state = NOT_STARTED;
+
+Callback* ctor_first;
+Callback* dtor_first;
+
+fn void runtime_startup() @public @export("__c3_runtime_startup")
+{
+	if (runtime_state != NOT_STARTED) return;
+	runtime_state = INIT;
+	_dyld_register_func_for_add_image(&dl_reg_callback);
+	assert(runtime_state == INIT);
+	runtime_state = RUN_CTORS;
+	Callback* ctor = ctor_first;
+	while (ctor)
+	{
+		ctor.xtor();
+		ctor = ctor.next;
+	}
+	assert(runtime_state == RUN_CTORS);
+	runtime_state = READ_DYLIB;
+	ctor_first = null;
+}
+
+fn void runtime_finalize() @public @export("__c3_runtime_finalize")
+{
+	if (runtime_state != READ_DYLIB) return;
+	runtime_state = RUN_DTORS;
+	Callback* dtor = dtor_first;
+	while (dtor)
+	{
+		dtor.xtor();
+		dtor = dtor.next;
+	}
+	assert(runtime_state == RUN_DTORS);
+	runtime_state = SHUTDOWN;
+}
+
+fn void append_xxlizer(Callback** ref, Callback* cb)
+{
+	while (Callback* current = *ref, current)
+	{
+		if (current.priority > cb.priority)
+		{
+			cb.next = current;
+			break;
+		}
+		ref = &current.next;
+	}
+	*ref = cb;
+}
+
+struct TypeId
+{
+	char type;
+	TypeId* parentof;
+	DynamicMethod* dtable;
+	usz sizeof;
+	TypeId* inner;
+	usz len;
+	typeid[?] additional;
+}
+
+fn void dl_reg_callback(MachHeader* mh, isz vmaddr_slide)
+{
+	usz size = 0;
+	assert(runtime_state == INIT || runtime_state == READ_DYLIB, "State was %s", runtime_state);
+	foreach (&dm : find_segment_section_body(mh, "__DATA", "__c3_dynamic", DynamicMethod))
+	{
+		TypeId* type = dm.type;
+		dm.next = type.dtable;
+		type.dtable = dm;
+		DynamicMethod* m = dm;
+		while (m)
+		{
+			m = m.next;
+		}
+	}
+	foreach (&cb : find_segment_section_body(mh, "__DATA", "__c3dtor", Callback))
+	{
+		append_xxlizer(&dtor_first, cb);
+	}
+	foreach (&cb : find_segment_section_body(mh, "__DATA", "__c3ctor", Callback))
+	{
+		append_xxlizer(&ctor_first, cb);
+	}
+	if (runtime_state != READ_DYLIB) return;
+	runtime_state = RUN_DYLIB_CTORS;
+	Callback* ctor = ctor_first;
+	ctor_first = null;
+	while (ctor)
+	{
+		ctor.xtor();
+		ctor = ctor.next;
+	}
+	assert(runtime_state == RUN_DYLIB_CTORS);
+	runtime_state = READ_DYLIB;
+}

lib/std/core/private/main_stub.c3

@@ -0,0 +1,180 @@
+module std::core::main_stub;
+
+macro usz _strlen(ptr) @private
+{
+	usz len = 0;
+	while (ptr[len]) len++;
+	return len;
+}
+
+macro int @main_to_err_main(#m, int, char**)
+{
+	if (catch #m()) return 1;
+	return 0;
+}
+macro int @main_to_int_main(#m, int, char**) => #m();
+macro int @main_to_void_main(#m, int, char**)
+{
+	#m();
+	return 0;
+}
+
+macro String[] args_to_strings(int argc, char** argv) @private
+{
+	String[] list = mem::alloc_array(String, argc);
+	for (int i = 0; i < argc; i++)
+	{
+		char* arg = argv[i];
+		usz len = 0;
+		list[i] = (String)arg[:_strlen(arg)];
+	}
+	return list;
+}
+
+
+macro int @main_to_err_main_args(#m, int argc, char** argv)
+{
+	String[] list = args_to_strings(argc, argv);
+	defer free(list.ptr);
+	if (catch #m(list)) return 1;
+	return 0;
+}
+
+macro int @main_to_int_main_args(#m, int argc, char** argv)
+{
+	String[] list = args_to_strings(argc, argv);
+	defer free(list.ptr);
+	return #m(list);
+}
+
+macro int @_main_runner(#m, int argc, char** argv)
+{
+	String[] list = args_to_strings(argc, argv);
+	defer free(list.ptr);
+	return #m(list) ? 0 : 1;
+}
+
+macro int @main_to_void_main_args(#m, int argc, char** argv)
+{
+	String[] list = args_to_strings(argc, argv);
+	defer free(list.ptr);
+	#m(list);
+	return 0;
+}
+
+module std::core::main_stub @if(env::WIN32);
+
+extern fn Char16** _win_command_line_to_argv_w(ushort* cmd_line, int* argc_ptr) @extern("CommandLineToArgvW");
+
+macro String[] win_command_line_to_strings(ushort* cmd_line) @private
+{
+	int argc;
+	Char16** argv = _win_command_line_to_argv_w(cmd_line, &argc);
+	return wargs_strings(argc, argv);
+}
+
+macro String[] wargs_strings(int argc, Char16** argv) @private
+{
+	String[] list = mem::alloc_array(String, argc);
+	for (int i = 0; i < argc; i++)
+	{
+		Char16* arg = argv[i];
+		Char16[] argstring = arg[:_strlen(arg)];
+		list[i] = string::new_from_utf16(argstring) ?? "?".copy();
+	}
+	return list[:argc];
+}
+
+macro void release_wargs(String[] list) @private
+{
+	foreach (s : list) free(s.ptr);
+	free(list.ptr);
+}
+
+macro int @win_to_err_main_noargs(#m, void* handle, void* prev_handle, Char16* cmd_line, int show_cmd)
+{
+	if (catch #m()) return 1;
+	return 0;
+}
+macro int @win_to_int_main_noargs(#m, void* handle, void* prev_handle, Char16* cmd_line, int show_cmd) => #m();
+macro int @win_to_void_main_noargs(#m, void* handle, void* prev_handle, Char16* cmd_line, int show_cmd)
+{
+	#m();
+	return 0;
+}
+
+macro int @win_to_err_main_args(#m, void* handle, void* prev_handle, Char16* cmd_line, int show_cmd)
+{
+	String[] args = win_command_line_to_strings(cmd_line);
+	defer release_wargs(args);
+	if (catch #m(args)) return 1;
+	return 0;
+}
+
+macro int @win_to_int_main_args(#m, void* handle, void* prev_handle, Char16* cmd_line, int show_cmd)
+{
+	String[] args = win_command_line_to_strings(cmd_line);
+	defer release_wargs(args);
+	return #m(args);
+}
+
+macro int @win_to_void_main_args(#m, void* handle, void* prev_handle, Char16* cmd_line, int show_cmd)
+{
+	String[] args = win_command_line_to_strings(cmd_line);
+	defer release_wargs(args);
+	#m(args);
+	return 0;
+}
+
+macro int @win_to_err_main(#m, void* handle, void* prev_handle, Char16* cmd_line, int show_cmd)
+{
+	String[] args = win_command_line_to_strings(cmd_line);
+	defer release_wargs(args);
+	if (catch #m(handle, prev_handle, args, show_cmd)) return 1;
+	return 0;
+}
+
+macro int @win_to_int_main(#m, void* handle, void* prev_handle, Char16* cmd_line, int show_cmd)
+{
+	String[] args = win_command_line_to_strings(cmd_line);
+	defer release_wargs(args);
+	return #m(handle, prev_handle, args, show_cmd);
+}
+
+macro int @win_to_void_main(#m, void* handle, void* prev_handle, Char16* cmd_line, int show_cmd)
+{
+	String[] args = win_command_line_to_strings(cmd_line);
+	defer release_wargs(args);
+	#m(handle, prev_handle, args, show_cmd);
+	return 0;
+}
+
+macro int @wmain_to_err_main_args(#m, int argc, Char16** argv)
+{
+	String[] args = wargs_strings(argc, argv);
+	defer release_wargs(args);
+	if (catch #m(args)) return 1;
+	return 1;
+}
+
+macro int @wmain_to_int_main_args(#m, int argc, Char16** argv)
+{
+	String[] args = wargs_strings(argc, argv);
+	defer release_wargs(args);
+	return #m(args);
+}
+
+macro int @_wmain_runner(#m, int argc, Char16** argv)
+{
+	String[] args = wargs_strings(argc, argv);
+	defer release_wargs(args);
+	return #m(args) ? 0 : 1;
+}
+
+macro int @wmain_to_void_main_args(#m, int argc, Char16** argv)
+{
+	String[] args = wargs_strings(argc, argv);
+	defer release_wargs(args);
+	#m(args);
+	return 0;
+}

lib/std/core/runtime.c3

@@ -0,0 +1,33 @@
+// Copyright (c) 2021 Christoffer Lerno. All rights reserved.
+// Use of this source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+module std::core::runtime;
+import libc, std::time, std::io, std::sort;
+
+struct ReflectedParam (Printable) @if(!$defined(ReflectedParam))
+{
+	String name;
+	typeid type;
+}
+
+struct AnyRaw
+{
+	void* ptr;
+	typeid type;
+}
+
+struct SliceRaw
+{
+	void* ptr;
+	usz len;
+}
+
+
+module std::core::runtime @if(WASM_NOLIBC);
+
+extern fn void __wasm_call_ctors();
+fn void wasm_initialize() @extern("_initialize") @wasm
+{
+	// The linker synthesizes this to call constructors.
+	__wasm_call_ctors();
+}

lib/std/core/runtime_benchmark.c3

@@ -0,0 +1,174 @@
+module std::core::runtime;
+import libc, std::time, std::io, std::sort;
+
+def BenchmarkFn = fn void!() @if($$OLD_TEST);
+def BenchmarkFn = fn void() @if(!$$OLD_TEST);
+
+struct BenchmarkUnit
+{
+	String name;
+	BenchmarkFn func;
+}
+
+fn BenchmarkUnit[] benchmark_collection_create(Allocator allocator = allocator::heap())
+{
+	BenchmarkFn[] fns = $$BENCHMARK_FNS;
+	String[] names = $$BENCHMARK_NAMES;
+	BenchmarkUnit[] benchmarks = allocator::alloc_array(allocator, BenchmarkUnit, names.len);
+	foreach (i, benchmark : fns)
+	{
+		benchmarks[i] = { names[i], fns[i] };
+	}
+	return benchmarks;
+}
+
+const DEFAULT_BENCHMARK_WARMUP_ITERATIONS = 3;
+const DEFAULT_BENCHMARK_MAX_ITERATIONS = 10000;
+
+uint benchmark_warmup_iterations @private = DEFAULT_BENCHMARK_WARMUP_ITERATIONS;
+uint benchmark_max_iterations @private = DEFAULT_BENCHMARK_MAX_ITERATIONS;
+
+fn void set_benchmark_warmup_iterations(uint value) @builtin
+{
+    benchmark_warmup_iterations = value;
+}
+
+fn void set_benchmark_max_iterations(uint value) @builtin
+{
+    assert(value > 0);
+    benchmark_max_iterations = value;
+}
+
+fn bool run_benchmarks(BenchmarkUnit[] benchmarks) @if($$OLD_TEST)
+{
+	int benchmarks_passed = 0;
+	int benchmark_count = benchmarks.len;
+	usz max_name;
+
+	foreach (&unit : benchmarks)
+	{
+		if (max_name < unit.name.len) max_name = unit.name.len;
+	}
+
+	usz len = max_name + 9;
+
+	DString name = dstring::temp_with_capacity(64);
+	name.append_repeat('-', len / 2);
+	name.append(" BENCHMARKS ");
+	name.append_repeat('-', len - len / 2);
+
+	io::printn(name);
+
+	name.clear();
+
+	long sys_clock_started;
+	long sys_clock_finished;
+	long sys_clocks;
+	Clock clock;
+	anyfault err;
+
+	foreach(unit : benchmarks)
+	{
+		defer name.clear();
+		name.appendf("Benchmarking %s ", unit.name);
+		name.append_repeat('.', max_name - unit.name.len + 2);
+		io::printf("%s ", name.str_view());
+
+		for (uint i = 0; i < benchmark_warmup_iterations; i++)
+		{
+			err = @catch(unit.func()) @inline;
+			@volatile_load(err);
+		}
+
+		clock = std::time::clock::now();
+		sys_clock_started = $$sysclock();
+
+		for (uint i = 0; i < benchmark_max_iterations; i++)
+		{
+			err = @catch(unit.func()) @inline;
+			@volatile_load(err);
+		}
+
+		sys_clock_finished = $$sysclock();
+		NanoDuration nano_seconds = clock.mark();
+		sys_clocks = sys_clock_finished - sys_clock_started;
+
+		if (err)
+		{
+			io::printfn("[failed] Failed due to: %s", err);
+			continue;
+		}
+
+		io::printfn("[ok] %.2f ns, %.2f CPU's clocks", (float)nano_seconds / benchmark_max_iterations, (float)sys_clocks / benchmark_max_iterations);
+		benchmarks_passed++;
+	}
+
+	io::printfn("\n%d benchmark%s run.\n", benchmark_count, benchmark_count > 1 ? "s" : "");
+	io::printfn("Benchmarks Result: %s. %d passed, %d failed.",
+		benchmarks_passed < benchmark_count ? "FAILED" : "ok",
+		benchmarks_passed,
+		benchmark_count - benchmarks_passed);
+
+	return benchmark_count == benchmarks_passed;
+}
+
+fn bool run_benchmarks(BenchmarkUnit[] benchmarks) @if(!$$OLD_TEST)
+{
+	usz max_name;
+
+	foreach (&unit : benchmarks)
+	{
+		if (max_name < unit.name.len) max_name = unit.name.len;
+	}
+
+	usz len = max_name + 9;
+
+	DString name = dstring::temp_with_capacity(64);
+	name.append_repeat('-', len / 2);
+	name.append(" BENCHMARKS ");
+	name.append_repeat('-', len - len / 2);
+
+	io::printn(name);
+
+	name.clear();
+
+	long sys_clock_started;
+	long sys_clock_finished;
+	long sys_clocks;
+	Clock clock;
+
+	foreach(unit : benchmarks)
+	{
+		defer name.clear();
+		name.appendf("Benchmarking %s ", unit.name);
+		name.append_repeat('.', max_name - unit.name.len + 2);
+		io::printf("%s ", name.str_view());
+
+		for (uint i = 0; i < benchmark_warmup_iterations; i++)
+		{
+			unit.func() @inline;
+		}
+
+		clock = std::time::clock::now();
+		sys_clock_started = $$sysclock();
+
+		for (uint i = 0; i < benchmark_max_iterations; i++)
+		{
+			unit.func() @inline;
+		}
+
+		sys_clock_finished = $$sysclock();
+		NanoDuration nano_seconds = clock.mark();
+		sys_clocks = sys_clock_finished - sys_clock_started;
+
+		io::printfn("[COMPLETE] %.2f ns, %.2f CPU's clocks", (float)nano_seconds / benchmark_max_iterations, (float)sys_clocks / benchmark_max_iterations);
+	}
+
+	io::printfn("\n%d benchmark%s run.\n", benchmarks.len, benchmarks.len > 1 ? "s" : "");
+	return true;
+}
+
+fn bool default_benchmark_runner(String[] args) => @pool()
+{
+	return run_benchmarks(benchmark_collection_create(allocator::temp()));
+}

lib/std/core/runtime_test.c3

@@ -0,0 +1,342 @@
+// Copyright (c) 2025 Christoffer Lerno. All rights reserved.
+// Use of this source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+module std::core::runtime;
+import std::core::test @public;
+import std::core::mem::allocator @public;
+import libc, std::time, std::io, std::sort;
+import std::os::env;
+
+def TestFn = fn void!() @if($$OLD_TEST);
+def TestFn = fn void() @if(!$$OLD_TEST);
+
+TestContext* test_context @private;
+
+struct TestContext
+{
+	JmpBuf buf;
+	// Allows filtering test cased or modules by substring, e.g. 'foo::', 'foo::test_add'
+	String test_filter;
+	// Triggers debugger breakpoint when assert or test:: checks failed
+	bool breakpoint_on_assert;
+
+	// internal state
+	bool assert_print_backtrace;
+	bool has_ansi_codes;
+	bool is_in_panic;
+	bool is_quiet_mode;
+	bool is_no_capture;
+	String current_test_name;
+	TestFn setup_fn;
+	TestFn teardown_fn;
+
+	char* error_buffer;
+	usz error_buffer_capacity;
+	File fake_stdout;
+	struct stored
+	{
+		File stdout;
+		File stderr;
+		Allocator allocator;
+	}
+}
+
+struct TestUnit
+{
+	String name;
+	TestFn func;
+}
+
+fn TestUnit[] test_collection_create(Allocator allocator = allocator::heap())
+{
+	TestFn[] fns = $$TEST_FNS;
+	String[] names = $$TEST_NAMES;
+	TestUnit[] tests = allocator::alloc_array(allocator, TestUnit, names.len);
+	foreach (i, test : fns)
+	{
+		tests[i] = { names[i], fns[i] };
+	}
+	return tests;
+}
+
+// Sort the tests by their name in ascending order.
+fn int cmp_test_unit(TestUnit a, TestUnit b)
+{
+	usz an = a.name.len;
+	usz bn = b.name.len;
+	if (an > bn) @swap(a, b);
+	foreach (i, ac : a.name)
+	{
+		char bc = b.name[i];
+		if (ac != bc) return an > bn ? bc - ac : ac - bc;
+	}
+	return (int)(an - bn);
+}
+
+fn bool terminal_has_ansi_codes() @local => @pool()
+{
+	if (try v = env::get_var_temp("TERM"))
+	{
+		if (v.contains("xterm") || v.contains("vt100") || v.contains("screen")) return true;
+	}
+	$if env::WIN32 || env::NO_LIBC:
+		return false;
+	$else
+		return io::stdout().isatty();
+	$endif
+}
+
+fn void test_panic(String message, String file, String function, uint line) @local
+{
+	if (test_context.is_in_panic) return;
+	test_context.is_in_panic = true;
+
+	unmute_output(true);
+	(void)io::stdout().flush();
+	if (test_context.assert_print_backtrace)
+	{
+		$if env::NATIVE_STACKTRACE:
+    		builtin::print_backtrace(message, 0);
+		$endif
+	}
+	io::printf("\nTest failed ^^^ ( %s:%s ) %s\n",  file, line, message);
+	test_context.assert_print_backtrace = true;
+
+	if (test_context.breakpoint_on_assert)
+	{
+		breakpoint();
+	}
+
+	if (test_context.teardown_fn)
+	{
+		test_context.teardown_fn();
+	}
+
+	test_context.is_in_panic = false;
+	allocator::thread_allocator = test_context.stored.allocator;
+	libc::longjmp(&test_context.buf, 1);
+}
+
+fn void mute_output() @local
+{
+	if (test_context.is_no_capture || !test_context.fake_stdout.file) return;
+	File* stdout = io::stdout();
+    File* stderr = io::stderr();
+	*stderr = test_context.fake_stdout;
+	*stdout = test_context.fake_stdout;
+	(void)test_context.fake_stdout.seek(0, Seek.SET)!!;
+}
+
+fn void unmute_output(bool has_error) @local
+{
+	if (test_context.is_no_capture || !test_context.fake_stdout.file) return;
+
+	File* stdout = io::stdout();
+	File* stderr = io::stderr();
+
+	*stderr = test_context.stored.stderr;
+	*stdout = test_context.stored.stdout;
+
+	usz log_size = test_context.fake_stdout.seek(0, Seek.CURSOR)!!;
+	if (has_error)
+	{
+		io::printf("\nTesting %s ", test_context.current_test_name);
+		io::printn(test_context.has_ansi_codes ? "[\e[0;31mFAIL\e[0m]" : "[FAIL]");
+	}
+
+	if (has_error && log_size > 0)
+	{
+		test_context.fake_stdout.write_byte('\n')!!;
+		test_context.fake_stdout.write_byte('\0')!!;
+		(void)test_context.fake_stdout.seek(0, Seek.SET)!!;
+
+		io::printfn("\n========== TEST LOG ============");
+		io::printfn("%s\n", test_context.current_test_name);
+		while (try c = test_context.fake_stdout.read_byte())
+		{
+			if (@unlikely(c == '\0'))
+			{
+				// ignore junk from previous tests
+				break;
+			}
+			libc::putchar(c);
+		}
+		io::printf("========== TEST END ============");
+	}
+	(void)stdout.flush();
+}
+
+fn bool run_tests(String[] args, TestUnit[] tests) @private
+{
+	usz max_name;
+	bool sort_tests = true;
+	bool check_leaks = true;
+	foreach (&unit : tests)
+	{
+		if (max_name < unit.name.len) max_name = unit.name.len;
+	}
+	TestContext context =
+	{
+		.assert_print_backtrace = true,
+		.breakpoint_on_assert = false,
+		.test_filter = "",
+		.has_ansi_codes = terminal_has_ansi_codes(),
+		.stored.allocator = allocator::heap(),
+		.stored.stderr = *io::stderr(),
+		.stored.stdout = *io::stdout(),
+	};
+	for (int i = 1; i < args.len; i++)
+	{
+		switch (args[i])
+		{
+			case "--test-breakpoint":
+				context.breakpoint_on_assert = true;
+			case "--test-nosort":
+				sort_tests = false;
+			case "--test-noleak":
+				check_leaks = false;
+			case "--test-nocapture":
+				context.is_no_capture = true;
+			case "--noansi":
+				context.has_ansi_codes = false;
+			case "--useansi":
+				context.has_ansi_codes = true;
+			case "--test-quiet":
+				context.is_quiet_mode = true;
+			case "--test-filter":
+				if (i == args.len - 1)
+				{
+					io::printn("Invalid arguments to test runner.");
+					return false;
+				}
+				context.test_filter = args[i + 1];
+				i++;
+			default:
+				io::printfn("Unknown argument: %s", args[i]);
+		}
+	}
+	test_context = &context;
+
+	if (sort_tests)
+	{
+		quicksort(tests, &cmp_test_unit);
+	}
+
+	// Buffer for hijacking the output
+	$if (!env::NO_LIBC):
+		context.fake_stdout.file = libc::tmpfile();
+	$endif
+	if (context.fake_stdout.file == null)
+	{
+		io::print("Failed to hijack stdout, tests will print everything");
+	}
+
+	PanicFn old_panic = builtin::panic;
+	defer builtin::panic = old_panic;
+	builtin::panic = &test_panic;
+	int tests_passed = 0;
+	int tests_skipped = 0;
+	int test_count = tests.len;
+	DString name = dstring::temp_with_capacity(64);
+	usz len = max_name + 9;
+	name.append_repeat('-', len / 2);
+	name.append(" TESTS ");
+	name.append_repeat('-', len - len / 2);
+	if (!context.is_quiet_mode) io::printn(name);
+	name.clear();
+	TempState temp_state = mem::temp_push();
+	defer mem::temp_pop(temp_state);
+	foreach(unit : tests)
+	{
+		mem::temp_pop(temp_state);
+		if (context.test_filter && !unit.name.contains(context.test_filter))
+		{
+			tests_skipped++;
+			continue;
+		}
+		context.setup_fn = null;
+		context.teardown_fn = null;
+		context.current_test_name = unit.name;
+
+		defer name.clear();
+		name.appendf("Testing %s ", unit.name);
+		name.append_repeat('.', max_name - unit.name.len + 2);
+		if (context.is_quiet_mode)
+		{
+			io::print(".");
+		}
+		else
+		{
+			io::printf("%s ", name.str_view());
+		}
+		(void)io::stdout().flush();
+		TrackingAllocator mem;
+
+		mem.init(context.stored.allocator);
+		if (libc::setjmp(&context.buf) == 0)
+		{
+			mute_output();
+			mem.clear();
+			if (check_leaks) allocator::thread_allocator = &mem;
+			$if(!$$OLD_TEST):
+				unit.func();
+			$else
+				if (catch err = unit.func())
+				{
+					io::printf("[FAIL] Failed due to: %s", err);
+					continue;
+				}
+			$endif
+			// track cleanup that may take place in teardown_fn
+			if (context.teardown_fn)
+			{
+				context.teardown_fn();
+			}
+			if (check_leaks) allocator::thread_allocator = context.stored.allocator;
+
+			unmute_output(false); // all good, discard output
+			if (mem.has_leaks())
+			{
+				if(context.is_quiet_mode) io::printf("\n%s ", context.current_test_name);
+				io::print(context.has_ansi_codes ? "[\e[0;31mFAIL\e[0m]" : "[FAIL]");
+				io::printn(" LEAKS DETECTED!");
+				mem.print_report();
+			}
+			else
+			{
+				if (!context.is_quiet_mode)
+				{
+					io::printfn(context.has_ansi_codes ? "[\e[0;32mPASS\e[0m]" : "[PASS]");
+				}
+				tests_passed++;
+			}
+		}
+		mem.free();
+	}
+	io::printfn("\n%d test%s run.\n", test_count-tests_skipped, test_count > 1 ? "s" : "");
+
+	int n_failed = test_count - tests_passed - tests_skipped;
+	io::printf("Test Result: %s%s%s: ",
+				context.has_ansi_codes ? (n_failed ? "\e[0;31m" : "\e[0;32m") : "",
+				n_failed ? "FAILED" : "PASSED",
+				context.has_ansi_codes ? "\e[0m" : "",
+				);
+
+	io::printfn("%d passed, %d failed, %d skipped.",
+				tests_passed,
+				n_failed,
+				tests_skipped);
+
+	// cleanup fake_stdout file
+	if (context.fake_stdout.file) libc::fclose(context.fake_stdout.file);
+	context.fake_stdout.file = null;
+
+	return n_failed == 0;
+}
+
+fn bool default_test_runner(String[] args) => @pool()
+{
+	assert(test_context == null, "test suite is already running");
+	return run_tests(args, test_collection_create(allocator::temp()));
+}
+

lib/std/core/sanitizer/asan.c3

@@ -0,0 +1,127 @@
+// Add this to your code to suppress leak detection or set other default options
+// fn ZString __asan_default_options() @export("__asan_default_options") @if(env::ADDRESS_SANITIZER)
+// {
+//   return "detect_leaks=0";
+// }
+
+// Add this to break on error
+// asan::set_error_report_callback(fn (ZString err)
+// {
+//   breakpoint();
+// });
+
+module std::core::sanitizer::asan;
+
+def ErrorCallback = fn void (ZString);
+
+<*
+ Marks a memory region ([addr, addr+size)) as unaddressable.
+
+ This memory must be previously allocated by your program. Instrumented
+ code is forbidden from accessing addresses in this region until it is
+ unpoisoned. This function is not guaranteed to poison the entire region -
+ it could poison only a subregion of [addr, addr+size) due to ASan
+ alignment restrictions.
+
+ NOTE This function is not thread-safe because no two threads can poison or
+ unpoison memory in the same memory region simultaneously.
+
+ @param addr "Start of memory region."
+ @param size "Size of memory region."
+*>
+macro poison_memory_region(void* addr, usz size)
+{
+	$if env::ADDRESS_SANITIZER: 
+		__asan_poison_memory_region(addr, size);
+	$endif 
+}
+
+<*
+ Marks a memory region ([addr, addr+size)) as addressable.
+
+ This memory must be previously allocated by your program. Accessing
+ addresses in this region is allowed until this region is poisoned again.
+ This function could unpoison a super-region of [addr, addr+size) due
+ to ASan alignment restrictions.
+
+ NOTE This function is not thread-safe because no two threads can
+ poison or unpoison memory in the same memory region simultaneously.
+
+ @param addr "Start of memory region."
+ @param size "Size of memory region."
+*>
+macro unpoison_memory_region(void* addr, usz size)
+{
+	$if env::ADDRESS_SANITIZER: 
+		__asan_unpoison_memory_region(addr, size);
+	$endif 
+}
+
+<*
+ Checks if an address is poisoned.
+ @return "True if 'addr' is poisoned (that is, 1-byte read/write access to this address would result in an error report from ASan). Otherwise returns false."
+ @param addr "Address to check."
+*>
+macro bool address_is_poisoned(void* addr)
+{
+	$if env::ADDRESS_SANITIZER:
+		return (bool)__asan_address_is_poisoned(addr);
+	$else
+		return false;
+	$endif
+}
+
+<*
+ Checks if a region is poisoned.
+
+ If at least one byte in [beg, beg+size) is poisoned, returns the
+ address of the first such byte. Otherwise returns 0.
+
+ @param beg "Start of memory region."
+ @param size "Start of memory region."
+ @return "Address of first poisoned byte."
+*>
+macro void* region_is_poisoned(void* beg, usz size)
+{
+	$if env::ADDRESS_SANITIZER:
+		return __asan_region_is_poisoned(beg, size);
+	$else
+		return null;
+	$endif
+}
+
+<*
+ Sets the callback function to be called during ASan error reporting.
+*>
+fn void set_error_report_callback(ErrorCallback callback)
+{
+	$if env::ADDRESS_SANITIZER:
+		__asan_set_error_report_callback(callback);
+	$endif
+}
+
+module std::core::sanitizer::asan @if(env::ADDRESS_SANITIZER);
+
+extern fn void __asan_poison_memory_region(void* addr, usz size);
+extern fn void __asan_unpoison_memory_region(void* addr, usz size);
+extern fn CInt __asan_address_is_poisoned(void* addr);
+extern fn void* __asan_region_is_poisoned(void* beg, usz size);
+extern fn void __asan_describe_address(void* addr);
+extern fn CInt __asan_report_present();
+extern fn void* __asan_get_report_pc();
+extern fn void* __asan_get_report_bp();
+extern fn void* __asan_get_report_sp();
+extern fn void* __asan_get_report_address();
+extern fn CInt __asan_get_report_access_type();
+extern fn usz __asan_get_report_access_size();
+extern fn ZString __asan_get_report_description();
+extern fn ZString __asan_locate_address(void* addr, char* name, usz name_size, void** region_address, usz* region_size);
+extern fn usz __asan_get_alloc_stack(void* addr, void** trace, usz size, CInt* thread_id);
+extern fn usz __asan_get_free_stack(void* addr, void** trace, usz size, CInt* thread_id);
+extern fn void __asan_get_shadow_mapping(usz* shadow_scale, usz* shadow_offset);
+extern fn void __asan_set_error_report_callback(ErrorCallback callback);
+extern fn void __asan_print_accumulated_stats();
+extern fn void* __asan_get_current_fake_stack();
+extern fn void* __asan_addr_is_in_fake_stack(void* fake_stack, void* addr, void** beg, void** end);
+extern fn void __asan_handle_no_return();
+extern fn CInt __asan_update_allocation_context(void* addr);

lib/std/core/sanitizer/sanitizer.c3

@@ -0,0 +1,80 @@
+module std::core::sanitizer;
+
+macro void annotate_contiguous_container(void* beg, void* end, void* old_mid, void* new_mid)
+{
+	$if env::ADDRESS_SANITIZER:
+		__sanitizer_annotate_contiguous_container(beg, end, old_mid, new_mid);
+	$endif
+}
+
+macro void annotate_double_ended_contiguous_container(void* storage_beg, void* storage_end, void* old_container_beg, void* old_container_end, void* new_container_beg, void* new_container_end)
+{
+	$if env::ADDRESS_SANITIZER:
+		__sanitizer_annotate_double_ended_contiguous_container(storage_beg, storage_end, old_container_beg, old_container_end, new_container_beg, new_container_end);
+	$endif
+}
+
+macro void print_stack_trace()
+{
+	$if env::ADDRESS_SANITIZER:
+		__sanitizer_print_stack_trace();
+	$endif
+}
+
+fn void set_death_callback(VoidFn callback)
+{
+	$if env::ANY_SANITIZER:
+		__sanitizer_set_death_callback(callback);
+	$endif
+}
+
+module std::core::sanitizer @if (env::ANY_SANITIZER);
+
+struct __Sanitizer_sandbox_arguments
+{
+	CInt coverage_sandboxed;
+	iptr coverage_fd;
+	CUInt coverage_max_block_size;
+}
+
+extern fn void __sanitizer_set_report_path(ZString path);
+extern fn void __sanitizer_set_report_fd(void* fd);
+extern fn ZString __sanitizer_get_report_path();
+extern fn void __sanitizer_sandbox_on_notify(__Sanitizer_sandbox_arguments* args);
+extern fn void __sanitizer_report_error_summary(ZString error_summary);
+extern fn ushort __sanitizer_unaligned_load16(void* p);
+extern fn uint __sanitizer_unaligned_load32(void* p);
+extern fn ulong __sanitizer_unaligned_load64(void* p);
+extern fn void __sanitizer_unaligned_store16(void* p, ushort x);
+extern fn void __sanitizer_unaligned_store32(void* p, uint x);
+extern fn void __sanitizer_unaligned_store64(void* p, ulong x);
+extern fn CInt __sanitizer_acquire_crash_state();
+extern fn void __sanitizer_annotate_contiguous_container(void* beg, void* end, void* old_mid, void* new_mid);
+extern fn void __sanitizer_annotate_double_ended_contiguous_container(void* storage_beg, void* storage_end,
+	void* old_container_beg, void* old_container_end,
+	void* new_container_beg, void* new_container_end);
+extern fn CInt __sanitizer_verify_contiguous_container(void* beg, void* mid, void* end);
+extern fn CInt __sanitizer_verify_double_ended_contiguous_container(
+	void* storage_beg, void* container_beg,
+	void* container_end, void* storage_end);
+extern fn void* __sanitizer_contiguous_container_find_bad_address(void* beg, void* mid, void* end);
+extern fn void* __sanitizer_double_ended_contiguous_container_find_bad_address(
+	void* storage_beg, void* container_beg,
+	void* container_end, void* storage_end);
+
+extern fn void __sanitizer_print_stack_trace();
+extern fn void __sanitizer_symbolize_pc(void* pc, ZString fmt, char* out_buf, usz out_buf_size);
+extern fn void __sanitizer_symbolize_global(void* data_ptr, ZString fmt, char* out_buf, usz out_buf_size);
+extern fn void __sanitizer_set_death_callback(VoidFn callback);
+extern fn void __sanitizer_weak_hook_memcmp(void* called_pc, void* s1, void* s2, usz n, CInt result);
+extern fn void __sanitizer_weak_hook_strncmp(void* called_pc, ZString s1, ZString s2, usz n, CInt result);
+extern fn void __sanitizer_weak_hook_strncasecmp(void* called_pc, ZString s1, ZString s2, usz n, CInt result);
+extern fn void __sanitizer_weak_hook_strcmp(void* called_pc, ZString s1, ZString s2, CInt result);
+extern fn void __sanitizer_weak_hook_strcasecmp(void* called_pc, ZString s1, ZString s2, CInt result);
+extern fn void __sanitizer_weak_hook_strstr(void* called_pc, ZString s1, ZString s2, char* result);
+extern fn void __sanitizer_weak_hook_strcasestr(void* called_pc, ZString s1, ZString s2, char* result);
+extern fn void __sanitizer_weak_hook_memmem(void* called_pc, void* s1, usz len1, void* s2, usz len2, void* result);
+extern fn void __sanitizer_print_memory_profile(usz top_percent, usz max_number_of_contexts);
+extern fn void __sanitizer_start_switch_fiber(void** fake_stack_save, void* bottom, usz size);
+extern fn void __sanitizer_finish_switch_fiber(void* fake_stack_save, void** bottom_old, usz* size_old);
+extern fn CInt __sanitizer_get_module_and_offset_for_pc(void* pc, char* module_path, usz module_path_len, void** pc_offset);

lib/std/core/sanitizer/tsan.c3

@@ -0,0 +1,39 @@
+module std::core::sanitizer::tsan;
+
+distinct MutexFlags = inline CUInt;
+
+const MutexFlags MUTEX_LINKER_INIT      = 1 << 0;
+const MutexFlags MUTEX_WRITE_REENTRANT  = 1 << 1;
+const MutexFlags MUTEX_READ_REENTRANT   = 1 << 2;
+const MutexFlags MUTEX_NOT_STATIC       = 1 << 8;
+const MutexFlags MUTEX_READ_LOCK        = 1 << 3;
+const MutexFlags MUTEX_TRY_LOCK         = 1 << 4;
+const MutexFlags MUTEX_TRY_LOCK_FAILED  = 1 << 5;
+const MutexFlags MUTEX_RECURSIVE_LOCK   = 1 << 6;
+const MutexFlags MUTEX_RECURSIVE_UNLOCK = 1 << 7;
+const MutexFlags MUTEX_TRY_READ_LOCK = MUTEX_READ_LOCK | MUTEX_TRY_LOCK;
+const MutexFlags MUTEX_TRY_READ_LOCK_FAILED = MUTEX_TRY_READ_LOCK | MUTEX_TRY_LOCK_FAILED;
+
+macro void mutex_create(void* addr, MutexFlags flags) { $if env::THREAD_SANITIZER: __tsan_mutex_create(addr, flags); $endif }
+macro void mutex_destroy(void* addr, MutexFlags flags) { $if env::THREAD_SANITIZER: __tsan_mutex_destroy(addr, flags); $endif }
+macro void mutex_pre_lock(void* addr, MutexFlags flags) { $if env::THREAD_SANITIZER: __tsan_mutex_pre_lock(addr, flags); $endif }
+macro void mutex_post_lock(void* addr, MutexFlags flags, CInt recursion) { $if env::THREAD_SANITIZER: __tsan_mutex_post_lock(addr, flags, recursion); $endif }
+macro CInt mutex_pre_unlock(void* addr, MutexFlags flags) { $if env::THREAD_SANITIZER: return __tsan_mutex_pre_unlock(addr, flags); $else return 0; $endif }
+macro void mutex_post_unlock(void* addr, MutexFlags flags) { $if env::THREAD_SANITIZER: __tsan_mutex_post_unlock(addr, flags); $endif }
+macro void mutex_pre_signal(void* addr, MutexFlags flags) { $if env::THREAD_SANITIZER: __tsan_mutex_pre_signal(addr, flags); $endif }
+macro void mutex_post_signal(void* addr, MutexFlags flags) { $if env::THREAD_SANITIZER: __tsan_mutex_post_signal(addr, flags); $endif }
+macro void mutex_pre_divert(void* addr, MutexFlags flags) { $if env::THREAD_SANITIZER: __tsan_mutex_pre_divert(addr, flags); $endif }
+macro void mutex_post_divert(void* addr, MutexFlags flags) { $if env::THREAD_SANITIZER: __tsan_mutex_post_divert(addr, flags); $endif }
+
+module std::core::sanitizer::tsan @if(env::THREAD_SANITIZER) @private;
+
+extern fn void __tsan_mutex_create(void* addr, CUInt flags);
+extern fn void __tsan_mutex_destroy(void* addr, CUInt flags);
+extern fn void __tsan_mutex_pre_lock(void* addr, CUInt flags);
+extern fn void __tsan_mutex_post_lock(void* addr, CUInt flags, CInt recursion);
+extern fn CInt __tsan_mutex_pre_unlock(void* addr, CUInt flags);
+extern fn void __tsan_mutex_post_unlock(void* addr, CUInt flags);
+extern fn void __tsan_mutex_pre_signal(void* addr, CUInt flags);
+extern fn void __tsan_mutex_post_signal(void* addr, CUInt flags);
+extern fn void __tsan_mutex_pre_divert(void* addr, CUInt flags);
+extern fn void __tsan_mutex_post_divert(void* addr, CUInt flags);

lib/std/core/string.c3

@@ -0,0 +1,879 @@
+module std::core::string;
+import std::ascii;
+import std::io;
+
+distinct String @if(!$defined(String)) = inline char[];
+distinct ZString = inline char*;
+distinct WString = inline Char16*;
+def Char32 = uint;
+def Char16 = ushort;
+
+fault UnicodeResult
+{
+	INVALID_UTF8,
+	INVALID_UTF16,
+	CONVERSION_FAILED,
+}
+
+const uint SURROGATE_OFFSET @private = 0x10000;
+const uint SURROGATE_GENERIC_MASK @private = 0xF800;
+const uint SURROGATE_MASK @private = 0xFC00;
+const uint SURROGATE_CODEPOINT_MASK @private = 0x03FF;
+const uint SURROGATE_BITS @private = 10;
+const uint SURROGATE_LOW_VALUE @private = 0xDC00;
+const uint SURROGATE_HIGH_VALUE @private = 0xD800;
+
+fault NumberConversion
+{
+	EMPTY_STRING,
+	NEGATIVE_VALUE,
+	MALFORMED_INTEGER,
+	INTEGER_OVERFLOW,
+	MALFORMED_FLOAT,
+	FLOAT_OUT_OF_RANGE,
+}
+
+
+<*
+ Return a temporary ZString created using the formatting function.
+
+ @param [in] fmt `The formatting string`
+*>
+fn ZString tformat_zstr(String fmt, args...)
+{
+	DString str = dstring::temp_with_capacity(fmt.len + args.len * 8);
+	str.appendf(fmt, ...args);
+	return str.zstr_view();
+}
+
+<*
+ Return a new String created using the formatting function.
+
+ @param [inout] allocator `The allocator to use`
+ @param [in] fmt `The formatting string`
+*>
+fn String format(String fmt, args..., Allocator allocator) => @pool(allocator)
+{
+	DString str = dstring::temp_with_capacity(fmt.len + args.len * 8);
+	str.appendf(fmt, ...args);
+	return str.copy_str(allocator);
+}
+
+<*
+ Return a heap allocated String created using the formatting function.
+
+ @param [in] fmt `The formatting string`
+*>
+fn String new_format(String fmt, args..., Allocator allocator = null) => format(fmt, ...args, allocator: allocator ?: allocator::heap());
+
+<*
+ Return a temporary String created using the formatting function.
+
+ @param [in] fmt `The formatting string`
+*>
+fn String tformat(String fmt, args...)
+{
+	DString str = dstring::temp_with_capacity(fmt.len + args.len * 8);
+	str.appendf(fmt, ...args);
+	return str.str_view();
+}
+
+<*
+ Return a new ZString created using the formatting function.
+
+ @param [in] fmt `The formatting string`
+ @param [inout] allocator `The allocator to use`
+*>
+fn ZString new_format_zstr(String fmt, args..., Allocator allocator = allocator::heap()) => @pool(allocator)
+{
+	DString str = dstring::temp_with_capacity(fmt.len + args.len * 8);
+	str.appendf(fmt, ...args);
+	return str.copy_zstr(allocator);
+}
+
+<*
+ Check if a character is in a set.
+
+ @param c `the character to check`
+ @param [in] set `The formatting string`
+ @pure
+ @return `True if a character is in the set`
+*>
+macro bool char_in_set(char c, String set)
+{
+	foreach (ch : set) if (ch == c) return true;
+	return false;
+}
+
+fn String join_new(String[] s, String joiner, Allocator allocator = allocator::heap())
+{
+	if (!s)
+	{
+		return (String)allocator::new_array(allocator, char, 2)[:0];
+	}
+
+	usz total_size = joiner.len * s.len;
+	foreach (String* &str : s)
+	{
+		total_size += str.len;
+	}
+	@pool(allocator)
+	{
+		DString res = dstring::temp_with_capacity(total_size);
+		res.append(s[0]);
+		foreach (String* &str : s[1..])
+		{
+			res.append(joiner);
+			res.append(*str);
+		}
+		return res.copy_str(allocator);
+	};
+}
+
+<*
+ Remove characters from the front and end of a string.
+
+ @param [in] string `The string to trim`
+ @param [in] to_trim `The set of characters to trim, defaults to whitespace`
+ @pure
+ @return `a substring of the string passed in`
+*>
+fn String String.trim(string, String to_trim = "\t\n\r ")
+{
+    return string.trim_left(to_trim).trim_right(to_trim);
+}
+
+<*
+ Remove characters from the front of a string.
+
+ @param [in] string `The string to trim`
+ @param [in] to_trim `The set of characters to trim, defaults to whitespace`
+ @pure
+ @return `a substring of the string passed in`
+*>
+fn String String.trim_left(string, String to_trim = "\t\n\r ")
+{
+	usz start = 0;
+	usz len = string.len;
+	while (start < len && char_in_set(string[start], to_trim)) start++;
+	if (start == len) return string[:0];
+	return string[start..];
+}
+
+<*
+ Remove characters from the end of a string.
+
+ @param [in] string `The string to trim`
+ @param [in] to_trim `The set of characters to trim, defaults to whitespace`
+ @pure
+ @return `a substring of the string passed in`
+*>
+fn String String.trim_right(string, String to_trim = "\t\n\r ")
+{
+	usz len = string.len;
+	while (len > 0 && char_in_set(string[len - 1], to_trim)) len--;
+	return string[:len];
+}
+
+<*
+ Check if the String starts with the needle.
+
+ @param [in] string
+ @param [in] needle
+ @pure
+ @return `'true' if the string starts with the needle`
+*>
+fn bool String.starts_with(string, String needle)
+{
+	if (needle.len > string.len) return false;
+	if (!needle.len) return true;
+	return string[:needle.len] == needle;
+}
+
+<*
+ Check if the String ends with the needle.
+
+ @param [in] string
+ @param [in] needle
+ @pure
+ @return `'true' if the string ends with the needle`
+*>
+fn bool String.ends_with(string, String needle)
+{
+	if (needle.len > string.len) return false;
+	if (!needle.len) return true;
+	return string[^needle.len..] == needle;
+}
+
+<*
+ Strip the front of the string if the prefix exists.
+
+ @param [in] string
+ @param [in] needle
+ @pure
+ @return `the substring with the prefix removed`
+*>
+fn String String.strip(string, String needle)
+{
+	if (!needle.len || !string.starts_with(needle)) return string;
+	return string[needle.len..];
+}
+
+<*
+ Strip the end of the string if the suffix exists.
+
+ @param [in] string
+ @param [in] needle
+ @pure
+ @return `the substring with the suffix removed`
+*>
+fn String String.strip_end(string, String needle)
+{
+	if (!needle.len || !string.ends_with(needle)) return string;
+	// Note that this is the safe way if we want to support zero length.
+	return string[:(string.len - needle.len)];
+}
+
+<*
+ Split a string into parts, e.g "a|b|c" split with "|" yields { "a", "b", "c" }
+
+ @param [in] s
+ @param [in] needle
+ @param max "Max number of elements, 0 means no limit, defaults to 0"
+ @param skip_empty "True to skip empty elements"
+ @param [&inout] allocator "The allocator to use for the String[]"
+
+ @require needle.len > 0 "The needle must be at least 1 character long"
+ @ensure return.len > 0
+*>
+fn String[] String.split(s, String needle, usz max = 0, Allocator allocator = allocator::heap(), bool skip_empty = false)
+{
+	usz capacity = 16;
+	usz i = 0;
+	String* holder = allocator::alloc_array(allocator, String, capacity);
+	bool no_more = false;
+	while (!no_more)
+	{
+		usz! index = i == max - 1 ? SearchResult.MISSING? : s.index_of(needle);
+		String res @noinit;
+		if (try index)
+		{
+			res = s[:index];
+			s = s[index + needle.len..];
+		}
+		else
+		{
+			res = s;
+			no_more = true;
+		}
+		if (!res.len && skip_empty)
+        {
+        	continue;
+        }
+
+		if (i == capacity)
+		{
+			capacity *= 2;
+			holder = allocator::realloc(allocator, holder, String.sizeof * capacity);
+		}
+		holder[i++] = res;
+	}
+	return holder[:i];
+}
+
+<*
+ Split a string into parts, e.g "a|b|c" split with "|" yields { "a", "b", "c" }, using the heap allocator
+ to store the parts.
+
+ @param [in] s
+ @param [in] needle
+ @param max "Max number of elements, 0 means no limit, defaults to 0"
+ @param skip_empty "True to skip empty elements"
+ @require needle.len > 0 "The needle must be at least 1 character long"
+ @ensure return.len > 0
+*>
+fn String[] String.new_split(s, String needle, usz max = 0, bool skip_empty) => s.split(needle, max, allocator::heap(), skip_empty) @inline;
+
+<*
+ This function is identical to String.split, but implicitly uses the
+ temporary allocator.
+
+ @param [in] s
+ @param [in] needle
+ @param max "Max number of elements, 0 means no limit, defaults to 0"
+ @param skip_empty "True to skip empty elements"
+*>
+fn String[] String.tsplit(s, String needle, usz max = 0, bool skip_empty = false) => s.split(needle, max, allocator::temp(), skip_empty) @inline;
+
+fault SplitResult { BUFFER_EXCEEDED }
+
+<*
+ Split a string into parts, e.g "a|b|c" split with "|" yields { "a", "b", "c" }
+
+ @param [in] s
+ @param [in] needle
+ @param [inout] buffer
+ @param max "Max number of elements, 0 means no limit, defaults to 0"
+ @require needle.len > 0 "The needle must be at least 1 character long"
+ @ensure return.len > 0
+ @return! SplitResult.BUFFER_EXCEEDED `If there are more elements than would fit the buffer`
+*>
+fn String[]! String.split_to_buffer(s, String needle, String[] buffer, usz max = 0, bool skip_empty = false)
+{
+	usz max_capacity = buffer.len;
+	usz i = 0;
+	bool no_more = false;
+	while (!no_more)
+	{
+		usz! index = i == max - 1 ? SearchResult.MISSING? : s.index_of(needle);
+		String res @noinit;
+		if (try index)
+		{
+			res = s[:index];
+			s = s[index + needle.len..];
+		}
+		else
+		{
+			res = s;
+			no_more = true;
+		}
+		if (!res.len && skip_empty)
+		{
+			continue;
+		}
+		if (i == max_capacity)
+		{
+			return SplitResult.BUFFER_EXCEEDED?;
+		}
+		buffer[i++] = res;
+	}
+	return buffer[:i];
+}
+
+<*
+ Check if a substring is found in the string.
+
+ @param [in] s
+ @param [in] needle "The string to look for."
+ @pure
+ @return "true if the string contains the substring, false otherwise"
+*>
+fn bool String.contains(s, String needle)
+{
+	return @ok(s.index_of(needle));
+}
+
+<*
+ Find the index of the first incidence of a string.
+
+ @param [in] s
+ @param needle "The character to look for"
+ @pure
+ @ensure return < s.len
+ @return "the index of the needle"
+ @return! SearchResult.MISSING "if the needle cannot be found"
+*>
+fn usz! String.index_of_char(s, char needle)
+{
+	foreach (i, c : s)
+	{
+		if (c == needle) return i;
+	}
+	return SearchResult.MISSING?;
+}
+
+<*
+ Find the index of the first incidence of a one of the chars.
+
+ @param [in] s
+ @param [in] needle "The characters to look for"
+ @pure
+ @ensure return < s.len
+ @return "the index of the needle"
+ @return! SearchResult.MISSING "if the needle cannot be found"
+*>
+fn usz! String.index_of_chars(String s, char[] needle)
+{
+    foreach (i, c : s)
+    {
+        foreach (j, pin : needle)
+        {
+            if (c == pin) return i;
+        }
+    }
+
+    return SearchResult.MISSING?;
+}
+
+<*
+ Find the index of the first incidence of a character.
+
+ @param [in] s
+ @param needle "The character to look for"
+ @param start_index "The index to start with, may exceed max index."
+ @pure
+ @ensure return < s.len
+ @return "the index of the needle"
+ @return! SearchResult.MISSING "if the needle cannot be found starting from the start_index"
+*>
+fn usz! String.index_of_char_from(s, char needle, usz start_index)
+{
+	usz len = s.len;
+	if (len <= start_index) return SearchResult.MISSING?;
+	for (usz i = start_index; i < len; i++)
+	{
+		if (s[i] == needle) return i;
+	}
+	return SearchResult.MISSING?;
+}
+
+<*
+ Find the index of the first incidence of a character starting from the end.
+
+ @param [in] s
+ @param needle "the character to find"
+ @pure
+ @ensure return < s.len
+ @return "the index of the needle"
+ @return! SearchResult.MISSING "if the needle cannot be found"
+*>
+fn usz! String.rindex_of_char(s, char needle)
+{
+	foreach_r (i, c : s)
+	{
+		if (c == needle) return i;
+	}
+	return SearchResult.MISSING?;
+}
+
+<*
+ Find the index of the first incidence of a string.
+
+ @param [in] s
+ @param [in] needle
+ @pure
+ @ensure return < s.len
+ @require needle.len > 0 : "The needle must be len 1 or more"
+ @return "the index of the needle"
+ @return! SearchResult.MISSING "if the needle cannot be found"
+*>
+fn usz! String.index_of(s, String needle)
+{
+	usz needed = needle.len;
+	if (needed > 0 && s.len >= needed)
+	{
+		char first = needle[0];
+		foreach (i, c: s[..^needed])
+		{
+			if (c == first && s[i:needed] == needle) return i;
+		}
+	}
+	return SearchResult.MISSING?;
+}
+
+<*
+ Find the index of the last incidence of a string.
+
+ @param [in] s
+ @param [in] needle
+ @pure
+ @ensure return < s.len
+ @require needle.len > 0 "The needle must be len 1 or more"
+ @return "the index of the needle"
+ @return! SearchResult.MISSING "if the needle cannot be found"
+*>
+fn usz! String.rindex_of(s, String needle)
+{
+	usz needed = needle.len;
+	if (needed > 0 && s.len >= needed)
+	{
+		char first = needle[0];
+		foreach_r (i, c: s[..^needed])
+		{
+			if (c == first && s[i:needed] == needle) return i;
+		}
+	}
+	return SearchResult.MISSING?;
+}
+
+fn String ZString.str_view(str)
+{
+	return (String)(str[:str.len()]);
+}
+
+fn usz ZString.char_len(str)
+{
+	usz len = 0;
+	char* ptr = (char*)str;
+	while (char c = ptr++[0])
+	{
+		if (c & 0xC0 != 0x80) len++;
+	}
+	return len;
+}
+
+fn usz ZString.len(str)
+{
+	usz len = 0;
+	char* ptr = (char*)str;
+	while (char c = ptr++[0]) len++;
+	return len;
+}
+
+
+fn ZString String.zstr_copy(s, Allocator allocator = allocator::heap())
+{
+	usz len = s.len;
+	char* str = allocator::malloc(allocator, len + 1);
+	mem::copy(str, s.ptr, len);
+	str[len] = 0;
+	return (ZString)str;
+}
+
+fn String String.concat(s1, String s2, Allocator allocator = allocator::heap())
+{
+	usz full_len = s1.len + s2.len;
+	char* str = allocator::malloc(allocator, full_len + 1);
+	usz s1_len = s1.len;
+	mem::copy(str, s1.ptr, s1_len);
+	mem::copy(str + s1_len, s2.ptr, s2.len);
+	str[full_len] = 0;
+	return (String)str[:full_len];
+}
+
+fn String String.tconcat(s1, String s2) => s1.concat(s2, allocator::temp());
+
+
+fn ZString String.zstr_tcopy(s) => s.zstr_copy(allocator::temp()) @inline;
+
+<*
+ Copy this string, by duplicating the string, always adding a zero byte
+ sentinel, so that it safely can be converted to a ZString by a
+ cast.
+*>
+fn String String.copy(s, Allocator allocator = allocator::heap())
+{
+	usz len = s.len;
+	char* str = allocator::malloc(allocator, len + 1);
+	mem::copy(str, s.ptr, len);
+	str[len] = 0;
+	return (String)str[:len];
+}
+
+fn void String.free(&s, Allocator allocator = allocator::heap())
+{
+	if (!s.ptr) return;
+	allocator::free(allocator, s.ptr);
+	*s = "";
+}
+
+fn String String.tcopy(s) => s.copy(allocator::temp()) @inline;
+
+fn String ZString.copy(z, Allocator allocator = allocator::heap())
+{
+	return z.str_view().copy(allocator) @inline;
+}
+
+fn String ZString.tcopy(z)
+{
+	return z.str_view().copy(allocator::temp()) @inline;
+}
+
+<*
+ Convert an UTF-8 string to UTF-16
+ @return "The UTF-16 string as a slice, allocated using the given allocator"
+ @return! UnicodeResult.INVALID_UTF8 "If the string contained an invalid UTF-8 sequence"
+ @return! AllocationFailure "If allocation of the string fails"
+*>
+fn Char16[]! String.to_new_utf16(s, Allocator allocator = allocator::heap())
+{
+	usz len16 = conv::utf16len_for_utf8(s);
+	Char16* data = allocator::alloc_array_try(allocator, Char16, len16 + 1)!;
+	conv::utf8to16_unsafe(s, data)!;
+	data[len16] = 0;
+	return data[:len16];
+}
+
+<*
+ Convert an UTF-8 string to UTF-16
+ @return "The UTF-16 string as a slice, allocated using the given allocator"
+ @return! UnicodeResult.INVALID_UTF8 "If the string contained an invalid UTF-8 sequence"
+ @return! AllocationFailure "If allocation of the string fails"
+*>
+fn Char16[]! String.to_temp_utf16(s)
+{
+	return s.to_new_utf16(allocator::temp());
+}
+
+fn WString! String.to_wstring(s, Allocator allocator)
+{
+	return (WString)s.to_new_utf16(allocator).ptr;
+}
+
+fn WString! String.to_temp_wstring(s) => s.to_wstring(allocator::temp());
+fn WString! String.to_new_wstring(s) => s.to_wstring(allocator::heap());
+
+fn Char32[]! String.to_utf32(s, Allocator allocator)
+{
+	usz codepoints = conv::utf8_codepoints(s);
+	Char32* data = allocator::alloc_array_try(allocator, Char32, codepoints + 1)!;
+	conv::utf8to32_unsafe(s, data)!;
+	data[codepoints] = 0;
+	return data[:codepoints];
+}
+
+fn Char32[]! String.to_new_utf32(s) => s.to_utf32(allocator::heap()) @inline;
+fn Char32[]! String.to_temp_utf32(s) => s.to_utf32(allocator::temp()) @inline;
+
+<*
+ Convert a string to ASCII lower case.
+
+ @param [inout] s
+ @pure
+*>
+fn void String.convert_ascii_to_lower(s)
+{
+	foreach (&c : s) if (c.is_upper() @pure) *c += 'a' - 'A';
+}
+
+fn String String.new_ascii_to_lower(s, Allocator allocator = allocator::heap())
+{
+	String copy = s.copy(allocator);
+	copy.convert_ascii_to_lower();
+	return copy;
+}
+
+fn String String.temp_ascii_to_lower(s)
+{
+	return s.new_ascii_to_lower(allocator::temp());
+}
+
+<*
+ Convert a string to ASCII upper case.
+
+ @param [inout] s
+ @pure
+*>
+fn void String.convert_ascii_to_upper(s)
+{
+	foreach (&c : s) if (c.is_lower() @pure) *c -= 'a' - 'A';
+}
+
+<*
+ Returns a string converted to ASCII upper case.
+
+ @param [in] s
+ @param [inout] allocator
+
+ @return `a new String converted to ASCII upper case.`
+*>
+fn String String.new_ascii_to_upper(s, Allocator allocator = allocator::heap())
+{
+	String copy = s.copy(allocator);
+	copy.convert_ascii_to_upper();
+	return copy;
+}
+
+fn StringIterator String.iterator(s)
+{
+	return { s, 0 };
+}
+
+<*
+ @param [in] s
+ @return `a temporary String converted to ASCII upper case.`
+*>
+fn String String.temp_ascii_to_upper(s)
+{
+	return s.new_ascii_to_upper(allocator::temp());
+}
+
+fn String! new_from_utf32(Char32[] utf32, Allocator allocator = allocator::heap())
+{
+	usz len = conv::utf8len_for_utf32(utf32);
+	char* data = allocator::malloc_try(allocator, len + 1)!;
+	defer catch allocator::free(allocator, data);
+	conv::utf32to8_unsafe(utf32, data);
+	data[len] = 0;
+	return (String)data[:len];
+}
+
+fn String! new_from_utf16(Char16[] utf16, Allocator allocator = allocator::heap())
+{
+	usz len = conv::utf8len_for_utf16(utf16);
+	char* data = allocator::malloc_try(allocator, len + 1)!;
+	defer catch allocator::free(allocator, data);
+	conv::utf16to8_unsafe(utf16, data)!;
+	data[len] = 0;
+	return (String)data[:len];
+}
+
+fn String! new_from_wstring(WString wstring, Allocator allocator = allocator::heap())
+{
+	usz utf16_len;
+	while (wstring[utf16_len] != 0) utf16_len++;
+	Char16[] utf16 = wstring[:utf16_len];
+	return new_from_utf16(utf16, allocator);
+}
+
+fn String! temp_from_wstring(WString wstring) => new_from_wstring(wstring, allocator::temp()) @inline;
+fn String! temp_from_utf16(Char16[] utf16) => new_from_utf16(utf16, allocator::temp()) @inline;
+
+fn usz String.utf8_codepoints(s)
+{
+	usz len = 0;
+	foreach (char c : s)
+	{
+		if (c & 0xC0 != 0x80) len++;
+	}
+	return len;
+}
+
+
+<*
+ @require (base <= 10 && base > 1) || base == 16 : "Unsupported base"
+*>
+macro String.to_integer(string, $Type, int base = 10)
+{
+	usz len = string.len;
+	usz index = 0;
+	char* ptr = string.ptr;
+	while (index < len && ascii::is_blank_m(ptr[index])) index++;
+	if (len == index) return NumberConversion.EMPTY_STRING?;
+	bool is_negative;
+	switch (string[index])
+	{
+		case '-':
+			if ($Type.min == 0) return NumberConversion.NEGATIVE_VALUE?;
+			is_negative = true;
+			index++;
+		case '+':
+			index++;
+		default:
+			break;
+	}
+	if (len == index) return NumberConversion.MALFORMED_INTEGER?;
+	$Type base_used = ($Type)base;
+	if (string[index] == '0' && base == 10)
+	{
+		index++;
+		if (index == len) return ($Type)0;
+		switch (string[index])
+		{
+			case 'x':
+			case 'X':
+				base_used = 16;
+				index++;
+			case 'b':
+			case 'B':
+				base_used = 2;
+				index++;
+			case 'o':
+			case 'O':
+				base_used = 8;
+				index++;
+			default:
+				break;
+		}
+		if (len == index) return NumberConversion.MALFORMED_INTEGER?;
+	}
+	$Type value = 0;
+	while (index != len)
+	{
+		char c = string[index++];
+		switch
+		{
+			case base_used != 16 || c < 'A':  c -= '0';
+			case c <= 'F':                    c -= 'A' - 10;
+			case c < 'a' || c > 'f':          return NumberConversion.MALFORMED_INTEGER?;
+			default:                          c -= 'a' - 10;
+		}
+		if (c >= base_used) return NumberConversion.MALFORMED_INTEGER?;
+		do
+		{
+			if (is_negative)
+			{
+				$Type new_value = value * base_used - c;
+				if (new_value > value) return NumberConversion.INTEGER_OVERFLOW?;
+				value = new_value;
+				break;
+			}
+			$Type new_value = value * base_used + c;
+			if (new_value < value) return NumberConversion.INTEGER_OVERFLOW?;
+			value = new_value;
+		};
+	}
+	return value;
+}
+
+fn int128! String.to_int128(s, int base = 10) => s.to_integer(int128, base);
+fn long! String.to_long(s, int base = 10) => s.to_integer(long, base);
+fn int! String.to_int(s, int base = 10) => s.to_integer(int, base);
+fn short! String.to_short(s, int base = 10) => s.to_integer(short, base);
+fn ichar! String.to_ichar(s, int base = 10) => s.to_integer(ichar, base);
+
+fn uint128! String.to_uint128(s, int base = 10) => s.to_integer(uint128, base);
+fn ulong! String.to_ulong(s, int base = 10) => s.to_integer(ulong, base);
+fn uint! String.to_uint(s, int base = 10) => s.to_integer(uint, base);
+fn ushort! String.to_ushort(s, int base = 10) => s.to_integer(ushort, base);
+fn char! String.to_uchar(s, int base = 10) => s.to_integer(char, base);
+
+fn double! String.to_double(s) => s.to_real(double);
+fn float! String.to_float(s) => s.to_real(float);
+
+fn Splitter String.splitter(self, String split)
+{
+	return { .string = self, .split = split };
+}
+
+fn Splitter String.tokenize(self, String split)
+{
+	return { .string = self, .split = split, .tokenize = true };
+}
+
+struct Splitter
+{
+	String string;
+	String split;
+	usz current;
+	bool tokenize;
+	int last_index;
+}
+
+fn void Splitter.reset(&self)
+{
+	self.current = 0;
+}
+
+fn String! Splitter.next(&self)
+{
+	while (true)
+	{
+		usz len = self.string.len;
+		usz current = self.current;
+		if (current >= len) return IteratorResult.NO_MORE_ELEMENT?;
+		String remaining = self.string[current..];
+		usz! next = remaining.index_of(self.split);
+		if (try next)
+		{
+			self.current = current + next + self.split.len;
+			if (!next && self.tokenize) continue;
+			return remaining[:next];
+		}
+		self.current = len;
+		return remaining;
+	}
+}
+
+macro String new_struct_to_str(x, Allocator allocator = allocator::heap())
+{
+	DString s;
+	@stack_mem(512; Allocator mem)
+	{
+		s.init(mem);
+		io::fprint(&s, x)!!;
+		return s.copy_str(allocator);
+	};
+}
+
+macro String temp_struct_to_str(x) => new_struct_to_str(x, allocator::temp());

lib/std/core/string_iterator.c3

@@ -0,0 +1,49 @@
+module std::core::string::iterator;
+
+struct StringIterator
+{
+	String utf8;
+	usz current;
+}
+
+fn void StringIterator.reset(&self)
+{
+	self.current = 0;
+}
+
+fn Char32! StringIterator.next(&self)
+{
+    usz len = self.utf8.len;
+    usz current = self.current;
+    if (current >= len) return IteratorResult.NO_MORE_ELEMENT?;
+    usz read = (len - current < 4 ? len - current : 4);
+    Char32 res = conv::utf8_to_char32(&self.utf8[current], &read)!;
+    self.current += read;
+    return res;
+}
+
+fn Char32! StringIterator.peek(&self)
+{
+    usz len = self.utf8.len;
+    usz current = self.current;
+    if (current >= len) return IteratorResult.NO_MORE_ELEMENT?;
+    usz read = (len - current < 4 ? len - current : 4);
+    Char32 res = conv::utf8_to_char32(&self.utf8[current], &read)!;
+    return res;
+}
+
+fn bool StringIterator.has_next(&self)
+{
+    return self.current < self.utf8.len;
+}
+
+fn Char32! StringIterator.get(&self)
+{
+    usz len = self.utf8.len;
+    usz current = self.current;
+    usz read = (len - current < 4 ? len - current : 4);
+    usz index = current > read ? current - read : 0;
+    if (index >= len) return IteratorResult.NO_MORE_ELEMENT?;
+    Char32 res = conv::utf8_to_char32(&self.utf8[index], &read)!;
+    return res;
+}

lib/std/core/string_to_real.c3

@@ -0,0 +1,484 @@
+module std::core::string;
+import std::math;
+
+// Float parsing based on code in Musl floatscan.c by Rich Felker.
+// Musl uses the MIT license, copied below:
+// ----------------------------------------------------------------------
+//   Copyright © 2005-2014 Rich Felker, et al.
+//
+//   Permission is hereby granted, free of charge, to any person obtaining
+//   a copy of this software and associated documentation files (the
+//   "Software"), to deal in the Software without restriction, including
+//   without limitation the rights to use, copy, modify, merge, publish,
+//   distribute, sublicense, and/or sell copies of the Software, and to
+//   permit persons to whom the Software is furnished to do so, subject to
+//   the following conditions:
+//
+//   The above copyright notice and this permission notice shall be
+//   included in all copies or substantial portions of the Software.
+//
+//   THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+//   EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+//   MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+//   IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
+//   CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+//   TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+//   SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+//   ----------------------------------------------------------------------
+
+const KMAX = 128;
+const MASK = KMAX - 1;
+const B1B_DIG = 2;
+const uint[2] B1B_MAX = { 9007199, 254740991 };
+
+<*
+ @require chars.len > 0
+*>
+macro double! decfloat(char[] chars, int $bits, int $emin, int sign)
+{
+	uint[KMAX] x;
+	const uint[2] TH = B1B_MAX;
+	int emax = - $emin - $bits + 3;
+
+	const int[?] P10S = { 10, 100, 1000, 10000, 100000, 1000000, 10000000, 100000000 };
+	usz index;
+	bool got_digit = chars[0] == '0';
+	bool got_rad;
+	long lrp, dc;
+	int k, j, lnz;
+	usz len = chars.len;
+	usz last_char = len - 1;
+
+	assert(len);
+
+	char c @noinit;
+	// Skip past first characters
+	while ((c = chars[index]) == '0')
+	{
+		if (index == last_char) return sign * 0.0;
+		index++;
+	}
+
+	if (c == '.')
+	{
+		got_rad = true;
+		if (index == last_char)
+		{
+			if (!got_digit) return NumberConversion.MALFORMED_FLOAT?;
+			return sign * 0.0;
+		}
+		if (index != last_char && (c = chars[++index]) == '0')
+		{
+			lrp--;
+			got_digit = true;
+			while (last_char != index && (c = chars[++index]) == '0')
+			{
+				lrp--;
+			}
+		}
+	}
+
+	while (c - '0' < 10u || c == '.')
+	{
+		switch
+		{
+			case c == '.':
+				if (got_rad) return NumberConversion.MALFORMED_FLOAT?;
+				got_rad = true;
+				lrp = dc;
+			case k < KMAX - 3:
+				dc++;
+				if (c != '0') lnz = (int)dc;
+				if (j)
+				{
+					x[k] = x[k] * 10 + c - '0';
+				}
+				else
+				{
+					x[k] = c - '0';
+				}
+				if (++j == 9)
+				{
+					k++;
+					j = 0;
+				}
+				got_digit = true;
+			default:
+				dc++;
+				if (c != '0') x[KMAX - 4] |= 1;
+
+		}
+		if (index == last_char) break;
+		assert(index < last_char);
+		c = chars[++index];
+	}
+	if (!got_rad) lrp = dc;
+	if (!got_digit) return NumberConversion.MALFORMED_FLOAT?;
+	if ((c | 32) == 'e')
+	{
+		if (last_char == index) return NumberConversion.MALFORMED_FLOAT?;
+		long e10 = String.to_long((String)chars[index + 1..]) ?? NumberConversion.MALFORMED_FLOAT?!;
+		lrp += e10;
+	}
+	else if (index != last_char)
+	{
+		return NumberConversion.MALFORMED_FLOAT?;
+	}
+	// Handle zero specially to avoid nasty special cases later
+	if (!x[0]) return sign * 0.0;
+
+	// Optimize small integers (w/no exponent) and over/under-flow
+	if (lrp == dc && dc < 10 && ($bits > 30 || (ulong)x[0] >> $bits == 0)) return sign * (double)x[0];
+	if (lrp > - $emin / 2) return NumberConversion.FLOAT_OUT_OF_RANGE?;
+	if (lrp < $emin - 2 * math::DOUBLE_MANT_DIG) return NumberConversion.FLOAT_OUT_OF_RANGE?;
+
+	// Align incomplete final B1B digit
+	if (j)
+	{
+		for (; j < 9; j++) x[k] *= 10;
+		k++;
+		j = 0;
+	}
+
+	int a;
+	int z = k;
+	int e2;
+	long rp = lrp;
+
+	// Optimize small to mid-size integers (even in exp. notation)
+	if (lnz < 9 && lnz <= rp && rp < 18)
+	{
+		if (rp == 9) return sign * (double)x[0];
+		if (rp < 9) return sign * (double)x[0] / P10S[8 - rp];
+		int bitlim = $bits - 3 * (int)(rp - 9);
+		if (bitlim > 30 || x[0] >> bitlim == 0) return sign * (double)x[0] * P10S[rp - 10];
+	}
+
+	// Align radix point to B1B digit boundary
+	if (rp % 9)
+	{
+		long rpm9 = rp >= 0 ? rp % 9 : rp % 9 + 9;
+		int p10 = P10S[8 - rpm9];
+		uint carry = 0;
+		for (k = a; k != z; k++)
+		{
+			uint tmp = x[k] % p10;
+			x[k] = x[k] / p10 + carry;
+			carry = 1000000000 / p10 * tmp;
+			if (k == a && !x[k])
+			{
+				a = (a + 1) & MASK;
+				rp -= 9;
+			}
+		}
+		if (carry) x[z++] = carry;
+		rp += 9 - rpm9;
+	}
+
+	// Upscale until desired number of bits are left of radix point
+	while (rp < 9 * B1B_DIG || (rp == 9 * B1B_DIG && x[a] < TH[0]))
+	{
+		uint carry = 0;
+		e2 -= 29;
+		for (k = (z - 1) & MASK; ; k = (k - 1) & MASK)
+		{
+			ulong tmp = (ulong)x[k] << 29 + carry;
+			if (tmp > 1000000000)
+			{
+				carry = (uint)(tmp / 1000000000);
+				x[k] = (uint)(tmp % 1000000000);
+			}
+			else
+			{
+				carry = 0;
+				x[k] = (uint)tmp;
+			}
+			if (k == (z - 1) & MASK && k != a && !x[k]) z = k;
+			if (k == a) break;
+		}
+		if (carry)
+		{
+			rp += 9;
+			a = (a - 1) & MASK;
+			if (a == z)
+			{
+				z = (z - 1) & MASK;
+				x[(z - 1) & MASK] |= x[z];
+			}
+			x[a] = carry;
+		}
+	}
+
+	// Downscale until exactly number of bits are left of radix point
+	while (true)
+	{
+		uint carry = 0;
+		int sh = 1;
+		int i;
+		for (i = 0; i < B1B_DIG; i++)
+		{
+			k = (a + i) & MASK;
+			if (k == z || x[k] < TH[i])
+			{
+				i = B1B_DIG;
+				break;
+			}
+			if (x[(a + i) & MASK] > TH[i]) break;
+		}
+		if (i == B1B_DIG && rp == 9 * B1B_DIG) break;
+		if (rp > 9 + 9 * B1B_DIG) sh = 9;
+		e2 += sh;
+		for (k = a; k != z; k = (k+1) & MASK)
+		{
+			uint tmp = x[k] & (1 << sh - 1);
+			x[k] = x[k] >> sh + carry;
+			carry = (1000000000 >> sh) * tmp;
+			if (k == a && !x[k])
+			{
+				a = (a + 1) & MASK;
+				i--;
+				rp -= 9;
+			}
+		}
+		if (carry)
+		{
+			if ((z + 1) & MASK != a)
+			{
+				x[z] = carry;
+				z = (z + 1) & MASK;
+			}
+			else
+			{
+				x[(z - 1) & MASK] |= 1;
+			}
+		}
+	}
+
+	// Assemble desired bits into floating point variable
+	double y;
+	int i;
+	for (i = 0; i < B1B_DIG; i++)
+	{
+		if ((a + i) & MASK == z) x[(z = (z + 1) & MASK) - 1] = 0;
+		y = 1000000000.0 * y + x[(a + i) & MASK];
+	}
+
+	y *= sign;
+
+	bool denormal;
+	// Limit precision for denormal results
+	uint bits = $bits;
+	if (bits > math::DOUBLE_MANT_DIG + e2 - $emin)
+	{
+		bits = math::DOUBLE_MANT_DIG + e2 - $emin;
+		if (bits < 0) bits = 0;
+		denormal = true;
+	}
+
+	// Calculate bias term to force rounding, move out lower bits
+	double bias;
+	double frac;
+	if (bits < math::DOUBLE_MANT_DIG)
+	{
+		bias = math::copysign(math::scalbn(1, 2 * math::DOUBLE_MANT_DIG - bits - 1), y);
+		frac = y % math::scalbn(1, math::DOUBLE_MANT_DIG - bits);
+		y -= frac;
+		y += bias;
+	}
+
+	// Process tail of decimal input so it can affect rounding
+	if ((a + i) & MASK != z)
+	{
+		uint t = x[(a + i) & MASK];
+		switch
+		{
+			case t < 500000000 && (t || (a + i + 1) & MASK != z):
+				frac += 0.25 * sign;
+			case t > 500000000:
+				frac += 0.75 * sign;
+			case t == 500000000:
+				if ((a + i + 1) & MASK == z)
+				{
+					frac += 0.5 * sign;
+				}
+				else
+				{
+					frac += 0.75 * sign;
+				}
+		}
+		if (math::DOUBLE_MANT_DIG - bits >= 2 && !(frac % 1)) frac++;
+	}
+
+	y += frac;
+	y -= bias;
+
+	if (((e2 + math::DOUBLE_MANT_DIG) & int.max) > emax - 5)
+	{
+		if (math::abs(y) >= 0x1p53)
+		{
+			if (denormal && bits == math::DOUBLE_MANT_DIG + e2 - $emin) denormal = false;
+			y *= 0.5;
+			e2++;
+		}
+		if (e2 + math::DOUBLE_MANT_DIG > emax || (denormal && frac)) return NumberConversion.MALFORMED_FLOAT?;
+	}
+	return math::scalbn(y, e2);
+}
+
+macro double! hexfloat(char[] chars, int $bits, int $emin, int sign)
+{
+	double scale = 1;
+	uint x;
+	long rp;
+	long dc;
+	char c @noinit;
+	bool got_rad;
+	bool got_digit;
+	bool got_tail;
+	usz len = chars.len;
+	usz last_char = len - 1;
+	usz index;
+	double y;
+
+	// Skip past first characters
+	while ((c = chars[index]) == '0')
+	{
+		if (index == last_char) return 0.0;
+		index++;
+	}
+	if (c == '.')
+	{
+		got_rad = true;
+		if (index == last_char)
+		{
+			if (!got_digit) return NumberConversion.MALFORMED_FLOAT?;
+			return sign * 0.0;
+		}
+		if (index != last_char && (c = chars[++index]) == '0')
+		{
+			rp--;
+			got_digit = true;
+			while (last_char != index && (c = chars[++index]) == '0')
+			{
+				rp--;
+			}
+		}
+	}
+
+	while ((c - '0') < 10u || ((c | 32) - 'a') < 6u || c == '.')
+	{
+		if (c == '.')
+		{
+			if (got_rad) return NumberConversion.MALFORMED_FLOAT?;
+			got_rad = true;
+			rp = dc;
+		}
+		else
+		{
+			got_digit = true;
+			int d = c > '9' ? ((c | 32) + 10 - 'a') : (c - '0');
+			switch
+			{
+				case dc < 8:
+					x = x * 16 + d;
+				case dc < math::DOUBLE_MANT_DIG / 4 + 1:
+					y += d * (scale /= 16);
+					got_tail = true;
+				case d && !got_tail:
+					y += 0.5 * scale;
+					got_tail = true;
+			}
+			dc++;
+		}
+		if (index == last_char) break;
+		c = chars[++index];
+	}
+	if (!got_digit) return NumberConversion.MALFORMED_FLOAT?;
+	if (!got_rad) rp = dc;
+	for (; dc < 8; dc++) x *= 16;
+
+	long e2;
+	if ((c | 32) == 'p')
+	{
+		long e2val = String.to_long((String)chars[index + 1..]) ?? (NumberConversion.MALFORMED_FLOAT?)!;
+		e2 = e2val;
+	}
+	e2 += 4 * rp - 32;
+	if (!x) return sign * 0.0;
+	if (e2 > -$emin) return NumberConversion.FLOAT_OUT_OF_RANGE?;
+	if (e2 < $emin - 2 * math::DOUBLE_MANT_DIG) return NumberConversion.FLOAT_OUT_OF_RANGE?;
+
+	while (x < 0x80000000)
+	{
+		if (y >= 0.5)
+		{
+			x += x + 1;
+			y += y - 1;
+		}
+		else
+		{
+			x += x;
+			y += y;
+		}
+		e2--;
+	}
+	int bits = $bits;
+	if ($bits > 32 + e2 - $emin)
+	{
+		bits = (int)(32 + e2 - $emin);
+		if (bits < 0) bits = 0;
+	}
+	double bias;
+	if (bits < math::DOUBLE_MANT_DIG)
+	{
+		bias = math::copysign(math::scalbn(1, 32 + math::DOUBLE_MANT_DIG - bits - 1), (double)sign);
+	}
+
+	if (bits < 32 && y && !(x & 1))
+	{
+		x++;
+		y = 0;
+	}
+	y = bias + sign * (double)x + sign * y;
+	y -= bias;
+	if (!y) return NumberConversion.FLOAT_OUT_OF_RANGE?;
+
+	return math::scalbn(y, (int)e2);
+}
+
+macro String.to_real(chars, $Type) @private
+{
+	int sign = 1;
+	$switch ($Type)
+	$case float:
+		const int BITS = math::FLOAT_MANT_DIG;
+		const int EMIN = math::FLOAT_MIN_EXP - BITS;
+	$case double:
+		const int BITS = math::DOUBLE_MANT_DIG;
+		const int EMIN = math::DOUBLE_MIN_EXP - BITS;
+	$case float128:
+		$error "Not yet supported";
+	$default:
+		$error "Unexpected type";
+	$endswitch
+
+	while (chars.len && chars[0] == ' ') chars = chars[1..];
+	if (!chars.len) return NumberConversion.MALFORMED_FLOAT?;
+	switch (chars[0])
+	{
+		case '-':
+			sign = -1;
+			nextcase;
+		case '+':
+			chars = chars[1..];
+	}
+	if (chars == "infinity" || chars == "INFINITY") return sign * $Type.inf;
+	if (chars == "NAN" || chars == "nan") return $Type.nan;
+
+	if (chars.len > 2 && chars[0] == '0' && (chars[1] | 32) == 'x')
+	{
+		return ($Type)hexfloat((char[])chars[2..], BITS, EMIN, sign);
+	}
+	return ($Type)decfloat((char[])chars, BITS, EMIN, sign);
+}
+

lib/std/core/test.c3

@@ -0,0 +1,223 @@
+<*
+Unit test module
+
+This module provides a toolset of macros for running unit test checks 
+
+Example:
+```c3
+module sample::m;
+import std::io;
+
+fault MathError
+{
+	DIVISION_BY_ZERO
+}
+
+fn double! divide(int a, int b)
+{
+	if (b == 0) return MathError.DIVISION_BY_ZERO?;
+	return (double)(a) / (double)(b);
+}
+
+fn void! test_div() @test
+{
+	test::eq(2, divide(6, 3)!);
+	test::ne(1, 2);
+	test::ge(3, 3);
+	test::gt(2, divide(3, 3)!);
+	test::lt(2, 3);
+	test::le(2, 3);
+	test::eq_approx(m::divide(1, 3)!, 0.333, places: 3);
+	test::@check(2 == 2, "divide: %d", divide(6, 3)!);
+	test::@error(m::divide(3, 0), MathError.DIVISION_BY_ZERO);
+}
+
+```
+*>
+// Copyright (c) 2025 Alex Veden <i@alexveden.com>. All rights reserved.
+// Use of this source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+module std::core::test;
+import std::core::runtime @public;
+import std::math, std::io, libc;
+
+<*
+ Initializes test case context.
+
+ @param setup_fn `initializer function for test case`
+ @param teardown_fn `cleanup function for test context (may be null)`
+
+ @require runtime::test_context != null "Only allowed in @test functions"
+ @require setup_fn != null "setup_fn must always be set"
+*>
+macro @setup(TestFn setup_fn, TestFn teardown_fn = null)
+{
+	runtime::test_context.setup_fn = setup_fn;
+	runtime::test_context.teardown_fn = teardown_fn;
+	runtime::test_context.setup_fn();
+}
+
+<*
+ Checks condition and fails assertion if not true
+
+ @param #condition `any boolean condition, will be expanded by text`
+ @param format `printf compatible format`
+ @param args `vargs for format`
+ @require runtime::test_context != null "Only allowed in @test functions"
+*>
+macro @check(#condition, String format = "", args...)
+{
+	if (!#condition)
+	{
+		@stack_mem(512; Allocator allocator)
+		{
+			DString s;
+			s.init(allocator);
+			s.appendf("check `%s` failed. ", $stringify(#condition));
+			s.appendf(format, ...args);
+			print_panicf(s.str_view());
+		};
+	}
+}
+
+<*
+ Check if function returns specific error
+
+ @param #funcresult `result of function execution`
+ @param error_expected `expected error of function execution`
+ @require runtime::test_context != null "Only allowed in @test functions"
+*>
+macro @error(#funcresult, anyfault error_expected)
+{
+	if (catch err = #funcresult)
+	{
+		if (err != error_expected)
+		{
+			print_panicf("`%s` expected to return error [%s], got [%s]",
+				$stringify(#funcresult), error_expected, err);
+		}
+		return;
+	}
+	print_panicf("`%s` error [%s] was not returned.", $stringify(#funcresult), error_expected);
+}
+
+<*
+ Check if left == right
+
+ @param left `left argument of any comparable type`
+ @param right `right argument of any comparable type`
+ @require runtime::test_context != null "Only allowed in @test functions"
+*>
+macro eq(left, right)
+{
+	if (!equals(left, right))
+	{
+		print_panicf("`%s` != `%s`", left, right);
+	}
+}
+
+<*
+ Check left floating point value is approximately equals to right value
+
+ @param places `number of decimal places to compare (default: 7)`
+ @param delta `minimal allowed difference (overrides places parameter)`
+ @param equal_nan `allows comparing nan values, if left and right both nans result is ok`
+
+ @require places > 0, places <= 20 "too many decimal places"
+ @require delta >= 0, delta <= 1 "delta must be a small number"
+ @require runtime::test_context != null "Only allowed in @test functions"
+*>
+macro void eq_approx(double left, double right, uint places = 7, double delta = 0, bool equal_nan = true)
+{
+	double diff = left - right;
+	double eps = delta;
+	if (eps == 0) eps = 1.0 / math::pow(10.0, places);
+
+	if (!math::is_approx(left, right, eps))
+	{
+		if (equal_nan && math::is_nan(left) && math::is_nan(right)) return;
+		print_panicf("Not almost equal: `%s` !~~ `%s` delta=%e diff: %e", left, right, eps, diff);
+	}
+}
+
+<*
+ Check if left != right
+
+ @param left `left argument of any comparable type`
+ @param right `right argument of any comparable type`
+ @require runtime::test_context != null "Only allowed in @test functions"
+*>
+macro void ne(left, right)
+{
+	if (equals(left, right))
+	{
+		print_panicf("`%s` == `%s`", left, right);
+	}
+}
+
+<*
+ Check if left > right
+
+ @param left `left argument of any comparable type`
+ @param right `right argument of any comparable type`
+ @require runtime::test_context != null "Only allowed in @test functions"
+*>
+macro gt(left, right)
+{
+	if (!builtin::greater(left, right))
+	{
+		print_panicf("`%s` <= `%s`", left, right);
+	}
+}
+
+<*
+ Check if left >= right
+
+ @param left `left argument of any comparable type`
+ @param right `right argument of any comparable type`
+ @require runtime::test_context != null "Only allowed in @test functions"
+*>
+macro ge(left, right)
+{
+	if (!builtin::greater_eq(left, right))
+	{
+		print_panicf("`%s` < `%s`", left, right);
+	}
+}
+
+<*
+ Check if left < right
+
+ @param left `left argument of any comparable type`
+ @param right `right argument of any comparable type`
+ @require runtime::test_context != null "Only allowed in @test functions"
+*>
+macro lt(left, right)
+{
+	if (!builtin::less(left, right))
+	{
+		print_panicf("`%s` >= `%s`", left, right);
+	}
+}
+
+<*
+ Check if left <= right
+
+ @param left `left argument of any comparable type`
+ @param right `right argument of any comparable type`
+ @require runtime::test_context != null "Only allowed in @test functions"
+*>
+macro le(left, right)
+{
+	if (!builtin::less_eq(left, right))
+	{
+		print_panicf("`%s` > `%s`", left, right);
+	}
+}
+
+macro void print_panicf(format, ...) @local
+{
+	runtime::test_context.assert_print_backtrace = false;
+	builtin::panicf(format, $$FILE, $$FUNC, $$LINE, $vasplat);
+}
+

lib/std/core/types.c3

@@ -0,0 +1,380 @@
+
+module std::core::types;
+import libc;
+
+
+fault ConversionResult
+{
+	VALUE_OUT_OF_RANGE,
+	VALUE_OUT_OF_UNSIGNED_RANGE,
+}
+
+<*
+ @require $Type.kindof.is_int() "Type was not an integer"
+ @require v.type.kindof == ENUM "Value was not an enum"
+*>
+macro any_to_enum_ordinal(any v, $Type)
+{
+	return any_to_int(v.as_inner(), $Type);
+}
+
+<*
+ @require $Type.kindof.is_int() "Type was not an integer"
+ @require v.type.kindof.is_int() "Value was not an integer"
+*>
+macro any_to_int(any v, $Type)
+{
+	typeid any_type = v.type;
+	TypeKind kind = any_type.kindof;
+	bool is_mixed_signed = $Type.kindof != any_type.kindof;
+	$Type max = $Type.max;
+	$Type min = $Type.min;
+	switch (any_type)
+	{
+		case ichar:
+			ichar c = *(char*)v.ptr;
+			if (is_mixed_signed && c < 0) return ConversionResult.VALUE_OUT_OF_UNSIGNED_RANGE?;
+			return ($Type)c;
+		case short:
+			short s = *(short*)v.ptr;
+			if (is_mixed_signed && s < 0) return ConversionResult.VALUE_OUT_OF_UNSIGNED_RANGE?;
+			if (s > max || s < min) return ConversionResult.VALUE_OUT_OF_RANGE?;
+			return ($Type)s;
+		case int:
+			int i = *(int*)v.ptr;
+			if (is_mixed_signed && i < 0) return ConversionResult.VALUE_OUT_OF_UNSIGNED_RANGE?;
+			if (i > max || i < min) return ConversionResult.VALUE_OUT_OF_RANGE?;
+			return ($Type)i;
+		case long:
+			long l = *(long*)v.ptr;
+			if (is_mixed_signed && l < 0) return ConversionResult.VALUE_OUT_OF_UNSIGNED_RANGE?;
+			if (l > max || l < min) return ConversionResult.VALUE_OUT_OF_RANGE?;
+			return ($Type)l;
+		case int128:
+			int128 i = *(int128*)v.ptr;
+			if (is_mixed_signed && i < 0) return ConversionResult.VALUE_OUT_OF_UNSIGNED_RANGE?;
+			if (i > max || i < min) return ConversionResult.VALUE_OUT_OF_RANGE?;
+			return ($Type)i;
+		case char:
+			char c = *(char*)v.ptr;
+			if (c > max) return ConversionResult.VALUE_OUT_OF_RANGE?;
+			return ($Type)c;
+		case ushort:
+			ushort s = *(ushort*)v.ptr;
+			if (s > max || s < min) return ConversionResult.VALUE_OUT_OF_RANGE?;
+			return ($Type)s;
+		case uint:
+			uint i = *(uint*)v.ptr;
+			if (i > max || i < min) return ConversionResult.VALUE_OUT_OF_RANGE?;
+			return ($Type)i;
+		case ulong:
+			ulong l = *(ulong*)v.ptr;
+			if (l > max || l < min) return ConversionResult.VALUE_OUT_OF_RANGE?;
+			return ($Type)l;
+		case uint128:
+			uint128 i = *(uint128*)v.ptr;
+			if (i > max || i < min) return ConversionResult.VALUE_OUT_OF_RANGE?;
+			return ($Type)i;
+		default:
+			unreachable();
+	}
+}
+
+fn bool typeid.is_subtype_of(self, typeid other)
+{
+	while (self != void.typeid)
+	{
+		if (self == other) return true;
+		self = self.parentof;
+	}
+	return false;
+}
+
+macro bool is_subtype_of($Type, $OtherType)
+{
+	var $typeid = $Type.typeid;
+	$switch ($Type)
+		$case $OtherType: return true;
+		$default: return false;
+	$endswitch
+}
+macro bool is_numerical($Type)
+{
+	var $kind = $Type.kindof;
+	$if $kind == TypeKind.DISTINCT:
+		return is_numerical($typefrom($Type.inner));
+	$else
+		return $kind == TypeKind.SIGNED_INT || $kind == TypeKind.UNSIGNED_INT || $kind == TypeKind.FLOAT
+			|| $kind == TypeKind.VECTOR;
+	$endif
+}
+
+fn bool TypeKind.is_int(kind) @inline
+{
+	return kind == TypeKind.SIGNED_INT || kind == TypeKind.UNSIGNED_INT;
+}
+
+macro bool is_slice_convertable($Type)
+{
+	$switch ($Type.kindof)
+		$case SLICE:
+			return true;
+		$case POINTER:
+			return $Type.inner.kindof == TypeKind.ARRAY;
+		$default:
+			return false;
+	$endswitch
+}
+
+macro bool is_bool($Type) @const => $Type.kindof == TypeKind.BOOL;
+macro bool is_int($Type) @const  => $Type.kindof == TypeKind.SIGNED_INT || $Type.kindof == TypeKind.UNSIGNED_INT;
+
+<*
+ @require is_numerical($Type) "Expected a numerical type"
+*>
+macro bool is_signed($Type) @const
+{
+	$switch (inner_kind($Type))
+		$case SIGNED_INT:
+		$case FLOAT:
+			return true;
+		$case VECTOR:
+			return is_signed($typefrom($Type.inner));
+		$default:
+			return false;
+	$endswitch
+}
+
+<*
+ @require is_numerical($Type) "Expected a numerical type"
+*>
+macro bool is_unsigned($Type) @const
+{
+	$switch (inner_kind($Type))
+		$case UNSIGNED_INT:
+			return true;
+		$case VECTOR:
+			return is_unsigned($typefrom($Type.inner));
+		$default:
+			return false;
+	$endswitch
+}
+
+macro bool is_indexable($Type) @const
+{
+	return $defined(($Type){}[0]);
+}
+
+macro bool is_ref_indexable($Type) @const
+{
+	return $defined(&($Type){}[0]);
+}
+
+macro bool is_intlike($Type) @const
+{
+	$switch ($Type.kindof)
+		$case SIGNED_INT:
+		$case UNSIGNED_INT:
+			return true;
+		$case VECTOR:
+			return $Type.inner.kindof == TypeKind.SIGNED_INT || $Type.inner.kindof == TypeKind.UNSIGNED_INT;
+		$default:
+			return false;
+	$endswitch
+}
+
+macro bool is_underlying_int($Type) @const
+{
+	$switch ($Type.kindof)
+		$case SIGNED_INT:
+		$case UNSIGNED_INT:
+			return true;
+		$case DISTINCT:
+			return is_underlying_int($typefrom($Type.inner));
+		$default:
+			return false;
+	$endswitch
+}
+
+macro bool is_float($Type) @const  => $Type.kindof == TypeKind.FLOAT;
+
+macro bool is_floatlike($Type) @const
+{
+	$switch ($Type.kindof)
+		$case FLOAT:
+			return true;
+		$case VECTOR:
+			return $Type.inner.kindof == TypeKind.FLOAT;
+		$default:
+			return false;
+	$endswitch
+}
+
+macro bool is_vector($Type) @const
+{
+	return $Type.kindof == TypeKind.VECTOR;
+}
+
+macro typeid inner_type($Type) @const
+{
+	$if $Type.kindof == TypeKind.DISTINCT:
+		return inner_type($typefrom($Type.inner));
+	$else
+		return $Type.typeid;
+	$endif
+}
+
+macro TypeKind inner_kind($Type) @const
+{
+	return inner_type($Type).kindof;
+}
+
+macro bool is_same($TypeA, $TypeB) @const
+{
+	return $TypeA.typeid == $TypeB.typeid;
+}
+
+macro bool @has_same(#a, #b, ...) @const
+{
+	var $type_a = @typeid(#a);
+	$if $type_a != @typeid(#b):
+		return false;
+	$endif
+	$for (var $i = 0; $i < $vacount; $i++)
+		$if @typeid($vaexpr[$i]) != $type_a:
+			return false;
+		$endif
+	$endfor
+	return true;
+}
+
+macro bool may_load_atomic($Type) @const
+{
+	$switch ($Type.kindof)
+	$case BOOL:
+	$case SIGNED_INT:
+	$case UNSIGNED_INT:
+	$case POINTER:
+	$case FLOAT:
+		return true;
+	$case DISTINCT:
+		return may_load_atomic($Type.inner);
+	$default:
+		return false;
+	$endswitch
+}
+
+macro lower_to_atomic_compatible_type($Type) @const
+{
+	$switch ($Type.kindof)
+	$case SIGNED_INT:
+	  $case UNSIGNED_INT:
+	  return $Type.typeid;
+		$case DISTINCT:
+			return lower_to_atomic_compatible_type($Type.inner);
+		$case FLOAT:
+			$switch ($Type)
+				$case float16: 
+					return ushort.typeid;
+				$case float: 
+					return uint.typeid;
+				$case double:
+					return ulong.typeid;
+				$case float128:
+					return uint128.typeid;
+				$default: 
+					return void.typeid;
+			$endswitch
+		$default:
+			return void.typeid;
+  $endswitch
+}
+
+macro bool is_promotable_to_floatlike($Type) @const  => types::is_floatlike($Type) || types::is_int($Type);
+macro bool is_promotable_to_float($Type) @const  => types::is_float($Type) || types::is_int($Type);
+
+macro bool is_same_vector_type($Type1, $Type2) @const
+{
+	$if $Type1.kindof != TypeKind.VECTOR:
+		return $Type2.kindof != TypeKind.VECTOR;
+	$else
+		return $Type1.inner == $Type2.inner && $Type1.len == $Type2.len;
+	$endif
+}
+
+macro bool is_equatable_type($Type) @const
+{
+	$if $defined($Type.less) || $defined($Type.compare_to) || $defined($Type.equals):
+		return true;
+	$else
+		return $Type.is_eq;
+	$endif
+}
+
+<*
+ Checks if a type implements the copy protocol.
+*>
+macro bool implements_copy($Type) @const
+{
+	return $defined($Type.copy) && $defined($Type.free);
+}
+
+macro bool is_equatable_value(value) @deprecated
+{
+	return is_equatable_type($typeof(value));
+}
+
+macro bool @equatable_value(#value) @const
+{
+	return is_equatable_type($typeof(#value));
+}
+
+macro bool @comparable_value(#value) @const
+{
+	$if $defined(#value.less) || $defined(#value.compare_to):
+		return true;
+	$else
+		return $typeof(#value).is_ordered;
+	$endif
+}
+
+macro bool is_comparable_value(value) @deprecated
+{
+	$if $defined(value.less) || $defined(value.compare_to):
+		return true;
+	$else
+		return $typeof(value).is_ordered;
+	$endif
+}
+
+enum TypeKind : char
+{
+	VOID,
+	BOOL,
+	SIGNED_INT,
+	UNSIGNED_INT,
+	FLOAT,
+	TYPEID,
+	ANYFAULT,
+	ANY,
+	ENUM,
+	FAULT,
+	STRUCT,
+	UNION,
+	BITSTRUCT,
+	FUNC,
+	OPTIONAL,
+	ARRAY,
+	SLICE,
+	VECTOR,
+	DISTINCT,
+	POINTER,
+	INTERFACE,
+}
+
+struct TypeEnum
+{
+	TypeKind type;
+	usz elements;
+}
+

lib/std/core/values.c3

@@ -0,0 +1,66 @@
+module std::core::values;
+
+macro typeid @typeid(#value) @const @builtin => $typeof(#value).typeid;
+macro TypeKind @typekind(#value) @const @builtin => $typeof(#value).kindof;
+macro bool @typeis(#value, $Type) @const @builtin => $typeof(#value).typeid == $Type.typeid;
+<*
+ Return true if two values have the same type before any conversions.
+*>
+macro bool @is_same_type(#value1, #value2) @const => $typeof(#value1).typeid == $typeof(#value2).typeid;
+macro bool @is_bool(#value) @const => types::is_bool($typeof(#value));
+macro bool @is_int(#value) @const => types::is_int($typeof(#value));
+macro bool @is_floatlike(#value) @const => types::is_floatlike($typeof(#value));
+macro bool @is_float(#value) @const => types::is_float($typeof(#value));
+macro bool @is_promotable_to_floatlike(#value) @const => types::is_promotable_to_floatlike($typeof(#value));
+macro bool @is_promotable_to_float(#value) @const => types::is_promotable_to_float($typeof(#value));
+macro bool @is_vector(#value) @const => types::is_vector($typeof(#value));
+macro bool @is_same_vector_type(#value1, #value2) @const => types::is_same_vector_type($typeof(#value1), $typeof(#value2));
+macro bool @assign_to(#value1, #value2) @const => $assignable(#value1, $typeof(#value2));
+macro bool @is_lvalue(#value) => $defined(#value = #value);
+
+macro promote_int(x)
+{
+	$if @is_int(x):
+		return (double)x;
+	$else
+		return x;
+	$endif
+}
+
+<*
+ Select between two values at compile time,
+ the values do not have to be of the same type.
+
+ This acts like `$bool ? #value_1 : #value_2` but at compile time.
+
+ @param $bool `true for picking the first value, false for the other`
+ @param #value_1
+ @param #value_2
+ @returns `The selected value.`
+*>
+macro @select(bool $bool, #value_1, #value_2) @builtin
+{
+	$if $bool:
+		return #value_1;
+	$else
+		return #value_2;
+	$endif
+}
+macro promote_int_same(x, y)
+{
+	$if @is_int(x):
+		$switch
+		$case @is_vector(y) &&& $typeof(y).inner == float.typeid:
+			return (float)x;
+		$case $typeof(y).typeid == float.typeid:
+			return (float)x;
+		$default:
+			return (double)x;
+		$endswitch
+	$else
+		return x;
+	$endif
+}
+
+macro TypeKind @inner_kind(#value) @const => types::inner_kind($typeof(#value));
+

lib/std/crypto/crypto.c3

@@ -0,0 +1,12 @@
+module std::crypto;
+
+fn bool safe_compare(void* data1, void* data2, usz len)
+{
+	char match = 0;
+	for (usz i = 0; i < len; i++)
+	{
+		match = match | (mem::@volatile_load(((char*)data1)[i]) ^ mem::@volatile_load(((char*)data2)[i]));
+	}
+	return match == 0;
+}
+

lib/std/crypto/dh.c3

@@ -0,0 +1,12 @@
+module std::crypto::dh;
+import std::math::bigint;
+
+fn BigInt generate_secret(BigInt p, BigInt x, BigInt y)
+{
+	return y.mod_pow(x, p);
+}
+
+fn BigInt public_key(BigInt p, BigInt g, BigInt x)
+{
+	return g.mod_pow(x, p);
+}

lib/std/crypto/rc4.c3

@@ -0,0 +1,75 @@
+module std::crypto::rc4;
+// Copyright (c) 2021 Christoffer Lerno. All rights reserved.
+// Use of this source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+
+struct Rc4
+{
+	uint i, j;
+	char[256] state;
+}
+
+<*
+ Initialize the RC4 state.
+
+ @param [in] key "The key to use"
+ @require key.len > 0 "The key must be at least 1 byte long"
+*>
+fn void Rc4.init(&self, char[] key)
+{
+	// Init the state matrix
+	foreach (char i, &c : self.state) *c = i;
+	for (int i = 0, int j = 0; i < 256; i++)
+	{
+		j = (j + self.state[i] + key[i % key.len]) & 0xFF;
+		@swap(self.state[i], self.state[j]);
+	}
+	self.i = 0;
+	self.j = 0;
+}
+
+<*
+ Run a single pass of en/decryption using a particular key.
+ @param [in] key
+ @param [inout] data
+*>
+fn void crypt(char[] key, char[] data)
+{
+	Rc4 rc4;
+	rc4.init(key);
+	rc4.crypt(data, data);
+}
+
+<*
+ Encrypt or decrypt a sequence of bytes.
+
+ @param [in] in "The input"
+ @param [out] out "The output"
+ @require in.len <= out.len "Output would overflow"
+*>
+fn void Rc4.crypt(&self, char[] in, char[] out)
+{
+	uint i = self.i;
+	uint j = self.j;
+	char* state = &self.state;
+	isz len = in.len;
+	foreach (idx, c : in)
+	{
+		i = (i + 1) & 0xFF;
+		j = (j + state[i]) & 0xFF;
+		@swap(state[i], state[j]);
+		out[idx] = in[idx] ^ state[(state[i] + state[j]) & 0xFF];
+	}
+	self.i = i;
+	self.j = j;
+}
+
+<*
+ Clear the rc4 state.
+
+ @param [&out] self "The RC4 State"
+*>
+fn void Rc4.destroy(&self)
+{
+	*self = {};
+}

lib/std/encoding/base32.c3

@@ -0,0 +1,406 @@
+module std::encoding::base32;
+
+// This module implements base32 encoding according to RFC 4648
+// (https://www.rfc-editor.org/rfc/rfc4648)
+
+struct Base32Alphabet
+{
+	char[32] encoding;
+	char[256] reverse;
+}
+
+const char NO_PAD = 0;
+const char DEFAULT_PAD = '=';
+
+<*
+ Encode the content of src into a newly allocated string
+ @param [in] src "The input to be encoded."
+ @param padding "The padding character or 0 if none"
+ @param alphabet "The alphabet to use"
+ @require padding < 0xFF "Invalid padding character"
+ @return "The encoded string."
+*>
+fn String! encode(char[] src, Allocator allocator, char padding = DEFAULT_PAD, Base32Alphabet* alphabet = &STANDARD)
+{
+	char[] dst = allocator::alloc_array(allocator, char, encode_len(src.len, padding));
+	return encode_buffer(src, dst, padding, alphabet);
+}
+
+<*
+ Decode the content of src into a newly allocated char array.
+ @param [in] src "The input to be encoded."
+ @param padding "The padding character or 0 if none"
+ @param alphabet "The alphabet to use"
+ @require padding < 0xFF "Invalid padding character"
+ @return "The decoded data."
+*>
+fn char[]! decode(char[] src, Allocator allocator, char padding = DEFAULT_PAD, Base32Alphabet* alphabet = &STANDARD)
+{
+	char[] dst = allocator::alloc_array(allocator, char, decode_len(src.len, padding));
+	return decode_buffer(src, dst, padding, alphabet);
+}
+
+fn String! encode_new(char[] code, char padding = DEFAULT_PAD, Base32Alphabet* alphabet = &STANDARD) @inline => encode(code, allocator::heap(), padding, alphabet);
+fn String! encode_temp(char[] code, char padding = DEFAULT_PAD, Base32Alphabet* alphabet = &STANDARD) @inline => encode(code, allocator::temp(), padding, alphabet);
+fn char[]! decode_new(char[] code, char padding = DEFAULT_PAD, Base32Alphabet* alphabet = &STANDARD) @inline => decode(code, allocator::heap(), padding, alphabet);
+fn char[]! decode_temp(char[] code, char padding = DEFAULT_PAD, Base32Alphabet* alphabet = &STANDARD) @inline => decode(code, allocator::temp(), padding, alphabet);
+
+<*
+ Calculate the length in bytes of the decoded data.
+ @param n "Length in bytes of input."
+ @param padding "The padding character or 0 if none"
+ @require padding < 0xFF "Invalid padding character"
+ @return "Length in bytes of the decoded data."
+*>
+fn usz decode_len(usz n, char padding)
+{
+	if (padding) return (n / 8) * 5;
+	// no padding
+	usz trailing = n % 8;
+	return n / 8 * 5 + (trailing * 5 ) / 8;
+}
+
+<*
+ Calculate the length in bytes of the encoded data.
+ @param n "Length in bytes on input."
+ @param padding "The padding character or 0 if none"
+ @require padding < 0xFF "Invalid padding character"
+ @return "Length in bytes of the encoded data."
+*>
+fn usz encode_len(usz n, char padding)
+{
+	// A character is encoded into 8 x 5-bit blocks.
+	if (padding) return (n + 4) / 5 * 8;
+
+	// no padding
+	usz trailing = n % 5;
+	return n / 5 * 8 + (trailing * 8 + 4) / 5;
+}
+
+<*
+ Decode the content of src into dst, which must be properly sized.
+ @param src "The input to be decoded."
+ @param dst "The decoded input."
+ @param padding "The padding character or 0 if none"
+ @param alphabet "The alphabet to use"
+ @require padding < 0xFF "Invalid padding character"
+ @require dst.len >= decode_len(src.len, padding) "Destination buffer too small"
+ @return "The resulting dst buffer"
+ @return! DecodingFailure
+*>
+fn char[]! decode_buffer(char[] src, char[] dst, char padding = DEFAULT_PAD, Base32Alphabet* alphabet = &STANDARD)
+{
+	if (src.len == 0) return dst[:0];
+	char* dst_ptr = dst;
+	usz dn = decode_len(src.len, padding);
+	usz n;
+	char[8] buf;
+	while (src.len > 0 && dst.len > 0)
+	{
+		usz i @noinit;
+		// load 8 bytes into buffer
+		for (i = 0; i < 8; i++)
+		{
+			if (src.len == 0)
+			{
+				if (padding > 0) return DecodingFailure.INVALID_PADDING?;
+				break;
+			}
+			if (src[0] == padding) break;
+			buf[i] = alphabet.reverse[src[0]];
+			if (buf[i] == INVALID) return DecodingFailure.INVALID_CHARACTER?;
+			src = src[1..];
+		}
+
+		// extract 5-bytes from the buffer which contains 8 x 5 bit chunks
+		switch (i)
+		{
+			case 8:
+				// |66677777|  dst[4]
+				// |   77777|  buf[7]
+				// |666     |  buf[6] << 5
+				dst[4] = buf[7] | buf[6] << 5;
+				n++;
+				nextcase 7;
+			case 7:
+				// |45555566|  dst[3]
+				// |      66|  buf[6] >> 3
+				// | 55555  |  buf[5] << 2
+				// |4       |  buf[4] << 7
+				dst[3] = buf[6] >> 3 | buf[5] << 2 | buf[4] << 7;
+				n++;
+				nextcase 5;
+			case 5:
+				// |33334444|  dst[2]
+				// |    4444|  buf[4] >> 1
+				// |3333    |  buf[3] << 4
+				dst[2] = buf[4] >> 1 | buf[3] << 4;
+				n++;
+				nextcase 4;
+			case 4:
+				// |11222223|  dst[1]
+				// |       3|  buf[3] >> 4
+				// |  22222 |  buf[2] << 1
+				// |11      |  buf[1] << 6
+				dst[1] = buf[3] >> 4 | buf[2] << 1 | buf[1] << 6;
+				n++;
+				nextcase 2;
+			case 2:
+				// |00000111|  dst[0]
+				// |     111|  buf[1] >> 2
+				// |00000   |  buf[0] << 3
+				dst[0] = buf[1] >> 2 | buf[0] << 3;
+				n++;
+			default:
+				return DecodingFailure.INVALID_CHARACTER?;
+		}
+		if (dst.len < 5) break;
+		dst = dst[5..];
+	}
+	return dst_ptr[:n];
+}
+
+<*
+ Encode the content of src into dst, which must be properly sized.
+ @param [in] src "The input to be encoded."
+ @param [inout] dst "The encoded input."
+ @param padding "The padding character or 0 if none"
+ @param alphabet "The alphabet to use"
+ @require padding < 0xFF "Invalid padding character"
+ @require dst.len >= encode_len(src.len, padding) "Destination buffer too small"
+ @return "The encoded size."
+*>
+fn String encode_buffer(char[] src, char[] dst, char padding = DEFAULT_PAD, Base32Alphabet* alphabet = &STANDARD)
+{
+	if (src.len == 0) return (String)dst[:0];
+
+	char* dst_ptr = dst;
+	usz n = (src.len / 5) * 5;
+	usz dn = encode_len(src.len, padding);
+
+	uint msb, lsb;
+	for (usz i = 0; i < n; i += 5)
+	{
+		// to fit 40 bits we need two 32-bit uints
+		msb = (uint)src[i] << 24 | (uint)src[i+1] << 16
+				| (uint)src[i+2] << 8 | (uint)src[i+3];
+		lsb = msb << 8 | (uint)src[i+4];
+
+		// now slice them into 5-bit chunks and translate to the
+		// alphabet.
+		dst[0] = alphabet.encoding[(msb >> 27) & MASK];
+		dst[1] = alphabet.encoding[(msb >> 22) & MASK];
+		dst[2] = alphabet.encoding[(msb >> 17) & MASK];
+		dst[3] = alphabet.encoding[(msb >> 12) & MASK];
+		dst[4] = alphabet.encoding[(msb >> 7) & MASK];
+		dst[5] = alphabet.encoding[(msb >> 2) & MASK];
+		dst[6] = alphabet.encoding[(lsb >> 5) & MASK];
+		dst[7] = alphabet.encoding[lsb & MASK];
+
+		dst = dst[8..];
+	}
+
+	usz trailing = src.len - n;
+	if (trailing == 0) return (String)dst_ptr[:dn];
+
+	msb = 0;
+	switch (trailing)
+	{
+		case 4:
+			msb |= (uint)src[n+3];
+			lsb = msb << 8;
+			dst[6] = alphabet.encoding[(lsb >> 5) & MASK];
+			dst[5] = alphabet.encoding[(msb >> 2) & MASK];
+			nextcase 3;
+		case 3:
+			msb |= (uint)src[n+2] << 8;
+			dst[4] = alphabet.encoding[(msb >> 7) & MASK];
+			nextcase 2;
+		case 2:
+			msb |= (uint)src[n+1] << 16;
+			dst[3] = alphabet.encoding[(msb >> 12) & MASK];
+			dst[2] = alphabet.encoding[(msb >> 17) & MASK];
+			nextcase 1;
+		case 1:
+			msb |= (uint)src[n] << 24;
+			dst[1] = alphabet.encoding[(msb >> 22) & MASK];
+			dst[0] = alphabet.encoding[(msb >> 27) & MASK];
+	}
+
+	// add the padding
+	if (padding > 0)
+	{
+		for (usz i = (trailing * 8 / 5) + 1; i < 8; i++)
+		{
+			dst[i] = padding;
+		}
+	}
+	return (String)dst_ptr[:dn];
+}
+
+const uint MASK @private = 0b11111;
+const char INVALID @private = 0xff;
+
+const int STD_PADDING = '=';
+const int NO_PADDING = -1;
+
+fault Base32Error
+{
+	DUPLICATE_IN_ALPHABET,
+	PADDING_IN_ALPHABET,
+	INVALID_CHARACTER_IN_ALPHABET,
+	DESTINATION_TOO_SMALL,
+	INVALID_PADDING,
+	CORRUPT_INPUT
+}
+
+struct Base32Encoder @deprecated
+{
+	Base32Alphabet alphabet;
+	char padding;
+}
+
+<*
+ @param encoder "The 32-character alphabet for encoding."
+ @param padding "Set to a negative value to disable padding."
+ @require padding < 256
+*>
+fn void! Base32Encoder.init(&self, Alphabet encoder = STD_ALPHABET, int padding = STD_PADDING)
+{
+	encoder.validate(padding)!;
+	*self = { .alphabet = { .encoding = (char[32])encoder }, .padding = padding < 0 ? (char)0 : (char)padding};
+}
+
+<*
+ Calculate the length in bytes of the encoded data.
+ @param n "Length in bytes on input."
+ @return "Length in bytes of the encoded data."
+*>
+fn usz Base32Encoder.encode_len(&self, usz n)
+{
+	return encode_len(n, self.padding);
+}
+
+<*
+ Encode the content of src into dst, which must be properly sized.
+ @param [in] src "The input to be encoded."
+ @param [inout] dst "The encoded input."
+ @return "The encoded size."
+ @return! Base32Error.DESTINATION_TOO_SMALL
+*>
+fn usz! Base32Encoder.encode(&self, char[] src, char[] dst)
+{
+	usz dn = self.encode_len(src.len);
+	if (dst.len < dn) return Base32Error.DESTINATION_TOO_SMALL?;
+	return encode_buffer(src, dst, self.padding, &self.alphabet).len;
+}
+
+struct Base32Decoder @deprecated
+{
+	Base32Alphabet alphabet;
+	char padding;
+}
+
+<*
+ @param decoder "The alphabet used for decoding."
+ @param padding "Set to a negative value to disable padding."
+ @require padding < 256
+*>
+fn void! Base32Decoder.init(&self, Alphabet decoder = STD_ALPHABET, int padding = STD_PADDING)
+{
+	decoder.validate(padding)!;
+	*self = { .alphabet = { .encoding = (char[32])decoder }, .padding = padding < 0 ? (char)0 : (char)padding };
+
+	self.alphabet.reverse[..] = INVALID;
+	foreach (char i, c : decoder)
+	{
+		self.alphabet.reverse[c] = i;
+	}
+}
+
+<*
+ Calculate the length in bytes of the decoded data.
+ @param n "Length in bytes of input."
+ @return "Length in bytes of the decoded data."
+*>
+fn usz Base32Decoder.decode_len(&self, usz n)
+{
+	return decode_len(n, self.padding);
+}
+
+<*
+ Decode the content of src into dst, which must be properly sized.
+ @param src "The input to be decoded."
+ @param dst "The decoded input."
+ @return "The decoded size."
+ @return! Base32Error.DESTINATION_TOO_SMALL, Base32Error.CORRUPT_INPUT
+*>
+fn usz! Base32Decoder.decode(&self, char[] src, char[] dst)
+{
+	if (src.len == 0) return 0;
+	usz dn = self.decode_len(src.len);
+	if (dst.len < dn) return Base32Error.DESTINATION_TOO_SMALL?;
+	return decode_buffer(src, dst, self.padding, &self.alphabet).len;
+}
+
+
+// Validate the 32-character alphabet to make sure that no character occurs
+// twice and that the padding is not present in the alphabet.
+fn void! Alphabet.validate(&self, int padding)
+{
+	bool[256] checked;
+	foreach (c : self)
+	{
+		if (checked[c])
+		{
+			return Base32Error.DUPLICATE_IN_ALPHABET?;
+		}
+		checked[c] = true;
+		if (c == '\r' || c == '\n')
+		{
+			return Base32Error.INVALID_CHARACTER_IN_ALPHABET?;
+		}
+	}
+	if (padding >= 0)
+	{
+		char pad = (char)padding;
+		if (pad == '\r' || pad == '\n')
+		{
+			return Base32Error.INVALID_PADDING?;
+		}
+		if (checked[pad])
+		{
+			return Base32Error.PADDING_IN_ALPHABET?;
+		}
+	}
+}
+
+distinct Alphabet = char[32];
+// Standard base32 Alphabet
+const Alphabet STD_ALPHABET = "ABCDEFGHIJKLMNOPQRSTUVWXYZ234567";
+// Extended Hex Alphabet
+const Alphabet HEX_ALPHABET = "0123456789ABCDEFGHIJKLMNOPQRSTUV";
+
+const Base32Alphabet STANDARD = {
+	.encoding = "ABCDEFGHIJKLMNOPQRSTUVWXYZ234567",
+	.reverse = x`ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff
+                 ffffffffffffffffffffffffffffffffffff1a1b1c1d1e1fffffffffffffffff
+                 ff000102030405060708090a0b0c0d0e0f10111213141516171819ffffffffff
+                 ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff
+                 ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff
+                 ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff
+                 ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff
+                 ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff`
+};
+
+const Base32Alphabet HEX = {
+	.encoding = "0123456789ABCDEFGHIJKLMNOPQRSTUV",
+	.reverse = x`ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff
+                 ffffffffffffffffffffffffffffffff00010203040506070809ffffffffffff
+                 ff0a0b0c0d0e0f101112131415161718191a1b1c1d1e1fffffffffffffffffff
+                 ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff
+                 ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff
+                 ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff
+                 ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff
+                 ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff`
+};

lib/std/encoding/base64.c3

@@ -0,0 +1,401 @@
+module std::encoding::base64;
+import std::core::bitorder;
+
+// The implementation is based on https://www.rfc-editor.org/rfc/rfc4648
+// Specifically this section:
+// https://www.rfc-editor.org/rfc/rfc4648#section-4
+
+const char NO_PAD = 0;
+const char DEFAULT_PAD = '=';
+
+struct Base64Alphabet
+{
+	char[64] encoding;
+	char[256] reverse;
+}
+
+const Base64Alphabet STANDARD = {
+	.encoding = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/",
+	.reverse =
+	x`ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff
+	  ffffffffffffffffffffff3effffff3f3435363738393a3b3c3dffffffffffff
+	  ff000102030405060708090a0b0c0d0e0f10111213141516171819ffffffffff
+	  ff1a1b1c1d1e1f202122232425262728292a2b2c2d2e2f30313233ffffffffff
+	  ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff
+	  ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff
+	  ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff
+	  ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff`
+};
+
+const Base64Alphabet URL = {
+	.encoding = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_",
+	.reverse =
+	x`ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff
+	  ffffffffffffffffffffffffff3effff3435363738393a3b3c3dffffffffffff
+	  ff000102030405060708090a0b0c0d0e0f10111213141516171819ffffffff3f
+	  ff1a1b1c1d1e1f202122232425262728292a2b2c2d2e2f30313233ffffffffff
+	  ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff
+	  ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff
+	  ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff
+	  ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff`
+};
+
+const STD_ALPHABET = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
+const URL_ALPHABET = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_";
+
+fn String encode(char[] src, Allocator allocator, char padding = DEFAULT_PAD, Base64Alphabet* alphabet = &STANDARD)
+{
+	char[] dst = allocator::alloc_array(allocator, char, encode_len(src.len, padding));
+	return encode_buffer(src, dst, padding, alphabet);
+}
+
+fn char[]! decode(char[] src, Allocator allocator, char padding = DEFAULT_PAD, Base64Alphabet* alphabet = &STANDARD)
+{
+	char[] dst = allocator::alloc_array(allocator, char, decode_len(src.len, padding))!;
+	return decode_buffer(src, dst, padding, alphabet);
+}
+
+fn String encode_new(char[] code, char padding = DEFAULT_PAD, Base64Alphabet* alphabet = &STANDARD) @inline => encode(code, allocator::heap(), padding, alphabet);
+fn String encode_temp(char[] code, char padding = DEFAULT_PAD, Base64Alphabet* alphabet = &STANDARD) @inline => encode(code, allocator::temp(), padding, alphabet);
+fn char[]! decode_new(char[] code, char padding = DEFAULT_PAD, Base64Alphabet* alphabet = &STANDARD) @inline => decode(code, allocator::heap(), padding, alphabet);
+fn char[]! decode_temp(char[] code, char padding = DEFAULT_PAD, Base64Alphabet* alphabet = &STANDARD) @inline => decode(code, allocator::temp(), padding, alphabet);
+
+
+<*
+ Calculate the size of the encoded data.
+ @param n "Size of the input to be encoded."
+ @param padding "The padding character or 0 if none"
+ @require padding < 0xFF "Invalid padding character"
+ @return "The size of the input once encoded."
+*>
+fn usz encode_len(usz n, char padding)
+{
+	if (padding) return (n + 2) / 3 * 4;
+	usz trailing = n % 3;
+	return n / 3 * 4 + (trailing * 4 + 2) / 3;
+}
+
+<*
+ Calculate the size of the decoded data.
+ @param n "Size of the input to be decoded."
+ @param padding "The padding character or 0 if none"
+ @require padding < 0xFF "Invalid padding character"
+ @return "The size of the input once decoded."
+ @return! DecodingFailure.INVALID_PADDING
+*>
+fn usz! decode_len(usz n, char padding)
+{
+	usz dn = n / 4 * 3;
+	usz trailing = n % 4;
+	if (padding)
+	{
+		if (trailing != 0) return DecodingFailure.INVALID_PADDING?;
+		// source size is multiple of 4
+		return dn;
+	}
+	if (trailing == 1) return DecodingFailure.INVALID_PADDING?;
+	return dn + trailing * 3 / 4;
+}
+
+<*
+ Encode the content of src into dst, which must be properly sized.
+ @param src "The input to be encoded."
+ @param dst "The encoded input."
+ @param padding "The padding character or 0 if none"
+ @param alphabet "The alphabet to use"
+ @require padding < 0xFF "Invalid padding character"
+ @return "The encoded size."
+ @return! Base64Error.DESTINATION_TOO_SMALL
+*>
+fn String encode_buffer(char[] src, char[] dst, char padding = DEFAULT_PAD, Base64Alphabet* alphabet = &STANDARD)
+{
+	if (src.len == 0) return (String)dst[:0];
+	usz dn = encode_len(src.len, padding);
+	char* dst_ptr = dst;
+	assert(dst.len >= dn);
+	usz trailing = src.len % 3;
+	char[] src3 = src[:^trailing];
+
+	while (src3.len > 0)
+	{
+		uint group = (uint)src3[0] << 16 | (uint)src3[1] << 8 | (uint)src3[2];
+		dst[0] = alphabet.encoding[group >> 18 & MASK];
+		dst[1] = alphabet.encoding[group >> 12 & MASK];
+		dst[2] = alphabet.encoding[group >> 6 & MASK];
+		dst[3] = alphabet.encoding[group & MASK];
+		dst = dst[4..];
+		src3 = src3[3..];
+	}
+
+	// Encode the remaining bytes according to:
+	// https://www.rfc-editor.org/rfc/rfc4648#section-3.5
+	switch (trailing)
+	{
+		case 1:
+			uint group = (uint)src[^1] << 16;
+			dst[0] = alphabet.encoding[group >> 18 & MASK];
+			dst[1] = alphabet.encoding[group >> 12 & MASK];
+			if (padding > 0)
+			{
+				dst[2] = padding;
+				dst[3] = padding;
+			}
+		case 2:
+			uint group = (uint)src[^2] << 16 | (uint)src[^1] << 8;
+			dst[0] = alphabet.encoding[group >> 18 & MASK];
+			dst[1] = alphabet.encoding[group >> 12 & MASK];
+			dst[2] = alphabet.encoding[group >> 6 & MASK];
+			if (padding > 0)
+			{
+				dst[3] = padding;
+			}
+		case 0:
+			break;
+		default:
+			unreachable();
+	}
+	return (String)dst_ptr[:dn];
+}
+
+<*
+ Decode the content of src into dst, which must be properly sized.
+ @param src "The input to be decoded."
+ @param dst "The decoded input."
+ @param padding "The padding character or 0 if none"
+ @param alphabet "The alphabet to use"
+ @require (decode_len(src.len, padding) ?? 0) <= dst.len "Destination buffer too small"
+ @require padding < 0xFF "Invalid padding character"
+ @return "The decoded data."
+ @return! DecodingFailure
+*>
+fn char[]! decode_buffer(char[] src, char[] dst, char padding = DEFAULT_PAD, Base64Alphabet* alphabet = &STANDARD)
+{
+	if (src.len == 0) return dst[:0];
+	usz dn = decode_len(src.len, padding)!;
+	assert(dst.len >= dn);
+
+	usz trailing = src.len % 4;
+	char* dst_ptr = dst;
+	char[] src4 = src;
+	switch
+	{
+		case !padding:
+			src4 = src[:^trailing];
+		default:
+			// If there is padding, keep the last 4 bytes for later.
+			// NB. src.len >= 4 as decode_len passed
+			trailing = 4;
+			if (src[^1] == padding) src4 = src[:^4];
+	}
+	while (src4.len > 0)
+	{
+		char c0 = alphabet.reverse[src4[0]];
+		char c1 = alphabet.reverse[src4[1]];
+		char c2 = alphabet.reverse[src4[2]];
+		char c3 = alphabet.reverse[src4[3]];
+		switch (0xFF)
+		{
+			case c0:
+			case c1:
+			case c2:
+			case c3:
+				return DecodingFailure.INVALID_CHARACTER?;
+		}
+		uint group = (uint)c0 << 18 | (uint)c1 << 12 | (uint)c2 << 6 | (uint)c3;
+		dst[0] = (char)(group >> 16);
+		dst[1] = (char)(group >> 8);
+		dst[2] = (char)group;
+		dst = dst[3..];
+		src4 = src4[4..];
+	}
+
+	if (trailing == 0) return dst_ptr[:dn];
+
+	src = src[^trailing..];
+	char c0 = alphabet.reverse[src[0]];
+	char c1 = alphabet.reverse[src[1]];
+	if (c0 == 0xFF || c1 == 0xFF) return DecodingFailure.INVALID_PADDING?;
+	if (!padding)
+	{
+		switch (src.len)
+		{
+			case 2:
+				uint group = (uint)c0 << 18 | (uint)c1 << 12;
+				dst[0] = (char)(group >> 16);
+			case 3:
+				char c2 = alphabet.reverse[src[2]];
+				if (c2 == 0xFF) return DecodingFailure.INVALID_CHARACTER?;
+				uint group = (uint)c0 << 18 | (uint)c1 << 12 | (uint)c2 << 6;
+				dst[0] = (char)(group >> 16);
+				dst[1] = (char)(group >> 8);
+		}
+	}
+	else
+	{
+		// Valid paddings are:
+		// 2: xx==
+		// 1: xxx=
+		switch (padding)
+		{
+			case src[2]:
+				if (src[3] != padding) return DecodingFailure.INVALID_PADDING?;
+				uint group = (uint)c0 << 18 | (uint)c1 << 12;
+				dst[0] = (char)(group >> 16);
+				dn -= 2;
+			case src[3]:
+				char c2 = alphabet.reverse[src[2]];
+				if (c2 == 0xFF) return DecodingFailure.INVALID_CHARACTER?;
+				uint group = (uint)c0 << 18 | (uint)c1 << 12 | (uint)c2 << 6;
+				dst[0] = (char)(group >> 16);
+				dst[1] = (char)(group >> 8);
+				dn -= 1;
+		}
+	}
+	return dst_ptr[:dn];
+}
+
+const MASK @private = 0b111111;
+
+struct Base64Encoder @deprecated
+{
+	char padding;
+	String alphabet;
+}
+
+fault Base64Error
+{
+	DUPLICATE_IN_ALPHABET,
+	PADDING_IN_ALPHABET,
+	DESTINATION_TOO_SMALL,
+	INVALID_PADDING,
+	INVALID_CHARACTER,
+}
+
+<*
+ @param alphabet "The alphabet used for encoding."
+ @param padding "Set to a negative value to disable padding."
+ @require alphabet.len == 64
+ @require padding < 256
+ @return! Base64Error.DUPLICATE_IN_ALPHABET, Base64Error.PADDING_IN_ALPHABET
+*>
+fn Base64Encoder*! Base64Encoder.init(&self, String alphabet, int padding = '=')
+{
+	check_alphabet(alphabet, padding)!;
+	*self = { .padding = padding < 0 ? 0 : (char)padding, .alphabet = alphabet };
+	return self;
+}
+
+<*
+ Calculate the size of the encoded data.
+ @param n "Size of the input to be encoded."
+ @return "The size of the input once encoded."
+*>
+fn usz Base64Encoder.encode_len(&self, usz n)
+{
+	return encode_len(n, self.padding);
+}
+
+<*
+ Encode the content of src into dst, which must be properly sized.
+ @param src "The input to be encoded."
+ @param dst "The encoded input."
+ @return "The encoded size."
+ @return! Base64Error.DESTINATION_TOO_SMALL
+*>
+fn usz! Base64Encoder.encode(&self, char[] src, char[] dst)
+{
+	if (src.len == 0) return 0;
+	usz dn = self.encode_len(src.len);
+	if (dst.len < dn) return Base64Error.DESTINATION_TOO_SMALL?;
+	Base64Alphabet a = { .encoding = self.alphabet[:64] };
+	return encode_buffer(src, dst, self.padding, &a).len;
+}
+
+struct Base64Decoder @deprecated
+{
+	char padding;
+	Base64Alphabet encoding;
+	bool init_done;
+}
+
+import std;
+<*
+ @param alphabet "The alphabet used for encoding."
+ @param padding "Set to a negative value to disable padding."
+ @require alphabet.len == 64
+ @require padding < 256
+ @return! Base64Error.DUPLICATE_IN_ALPHABET, Base64Error.PADDING_IN_ALPHABET
+*>
+fn void! Base64Decoder.init(&self, String alphabet, int padding = '=')
+{
+	self.init_done = true;
+	check_alphabet(alphabet, padding)!;
+	*self = { .padding = padding < 0 ? 0 : (char)padding, .encoding.encoding = alphabet[:64] };
+
+	self.encoding.reverse[..] = 0xFF;
+
+	foreach (i, c : alphabet)
+	{
+		self.encoding.reverse[c] = (char)i;
+	}
+}
+
+<*
+ Calculate the size of the decoded data.
+ @param n "Size of the input to be decoded."
+ @return "The size of the input once decoded."
+ @return! Base64Error.INVALID_PADDING
+*>
+fn usz! Base64Decoder.decode_len(&self, usz n)
+{
+	return decode_len(n, self.padding) ?? Base64Error.INVALID_PADDING?;
+}
+
+<*
+ Decode the content of src into dst, which must be properly sized.
+ @param src "The input to be decoded."
+ @param dst "The decoded input."
+ @return "The decoded size."
+ @return! Base64Error.DESTINATION_TOO_SMALL, Base64Error.INVALID_PADDING, Base64Error.INVALID_CHARACTER
+*>
+fn usz! Base64Decoder.decode(&self, char[] src, char[] dst)
+{
+	if (src.len == 0) return 0;
+	usz dn = self.decode_len(src.len)!;
+	if (dst.len < dn) return Base64Error.DESTINATION_TOO_SMALL?;
+	char[]! decoded = decode_buffer(src, dst, self.padding, &self.encoding);
+	if (catch err = decoded)
+	{
+		case DecodingFailure.INVALID_PADDING:
+			return Base64Error.INVALID_PADDING?;
+		case DecodingFailure.INVALID_CHARACTER:
+			return Base64Error.INVALID_CHARACTER?;
+		default:
+			return err?;
+	}
+	return decoded.len;
+}
+
+// Make sure that all bytes in the alphabet are unique and
+// the padding is not present in the alphabet.
+fn void! check_alphabet(String alphabet, int padding) @local
+{
+	bool[256] checked;
+	if (padding < 0)
+	{
+		foreach (c : alphabet)
+		{
+			if (checked[c]) return Base64Error.DUPLICATE_IN_ALPHABET?;
+			checked[c] = true;
+		}
+		return;
+	}
+	char pad = (char)padding;
+	foreach (c : alphabet)
+	{
+		if (c == pad) return Base64Error.PADDING_IN_ALPHABET?;
+		if (checked[c]) return Base64Error.DUPLICATE_IN_ALPHABET?;
+		checked[c] = true;
+	}
+}
+

lib/std/encoding/csv.c3

@@ -0,0 +1,95 @@
+module std::encoding::csv;
+import std::io;
+
+
+struct CsvReader
+{
+	InStream stream;
+	String separator;
+}
+
+struct CsvRow (Printable)
+{
+	String[] list;
+	String row;
+	Allocator allocator;
+}
+
+fn usz! CsvRow.to_format(&self, Formatter* f) @dynamic
+{
+	return f.printf("%s", self.list);
+}
+
+fn usz CsvRow.len(&self) @operator(len)
+{
+	return self.list.len;
+}
+
+<*
+ @require col < self.list.len
+*>
+fn String CsvRow.get_col(&self, usz col) @operator([])
+{
+	return self.list[col];
+}
+
+fn void CsvReader.init(&self, InStream stream, String separator = ",")
+{
+	self.stream = stream;
+	self.separator = separator;
+}
+
+fn CsvRow! CsvReader.read_new_row(self)
+{
+	return self.read_row(allocator::heap()) @inline;
+}
+
+<*
+ @param [&inout] allocator
+*>
+fn CsvRow! CsvReader.read_row(self, Allocator allocator)
+{
+	String row = io::readline(self.stream, allocator: allocator)!;
+	defer catch allocator::free(allocator, row);
+	String[] list = row.split(self.separator, allocator: allocator);
+	return { list, row, allocator };
+}
+
+fn CsvRow! CsvReader.read_temp_row(self)
+{
+	return self.read_row(allocator::temp()) @inline;
+}
+
+<*
+ @require self.allocator != null  `Row already freed`
+*>
+fn void CsvRow.free(&self)
+{
+	allocator::free(self.allocator, self.list);
+	allocator::free(self.allocator, self.row);
+	self.allocator = null;
+}
+
+fn void! CsvReader.skip_row(self) @maydiscard => @pool()
+{
+	(void)io::treadline(self.stream);
+}
+
+macro void! CsvReader.@each_row(self, int rows = int.max; @body(String[] row)) @maydiscard
+{
+	InStream stream = self.stream;
+	String sep = self.separator;
+	while (rows--)
+	{
+		@stack_mem(512; Allocator mem)
+		{
+			String! s = io::readline(stream, mem);
+			if (catch err = s)
+			{
+				if (err == IoError.EOF) return;
+				return err?;
+			}
+			@body(s.split(sep, allocator: mem));
+		};
+	}
+}

lib/std/encoding/encoding.c3

@@ -0,0 +1,7 @@
+module std::encoding;
+
+fault DecodingFailure
+{
+	INVALID_CHARACTER,
+	INVALID_PADDING,
+}

lib/std/encoding/hex.c3

@@ -0,0 +1,109 @@
+module std::encoding::hex;
+import std::encoding @norecurse;
+
+// The implementation is based on https://www.rfc-editor.org/rfc/rfc4648
+
+fn String encode_buffer(char[] code, char[] buffer)
+{
+	return (String)buffer[:encode_bytes(code, buffer)];
+}
+
+fn char[]! decode_buffer(char[] code, char[] buffer)
+{
+	return buffer[:decode_bytes(code, buffer)!];
+}
+
+fn String encode(char[] code, Allocator allocator)
+{
+	char[] data = allocator::alloc_array(allocator, char, encode_len(code.len));
+	return (String)data[:encode_bytes(code, data)];
+}
+
+fn char[]! decode(char[] code, Allocator allocator)
+{
+	char[] data = allocator::alloc_array(allocator, char, decode_len(code.len));
+	return data[:decode_bytes(code, data)!];
+}
+
+fn String encode_new(char[] code) @inline => encode(code, allocator::heap());
+fn String encode_temp(char[] code) @inline => encode(code, allocator::temp());
+fn char[]! decode_new(char[] code) @inline => decode(code, allocator::heap());
+fn char[]! decode_temp(char[] code) @inline => decode(code, allocator::temp());
+
+<*
+ Calculate the size of the encoded data.
+ @param n "Size of the input to be encoded."
+ @return "The size of the input once encoded."
+*>
+fn usz encode_len(usz n) => n * 2;
+
+<*
+ Encode the content of src into dst, which must be properly sized.
+ @param src "The input to be encoded."
+ @param dst "The encoded input."
+ @return "The encoded size."
+ @require dst.len >= encode_len(src.len) "Destination array is not large enough"
+*>
+fn usz encode_bytes(char[] src, char[] dst)
+{
+	usz j = 0;
+	foreach (v : src)
+	{
+		dst[j] = HEXALPHABET[v >> 4];
+		dst[j + 1] = HEXALPHABET[v & 0x0f];
+		j = j + 2;
+	}
+	return src.len * 2;
+}
+
+<*
+ Calculate the size of the decoded data.
+ @param n "Size of the input to be decoded."
+ @return "The size of the input once decoded."
+*>
+macro usz decode_len(usz n) => n / 2;
+
+<*
+ Decodes src into bytes. Returns the actual number of bytes written to dst.
+
+ Expects that src only contains hexadecimal characters and that src has even
+ length.
+
+ @param src "The input to be decoded."
+ @param dst "The decoded input."
+ @require src.len % 2 == 0 "src is not of even length"
+ @require dst.len >= decode_len(src.len) "Destination array is not large enough"
+ @return! DecodingFailure.INVALID_CHARACTER
+*>
+fn usz! decode_bytes(char[] src, char[] dst)
+{
+	usz i;
+	for (usz j = 1; j < src.len; j += 2)
+	{
+		char a = HEXREVERSE[src[j - 1]];
+		char b = HEXREVERSE[src[j]];
+		if (a > 0x0f || b > 0x0f) return DecodingFailure.INVALID_CHARACTER?;
+		dst[i] = (a << 4) | b;
+		i++;
+	}
+	return i;
+}
+
+const char[?] HEXALPHABET @private = "0123456789abcdef";
+const char[?] HEXREVERSE @private =
+x`ffffffffffffffffffffffffffffffff
+  ffffffffffffffffffffffffffffffff
+  ffffffffffffffffffffffffffffffff
+  00010203040506070809ffffffffffff
+  ff0a0b0c0d0e0fffffffffffffffffff
+  ffffffffffffffffffffffffffffffff
+  ff0a0b0c0d0e0fffffffffffffffffff
+  ffffffffffffffffffffffffffffffff
+  ffffffffffffffffffffffffffffffff
+  ffffffffffffffffffffffffffffffff
+  ffffffffffffffffffffffffffffffff
+  ffffffffffffffffffffffffffffffff
+  ffffffffffffffffffffffffffffffff
+  ffffffffffffffffffffffffffffffff
+  ffffffffffffffffffffffffffffffff
+  ffffffffffffffffffffffffffffffff`;

lib/std/encoding/json.c3

@@ -0,0 +1,392 @@
+// Copyright (c) 2023 Christoffer Lerno. All rights reserved.
+// Use of this source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+module std::encoding::json;
+import std::io;
+import std::ascii;
+import std::collections::object;
+
+fault JsonParsingError
+{
+	EOF,
+	UNEXPECTED_CHARACTER,
+	INVALID_ESCAPE_SEQUENCE,
+	DUPLICATE_MEMBERS,
+	INVALID_NUMBER,
+}
+
+fn Object*! parse_string(String s, Allocator allocator = allocator::heap())
+{
+	return parse((ByteReader){}.init(s), allocator);
+}
+
+fn Object*! temp_parse_string(String s)
+{
+	return parse((ByteReader){}.init(s), allocator::temp());
+}
+
+fn Object*! parse(InStream s, Allocator allocator = allocator::heap())
+{
+	@stack_mem(512; Allocator mem)
+	{
+		JsonContext context = { .last_string = dstring::new_with_capacity(64, mem), .stream = s, .allocator = allocator };
+		@pool(allocator)
+		{
+			return parse_any(&context);
+		};
+	};
+}
+
+fn Object*! temp_parse(InStream s)
+{
+	return parse(s, allocator::temp());
+}
+
+// -- Implementation follows --
+
+enum JsonTokenType @local
+{
+	NO_TOKEN,
+	LBRACE,
+	LBRACKET,
+	COMMA,
+	COLON,
+	RBRACE,
+	RBRACKET,
+	STRING,
+	NUMBER,
+	TRUE,
+	FALSE,
+	NULL,
+	EOF,
+}
+
+struct JsonContext @local
+{
+	uint line;
+	InStream stream;
+	Allocator allocator;
+	JsonTokenType token;
+	DString last_string;
+	double last_number;
+	char current;
+	bitstruct : char {
+		bool skip_comments;
+		bool reached_end;
+		bool pushed_back;
+	}
+}
+
+
+fn Object*! parse_from_token(JsonContext* context, JsonTokenType token) @local
+{
+	switch (token)
+	{
+		case NO_TOKEN: unreachable();
+		case LBRACE: return parse_map(context);
+		case LBRACKET: return parse_array(context);
+		case COMMA:
+		case RBRACE:
+		case RBRACKET:
+		case COLON: return JsonParsingError.UNEXPECTED_CHARACTER?;
+		case STRING: return object::new_string(context.last_string.str_view(), context.allocator);
+		case NUMBER: return object::new_float(context.last_number, context.allocator);
+		case TRUE: return object::new_bool(true);
+		case FALSE: return object::new_bool(false);
+		case NULL: return object::new_null();
+		case EOF: return JsonParsingError.EOF?;
+	}
+}
+fn Object*! parse_any(JsonContext* context) @local
+{
+	return parse_from_token(context, advance(context));
+}
+
+fn JsonTokenType! lex_number(JsonContext *context, char c) @local
+{
+	@stack_mem(256; Allocator mem)
+	{
+		DString t = dstring::new_with_capacity(32, allocator: mem);
+		bool negate = c == '-';
+		if (negate)
+		{
+			t.append(c);
+			c = read_next(context)!;
+		}
+		while (c.is_digit())
+		{
+			t.append(c);
+			c = read_next(context)!;
+		}
+		if (c == '.')
+		{
+			t.append(c);
+			while (c = read_next(context)!, c.is_digit())
+			{
+				t.append(c);
+			}
+		}
+		if ((c | 32) == 'e')
+		{
+			t.append(c);
+			c = read_next(context)!;
+			switch (c)
+			{
+				case '-':
+				case '+':
+					t.append(c);
+					c = read_next(context)!;
+			}
+			if (!c.is_digit()) return JsonParsingError.INVALID_NUMBER?;
+			while (c.is_digit())
+			{
+				t.append(c);
+				c = read_next(context)!;
+			}
+		}
+		pushback(context, c);
+		double! d = t.str_view().to_double() ?? JsonParsingError.INVALID_NUMBER?;
+		context.last_number = d!;
+		return NUMBER;
+	};
+}
+
+fn Object*! parse_map(JsonContext* context) @local
+{
+	Object* map = object::new_obj(context.allocator);
+	defer catch map.free();
+	JsonTokenType token = advance(context)!;
+
+	@stack_mem(256; Allocator mem)
+	{
+		DString temp_key = dstring::new_with_capacity(32, mem);
+		while (token != JsonTokenType.RBRACE)
+		{
+			if (token != JsonTokenType.STRING) return JsonParsingError.UNEXPECTED_CHARACTER?;
+			DString string = context.last_string;
+			if (map.has_key(string.str_view())) return JsonParsingError.DUPLICATE_MEMBERS?;
+			// Copy the key to our temp holder, since our
+			// last_string may be used in parse_any
+			temp_key.clear();
+			temp_key.append(string);
+			parse_expected(context, COLON)!;
+			Object* element = parse_any(context)!;
+			map.set(temp_key.str_view(), element);
+			token = advance(context)!;
+			if (token == JsonTokenType.COMMA)
+			{
+				token = advance(context)!;
+				continue;
+			}
+			if (token != JsonTokenType.RBRACE) return JsonParsingError.UNEXPECTED_CHARACTER?;
+		}
+		return map;
+	};
+}
+
+fn Object*! parse_array(JsonContext* context) @local
+{
+	Object* list = object::new_obj(context.allocator);
+	defer catch list.free();
+	JsonTokenType token = advance(context)!;
+	while (token != JsonTokenType.RBRACKET)
+	{
+		Object* element = parse_from_token(context, token)!;
+		list.push(element);
+		token = advance(context)!;
+		if (token == JsonTokenType.COMMA)
+		{
+			token = advance(context)!;
+			continue;
+		}
+		if (token != JsonTokenType.RBRACKET) return JsonParsingError.UNEXPECTED_CHARACTER?;
+	}
+	return list;
+}
+
+fn void pushback(JsonContext* context, char c) @local
+{
+	if (!context.reached_end)
+	{
+		assert(!context.pushed_back);
+		context.pushed_back = true;
+		context.current = c;
+	}
+}
+
+fn char! read_next(JsonContext* context) @local
+{
+	if (context.reached_end) return '\0';
+	if (context.pushed_back)
+	{
+		context.pushed_back = false;
+		return context.current;
+	}
+	char! c = context.stream.read_byte();
+	if (catch err = c)
+	{
+		case IoError.EOF:
+			context.reached_end = true;
+			return '\0';
+		default:
+			return err?;
+	}
+	if (c == 0)
+	{
+		context.reached_end = true;
+	}
+	return c;
+}
+
+fn JsonTokenType! advance(JsonContext* context) @local
+{
+	char c;
+	// Skip whitespace
+	while WS: (c = read_next(context)!)
+	{
+		switch (c)
+		{
+			case '\n':
+				context.line++;
+				nextcase;
+			case ' ':
+			case '\t':
+			case '\r':
+			case '\v':
+				continue;
+			case '/':
+				if (!context.skip_comments) break;
+				c = read_next(context)!;
+				if (c != '*')
+				{
+					pushback(context, c);
+					break WS;
+				}
+				while COMMENT: (true)
+				{
+					// Skip to */
+					while (c = read_next(context)!)
+					{
+						if (c == '\n') context.line++;
+						if (c != '*') continue;
+						// Skip through all the '*'
+						while (c = read_next(context)!)
+						{
+							if (c == '\n') context.line++;
+							if (c != '*') break;
+						}
+						if (c == '/') break COMMENT;
+					}
+				}
+				continue;
+			default:
+				break WS;
+		}
+	}
+	switch (c)
+	{
+		case '\0':
+			return IoError.EOF?;
+		case '{':
+			return LBRACE;
+		case '}':
+			return RBRACE;
+		case '[':
+			return LBRACKET;
+		case ']':
+			return RBRACKET;
+		case ':':
+			return COLON;
+		case ',':
+			return COMMA;
+		case '"':
+			return lex_string(context);
+		case '-':
+		case '0'..'9':
+			return lex_number(context, c);
+		case 't':
+			match(context, "rue")!;
+			return TRUE;
+		case 'f':
+			match(context, "alse")!;
+			return FALSE;
+		case 'n':
+			match(context, "ull")!;
+			return NULL;
+		default:
+			return JsonParsingError.UNEXPECTED_CHARACTER?;
+	}
+}
+
+fn void! match(JsonContext* context, String str) @local
+{
+	foreach (c : str)
+	{
+		char l = read_next(context)!;
+		if (l != c) return JsonParsingError.UNEXPECTED_CHARACTER?;
+	}
+}
+
+fn void! parse_expected(JsonContext* context, JsonTokenType token) @local
+{
+	if (advance(context)! != token) return JsonParsingError.UNEXPECTED_CHARACTER?;
+}
+
+fn JsonTokenType! lex_string(JsonContext* context)
+{
+	context.last_string.clear();
+	while LOOP: (true)
+	{
+		char c = read_next(context)!;
+		switch (c)
+		{
+			case '\0':
+				return JsonParsingError.EOF?;
+			case 1..31:
+				return JsonParsingError.UNEXPECTED_CHARACTER?;
+			case '"':
+				break LOOP;
+			case '\\':
+				break;
+			default:
+				context.last_string.append(c);
+				continue;
+		}
+		c = read_next(context)!;
+		switch (c)
+		{
+			case '\0':
+				return JsonParsingError.EOF?;
+			case 1..31:
+				return JsonParsingError.UNEXPECTED_CHARACTER?;
+			case '"':
+			case '\\':
+			case '/':
+				break;
+			case 'b':
+				c = '\b';
+			case 'f':
+				c = '\f';
+			case 'n':
+				c = '\n';
+			case 'r':
+				c = '\r';
+			case 't':
+				c = '\t';
+			case 'u':
+				uint val;
+				for (int i = 0; i < 4; i++)
+				{
+					c = read_next(context)!;
+					if (!c.is_xdigit()) return JsonParsingError.INVALID_ESCAPE_SEQUENCE?;
+					val = val << 4 + (c > '9' ? (c | 32) - 'a' + 10 : c - '0');
+				}
+				context.last_string.append_char32(val);
+				continue;
+			default:
+				 return JsonParsingError.INVALID_ESCAPE_SEQUENCE?;
+		}
+		context.last_string.append(c);
+	}
+	return STRING;
+}

lib/std/enumset.c3

@@ -1,80 +0,0 @@
-// TODO: ensure the type is an enum first.
-module enumset<Enum>;
-
-$assert(Enum.min < Enum.max, "Only strictly increasing enums may be used with enum sets.");
-$assert(Enum.max < 64, "Maximum value of an enum used as enum set is 63");
-$assert(Enum.min >= 0, "Minimum value of an enum used as enum set is 0");
-
-$switch ($$C_INT_SIZE):
-$case 64:
-    private define EnumSetType = ulong;
-$case 32:
-    $if (Enum.max < 32):
-        private define EnumSetType = uint;
-    $else:
-        private define EnumSetType = ulong;
-    $endif;
-$default:
-    $if (Enum.max < 16):
-        private define EnumSetType = ushort;
-    $elif (Enum.max < 31):
-        private define EnumSetType = uint;
-    $else:
-        private define EnumSetType = ulong;
-    $endif;
-$endswitch;
-
-define EnumSet = distinct EnumSetType;
-
-fn void EnumSet.add(EnumSet *this, Enum v)
-{
-    *this = (EnumSet)((EnumSetType)*this | 1u << (EnumSetType)v);
-}
-
-fn void EnumSet.clear(EnumSet *this)
-{
-    *this = 0;
-}
-
-fn bool EnumSet.remove(EnumSet *this, Enum v)
-{
-    EnumSetType old = (EnumSetType)*this;
-    EnumSetType new = old & ~(1u << (EnumSetType)v);
-    *this = (EnumSet)new;
-    return old != new;
-}
-
-fn bool EnumSet.has(EnumSet *this, Enum v)
-{
-    return ((EnumSetType)*this & (1u << (EnumSetType)v)) != 0;
-}
-
-fn void EnumSet.add_all(EnumSet *this, EnumSet s)
-{
-    *this = (EnumSet)((EnumSetType)*this | (EnumSetType)s);
-}
-
-fn void EnumSet.retain_all(EnumSet *this, EnumSet s)
-{
-    *this = (EnumSet)((EnumSetType)*this & (EnumSetType)s);
-}
-
-fn EnumSet EnumSet.and_of(EnumSet *this, EnumSet s)
-{
-    return (EnumSet)((EnumSetType)*this & (EnumSetType)s);
-}
-
-fn EnumSet EnumSet.or_of(EnumSet *this, EnumSet s)
-{
-    return (EnumSet)((EnumSetType)*this | (EnumSetType)s);
-}
-
-fn EnumSet EnumSet.diff_of(EnumSet *this, EnumSet s)
-{
-    return (EnumSet)((EnumSetType)*this & ~(EnumSetType)s);
-}
-
-fn EnumSet EnumSet.xor_of(EnumSet *this, EnumSet s)
-{
-    return (EnumSet)((EnumSetType)*this ^ (EnumSetType)s);
-}

lib/std/env.c3

@@ -1,58 +0,0 @@
-// Copyright (c) 2021 Christoffer Lerno. All rights reserved.
-// Use of this source code is governed by the MIT license
-// a copy of which can be found in the LICENSE_STDLIB file.
-module std::env;
-
-enum CompilerOptLevel
-{
-    O0,
-    O1,
-    O2,
-    O3
-}
-
-enum OsType
-{
-	UNKNOWN,
-	NONE,
-	ANANAS,
-	CLOUD_ABI,
-	DRAGON_FLY,
-	FREEBSD,
-	FUCHSIA,
-	IOS,
-	KFREEBSD,
-	LINUX,
-	PS3,
-	MACOSX,
-	NETBSD,
-	OPENBSD,
-	SOLARIS,
-	WIN32,
-	HAIKU,
-	MINIX,
-	RTEMS,
-	NACL,       // Native Client
-	CNK,        // BG/P Compute-Node Kernel
-	AIX,
-	CUDA,
-	NVOPENCL,
-	AMDHSA,
-	PS4,
-	ELFIAMCU,
-	TVOS,
-	WATCHOS,
-	MESA3D,
-	CONTIKI,
-	AMDPAL,
-	HERMITCORE,
-	HURD,
-	WASI,
-	EMSCRIPTEN,
-}
-
-const OsType OS_TYPE = (OsType)($$OS_TYPE);
-const CompilerOptLevel COMPILER_OPT_LEVEL = (CompilerOptLevel)($$COMPILER_OPT_LEVEL);
-const bool BIG_ENDIAN = $$PLATFORM_BIG_ENDIAN;
-const bool I128_SUPPORT = $$PLATFORM_I128_SUPPORTED;
-const bool COMPILER_SAFE_MODE = $$COMPILER_SAFE_MODE;

lib/std/experimental/FrameScheduler.c3

@@ -0,0 +1,94 @@
+module std::experimental::scheduler(<Event>);
+import std::collections, std::thread, std::time;
+
+struct DelayedSchedulerEvent @local
+{
+	inline Event event;
+	Clock execution_time;
+}
+
+fn int DelayedSchedulerEvent.compare_to(self, DelayedSchedulerEvent other) @local
+{
+	switch
+	{
+		case self.execution_time < other.execution_time: return -1;
+		case self.execution_time > other.execution_time: return 1;
+		default: return 0;
+	}
+}
+
+struct FrameScheduler
+{
+	PriorityQueue(<DelayedSchedulerEvent>) delayed_events;
+	List(<Event>) events;
+	List(<Event>) pending_events;
+	bool pending;
+	Mutex mtx;
+}
+
+fn void FrameScheduler.init(&self)
+{
+	self.events.init(mem);
+	self.pending_events.init(mem);
+	self.delayed_events.init(mem);
+	(void)self.mtx.init();
+	bool pending;
+}
+
+macro void FrameScheduler.@destroy(&self; @destruct(Event e))
+{
+	foreach (e : self.events) @destruct(e);
+	foreach (e : self.pending_events) @destruct(e);
+	foreach (e : self.delayed_events.heap) @destruct(e.event);
+	self.events.free();
+	self.pending_events.free();
+	self.delayed_events.free();
+	(void)self.mtx.destroy();
+}
+
+fn void FrameScheduler.queue_delayed_event(&self, Event event, Duration delay)
+{
+	self.mtx.@in_lock()
+	{
+		self.delayed_events.push({ event, clock::now().add_duration(delay)});
+		@atomic_store(self.pending, true);
+	};
+}
+
+fn bool FrameScheduler.has_delayed(&self)
+{
+	self.mtx.@in_lock()
+    {
+        return @ok(self.delayed_events.first());
+    };
+}
+
+fn void FrameScheduler.queue_event(&self, Event event)
+{
+	self.mtx.@in_lock()
+	{
+		self.pending_events.push(event);
+		@atomic_store(self.pending, true);
+	};
+}
+fn Event! FrameScheduler.pop_event(&self)
+{
+	while (true)
+	{
+		if (try event = self.events.pop()) return event;
+    	if (!@atomic_load(self.pending)) return IteratorResult.NO_MORE_ELEMENT?;
+        self.mtx.@in_lock()
+        {
+            self.events.add_all(&self.pending_events);
+            self.pending_events.clear();
+            Clock c = clock::now();
+            while (try top = self.delayed_events.first())
+            {
+                if (top.execution_time > c) break;
+                self.events.push(self.delayed_events.pop()!!);
+            }
+            @atomic_store(self.pending, self.delayed_events.len() > 0);
+            if (!self.events.len()) return IteratorResult.NO_MORE_ELEMENT?;
+        };
+	}
+}

lib/std/hash/adler32.c3

@@ -4,50 +4,50 @@
 
 module std::hash::adler32;
 
-private const uint ADLER_CONST = 65521;
+const uint ADLER_CONST @private = 65521;
 
 struct Adler32
 {
-    uint a;
-    uint b;
+	uint a;
+	uint b;
 }
 
-fn void Adler32.init(Adler32 *this)
+fn void Adler32.init(&self)
 {
-    *this = { 1, 0 };
+	*self = { 1, 0 };
 }
 
-fn void Adler32.updatec(Adler32* this, char c)
+fn void Adler32.updatec(&self, char c)
 {
-    this.a = (this.a + c) % ADLER_CONST;
-    this.b = (this.b + this.a) % ADLER_CONST;
+	self.a = (self.a + c) % ADLER_CONST;
+	self.b = (self.b + self.a) % ADLER_CONST;
 }
 
-fn void Adler32.update(Adler32* this, char[] data)
+fn void Adler32.update(&self, char[] data)
 {
-    uint a = this.a;
-    uint b = this.b;
-    foreach (char x : data)
-    {
-        a = (a + x) % ADLER_CONST;
-        b = (b + a) % ADLER_CONST;
-    }
-    *this = { a, b };
+	uint a = self.a;
+	uint b = self.b;
+	foreach (char x : data)
+	{
+		a = (a + x) % ADLER_CONST;
+		b = (b + a) % ADLER_CONST;
+	}
+	*self = { a, b };
 }
 
-fn uint Adler32.final(Adler32* this)
+fn uint Adler32.final(&self)
 {
-    return (this.b << 16) | this.a;
+	return (self.b << 16) | self.a;
 }
 
 fn uint encode(char[] data)
 {
-    uint a = 1;
-    uint b = 0;
-    foreach (char x : data)
-    {
-        a = (a + x) % ADLER_CONST;
-        b = (b + a) % ADLER_CONST;
-    }
-    return (b << 16) | a;
+	uint a = 1;
+	uint b = 0;
+	foreach (char x : data)
+	{
+		a = (a + x) % ADLER_CONST;
+		b = (b + a) % ADLER_CONST;
+	}
+	return (b << 16) | a;
 }

lib/std/hash/crc32.c3

@@ -5,45 +5,45 @@ module std::hash::crc32;
 
 struct Crc32
 {
-    uint result;
+	uint result;
 }
 
-fn void Crc32.init(Crc32* this, uint seed = 0)
+fn void Crc32.init(&self, uint seed = 0)
 {
-    this.result = ~seed;
+	self.result = ~seed;
 }
 
-fn void Crc32.updatec(Crc32* this, char c)
+fn void Crc32.updatec(&self, char c)
 {
-    this.result = (this.result >> 8) ^ CRC32_TABLE[(this.result ^ c) & 0xFF];
+	self.result = (self.result >> 8) ^ CRC32_TABLE[(self.result ^ c) & 0xFF];
 }
 
-fn void Crc32.update(Crc32* this, char[] data)
+fn void Crc32.update(&self, char[] data)
 {
-    uint result = this.result;
-    foreach (char x : data)
-    {
-        result = (result >> 8) ^ CRC32_TABLE[(result ^ x) & 0xFF];
-    }
-    this.result = result;
+	uint result = self.result;
+	foreach (char x : data)
+	{
+		result = (result >> 8) ^ CRC32_TABLE[(result ^ x) & 0xFF];
+	}
+	self.result = result;
 }
 
-fn uint Crc32.final(Crc32* this)
+fn uint Crc32.final(&self)
 {
-    return ~this.result;
+	return ~self.result;
 }
 
 fn uint encode(char[] data)
 {
-    uint result = ~(uint)(0);
-    foreach (char x : data)
-    {
-        result = (result >> 8) ^ CRC32_TABLE[(result ^ x) & 0xFF];
-    }
+	uint result = ~(uint)(0);
+	foreach (char x : data)
+	{
+		result = (result >> 8) ^ CRC32_TABLE[(result ^ x) & 0xFF];
+	}
 	return ~result;
 }
 
-private const uint[256] CRC32_TABLE = {
+const uint[256] CRC32_TABLE @private = {
 	0x00000000, 0x77073096, 0xee0e612c, 0x990951ba,
 	0x076dc419, 0x706af48f, 0xe963a535, 0x9e6495a3,
 	0x0edb8832, 0x79dcb8a4, 0xe0d5e91e, 0x97d2d988,

lib/std/hash/crc64.c3

@@ -5,107 +5,107 @@ module std::hash::crc64;
 
 struct Crc64
 {
-    ulong result;
+	ulong result;
 }
 
-fn void Crc64.init(Crc64* this, uint seed = 0)
+fn void Crc64.init(&self, uint seed = 0)
 {
-    this.result = seed;
+	self.result = seed;
 }
 
-fn void Crc64.updatec(Crc64* this, char c)
+fn void Crc64.updatec(&self, char c)
 {
-    this.result = (this.result << 8) ^ CRC64_TABLE[(char)((this.result >> 56) ^ c)];
+	self.result = (self.result << 8) ^ CRC64_TABLE[(char)((self.result >> 56) ^ c)];
 }
 
-fn void Crc64.update(Crc64* this, char[] data)
+fn void Crc64.update(&self, char[] data)
 {
-    ulong result = this.result;
-    foreach (char x : data)
-    {
-        result = (result << 8) ^ CRC64_TABLE[(char)((result >> 56) ^ x)];
-    }
-    this.result = result;
+	ulong result = self.result;
+	foreach (char x : data)
+	{
+		result = (result << 8) ^ CRC64_TABLE[(char)((result >> 56) ^ x)];
+	}
+	self.result = result;
 }
 
-fn ulong Crc64.final(Crc64* this)
+fn ulong Crc64.final(&self)
 {
-    return this.result;
+	return self.result;
 }
 
 fn ulong encode(char[] data)
 {
-    ulong result = (ulong)(0);
-    foreach (char x : data)
-    {
-        result = (result << 8) ^ CRC64_TABLE[(char)((result >> 56) ^ x)];
-    }
+	ulong result = (ulong)(0);
+	foreach (char x : data)
+	{
+		result = (result << 8) ^ CRC64_TABLE[(char)((result >> 56) ^ x)];
+	}
 	return result;
 }
 
-private const ulong[256] CRC64_TABLE = {
-    0x0000000000000000, 0x42f0e1eba9ea3693, 0x85e1c3d753d46d26, 0xc711223cfa3e5bb5,
-   	0x493366450e42ecdf, 0x0bc387aea7a8da4c, 0xccd2a5925d9681f9, 0x8e224479f47cb76a,
-   	0x9266cc8a1c85d9be, 0xd0962d61b56fef2d, 0x17870f5d4f51b498, 0x5577eeb6e6bb820b,
-   	0xdb55aacf12c73561, 0x99a54b24bb2d03f2, 0x5eb4691841135847, 0x1c4488f3e8f96ed4,
-   	0x663d78ff90e185ef, 0x24cd9914390bb37c, 0xe3dcbb28c335e8c9, 0xa12c5ac36adfde5a,
-   	0x2f0e1eba9ea36930, 0x6dfeff5137495fa3, 0xaaefdd6dcd770416, 0xe81f3c86649d3285,
-   	0xf45bb4758c645c51, 0xb6ab559e258e6ac2, 0x71ba77a2dfb03177, 0x334a9649765a07e4,
-   	0xbd68d2308226b08e, 0xff9833db2bcc861d, 0x388911e7d1f2dda8, 0x7a79f00c7818eb3b,
-   	0xcc7af1ff21c30bde, 0x8e8a101488293d4d, 0x499b3228721766f8, 0x0b6bd3c3dbfd506b,
-   	0x854997ba2f81e701, 0xc7b97651866bd192, 0x00a8546d7c558a27, 0x4258b586d5bfbcb4,
-   	0x5e1c3d753d46d260, 0x1cecdc9e94ace4f3, 0xdbfdfea26e92bf46, 0x990d1f49c77889d5,
-   	0x172f5b3033043ebf, 0x55dfbadb9aee082c, 0x92ce98e760d05399, 0xd03e790cc93a650a,
-   	0xaa478900b1228e31, 0xe8b768eb18c8b8a2, 0x2fa64ad7e2f6e317, 0x6d56ab3c4b1cd584,
-   	0xe374ef45bf6062ee, 0xa1840eae168a547d, 0x66952c92ecb40fc8, 0x2465cd79455e395b,
-   	0x3821458aada7578f, 0x7ad1a461044d611c, 0xbdc0865dfe733aa9, 0xff3067b657990c3a,
-   	0x711223cfa3e5bb50, 0x33e2c2240a0f8dc3, 0xf4f3e018f031d676, 0xb60301f359dbe0e5,
-   	0xda050215ea6c212f, 0x98f5e3fe438617bc, 0x5fe4c1c2b9b84c09, 0x1d14202910527a9a,
-   	0x93366450e42ecdf0, 0xd1c685bb4dc4fb63, 0x16d7a787b7faa0d6, 0x5427466c1e109645,
-   	0x4863ce9ff6e9f891, 0x0a932f745f03ce02, 0xcd820d48a53d95b7, 0x8f72eca30cd7a324,
-   	0x0150a8daf8ab144e, 0x43a04931514122dd, 0x84b16b0dab7f7968, 0xc6418ae602954ffb,
-   	0xbc387aea7a8da4c0, 0xfec89b01d3679253, 0x39d9b93d2959c9e6, 0x7b2958d680b3ff75,
-   	0xf50b1caf74cf481f, 0xb7fbfd44dd257e8c, 0x70eadf78271b2539, 0x321a3e938ef113aa,
-   	0x2e5eb66066087d7e, 0x6cae578bcfe24bed, 0xabbf75b735dc1058, 0xe94f945c9c3626cb,
-   	0x676dd025684a91a1, 0x259d31cec1a0a732, 0xe28c13f23b9efc87, 0xa07cf2199274ca14,
-   	0x167ff3eacbaf2af1, 0x548f120162451c62, 0x939e303d987b47d7, 0xd16ed1d631917144,
-   	0x5f4c95afc5edc62e, 0x1dbc74446c07f0bd, 0xdaad56789639ab08, 0x985db7933fd39d9b,
-   	0x84193f60d72af34f, 0xc6e9de8b7ec0c5dc, 0x01f8fcb784fe9e69, 0x43081d5c2d14a8fa,
-   	0xcd2a5925d9681f90, 0x8fdab8ce70822903, 0x48cb9af28abc72b6, 0x0a3b7b1923564425,
-   	0x70428b155b4eaf1e, 0x32b26afef2a4998d, 0xf5a348c2089ac238, 0xb753a929a170f4ab,
-   	0x3971ed50550c43c1, 0x7b810cbbfce67552, 0xbc902e8706d82ee7, 0xfe60cf6caf321874,
-   	0xe224479f47cb76a0, 0xa0d4a674ee214033, 0x67c58448141f1b86, 0x253565a3bdf52d15,
-   	0xab1721da49899a7f, 0xe9e7c031e063acec, 0x2ef6e20d1a5df759, 0x6c0603e6b3b7c1ca,
-   	0xf6fae5c07d3274cd, 0xb40a042bd4d8425e, 0x731b26172ee619eb, 0x31ebc7fc870c2f78,
-   	0xbfc9838573709812, 0xfd39626eda9aae81, 0x3a28405220a4f534, 0x78d8a1b9894ec3a7,
-   	0x649c294a61b7ad73, 0x266cc8a1c85d9be0, 0xe17dea9d3263c055, 0xa38d0b769b89f6c6,
-   	0x2daf4f0f6ff541ac, 0x6f5faee4c61f773f, 0xa84e8cd83c212c8a, 0xeabe6d3395cb1a19,
-   	0x90c79d3fedd3f122, 0xd2377cd44439c7b1, 0x15265ee8be079c04, 0x57d6bf0317edaa97,
-   	0xd9f4fb7ae3911dfd, 0x9b041a914a7b2b6e, 0x5c1538adb04570db, 0x1ee5d94619af4648,
-   	0x02a151b5f156289c, 0x4051b05e58bc1e0f, 0x87409262a28245ba, 0xc5b073890b687329,
-   	0x4b9237f0ff14c443, 0x0962d61b56fef2d0, 0xce73f427acc0a965, 0x8c8315cc052a9ff6,
-   	0x3a80143f5cf17f13, 0x7870f5d4f51b4980, 0xbf61d7e80f251235, 0xfd913603a6cf24a6,
-   	0x73b3727a52b393cc, 0x31439391fb59a55f, 0xf652b1ad0167feea, 0xb4a25046a88dc879,
-   	0xa8e6d8b54074a6ad, 0xea16395ee99e903e, 0x2d071b6213a0cb8b, 0x6ff7fa89ba4afd18,
-   	0xe1d5bef04e364a72, 0xa3255f1be7dc7ce1, 0x64347d271de22754, 0x26c49cccb40811c7,
-   	0x5cbd6cc0cc10fafc, 0x1e4d8d2b65facc6f, 0xd95caf179fc497da, 0x9bac4efc362ea149,
-   	0x158e0a85c2521623, 0x577eeb6e6bb820b0, 0x906fc95291867b05, 0xd29f28b9386c4d96,
-   	0xcedba04ad0952342, 0x8c2b41a1797f15d1, 0x4b3a639d83414e64, 0x09ca82762aab78f7,
-   	0x87e8c60fded7cf9d, 0xc51827e4773df90e, 0x020905d88d03a2bb, 0x40f9e43324e99428,
-   	0x2cffe7d5975e55e2, 0x6e0f063e3eb46371, 0xa91e2402c48a38c4, 0xebeec5e96d600e57,
-   	0x65cc8190991cb93d, 0x273c607b30f68fae, 0xe02d4247cac8d41b, 0xa2dda3ac6322e288,
-   	0xbe992b5f8bdb8c5c, 0xfc69cab42231bacf, 0x3b78e888d80fe17a, 0x7988096371e5d7e9,
-   	0xf7aa4d1a85996083, 0xb55aacf12c735610, 0x724b8ecdd64d0da5, 0x30bb6f267fa73b36,
-   	0x4ac29f2a07bfd00d, 0x08327ec1ae55e69e, 0xcf235cfd546bbd2b, 0x8dd3bd16fd818bb8,
-   	0x03f1f96f09fd3cd2, 0x41011884a0170a41, 0x86103ab85a2951f4, 0xc4e0db53f3c36767,
-   	0xd8a453a01b3a09b3, 0x9a54b24bb2d03f20, 0x5d45907748ee6495, 0x1fb5719ce1045206,
-   	0x919735e51578e56c, 0xd367d40ebc92d3ff, 0x1476f63246ac884a, 0x568617d9ef46bed9,
-   	0xe085162ab69d5e3c, 0xa275f7c11f7768af, 0x6564d5fde549331a, 0x279434164ca30589,
-   	0xa9b6706fb8dfb2e3, 0xeb46918411358470, 0x2c57b3b8eb0bdfc5, 0x6ea7525342e1e956,
-   	0x72e3daa0aa188782, 0x30133b4b03f2b111, 0xf7021977f9cceaa4, 0xb5f2f89c5026dc37,
-   	0x3bd0bce5a45a6b5d, 0x79205d0e0db05dce, 0xbe317f32f78e067b, 0xfcc19ed95e6430e8,
-   	0x86b86ed5267cdbd3, 0xc4488f3e8f96ed40, 0x0359ad0275a8b6f5, 0x41a94ce9dc428066,
-   	0xcf8b0890283e370c, 0x8d7be97b81d4019f, 0x4a6acb477bea5a2a, 0x089a2aacd2006cb9,
-   	0x14dea25f3af9026d, 0x562e43b4931334fe, 0x913f6188692d6f4b, 0xd3cf8063c0c759d8,
-   	0x5dedc41a34bbeeb2, 0x1f1d25f19d51d821, 0xd80c07cd676f8394, 0x9afce626ce85b507,
+const ulong[256] CRC64_TABLE @private = {
+	0x0000000000000000, 0x42f0e1eba9ea3693, 0x85e1c3d753d46d26, 0xc711223cfa3e5bb5,
+	0x493366450e42ecdf, 0x0bc387aea7a8da4c, 0xccd2a5925d9681f9, 0x8e224479f47cb76a,
+	0x9266cc8a1c85d9be, 0xd0962d61b56fef2d, 0x17870f5d4f51b498, 0x5577eeb6e6bb820b,
+	0xdb55aacf12c73561, 0x99a54b24bb2d03f2, 0x5eb4691841135847, 0x1c4488f3e8f96ed4,
+	0x663d78ff90e185ef, 0x24cd9914390bb37c, 0xe3dcbb28c335e8c9, 0xa12c5ac36adfde5a,
+	0x2f0e1eba9ea36930, 0x6dfeff5137495fa3, 0xaaefdd6dcd770416, 0xe81f3c86649d3285,
+	0xf45bb4758c645c51, 0xb6ab559e258e6ac2, 0x71ba77a2dfb03177, 0x334a9649765a07e4,
+	0xbd68d2308226b08e, 0xff9833db2bcc861d, 0x388911e7d1f2dda8, 0x7a79f00c7818eb3b,
+	0xcc7af1ff21c30bde, 0x8e8a101488293d4d, 0x499b3228721766f8, 0x0b6bd3c3dbfd506b,
+	0x854997ba2f81e701, 0xc7b97651866bd192, 0x00a8546d7c558a27, 0x4258b586d5bfbcb4,
+	0x5e1c3d753d46d260, 0x1cecdc9e94ace4f3, 0xdbfdfea26e92bf46, 0x990d1f49c77889d5,
+	0x172f5b3033043ebf, 0x55dfbadb9aee082c, 0x92ce98e760d05399, 0xd03e790cc93a650a,
+	0xaa478900b1228e31, 0xe8b768eb18c8b8a2, 0x2fa64ad7e2f6e317, 0x6d56ab3c4b1cd584,
+	0xe374ef45bf6062ee, 0xa1840eae168a547d, 0x66952c92ecb40fc8, 0x2465cd79455e395b,
+	0x3821458aada7578f, 0x7ad1a461044d611c, 0xbdc0865dfe733aa9, 0xff3067b657990c3a,
+	0x711223cfa3e5bb50, 0x33e2c2240a0f8dc3, 0xf4f3e018f031d676, 0xb60301f359dbe0e5,
+	0xda050215ea6c212f, 0x98f5e3fe438617bc, 0x5fe4c1c2b9b84c09, 0x1d14202910527a9a,
+	0x93366450e42ecdf0, 0xd1c685bb4dc4fb63, 0x16d7a787b7faa0d6, 0x5427466c1e109645,
+	0x4863ce9ff6e9f891, 0x0a932f745f03ce02, 0xcd820d48a53d95b7, 0x8f72eca30cd7a324,
+	0x0150a8daf8ab144e, 0x43a04931514122dd, 0x84b16b0dab7f7968, 0xc6418ae602954ffb,
+	0xbc387aea7a8da4c0, 0xfec89b01d3679253, 0x39d9b93d2959c9e6, 0x7b2958d680b3ff75,
+	0xf50b1caf74cf481f, 0xb7fbfd44dd257e8c, 0x70eadf78271b2539, 0x321a3e938ef113aa,
+	0x2e5eb66066087d7e, 0x6cae578bcfe24bed, 0xabbf75b735dc1058, 0xe94f945c9c3626cb,
+	0x676dd025684a91a1, 0x259d31cec1a0a732, 0xe28c13f23b9efc87, 0xa07cf2199274ca14,
+	0x167ff3eacbaf2af1, 0x548f120162451c62, 0x939e303d987b47d7, 0xd16ed1d631917144,
+	0x5f4c95afc5edc62e, 0x1dbc74446c07f0bd, 0xdaad56789639ab08, 0x985db7933fd39d9b,
+	0x84193f60d72af34f, 0xc6e9de8b7ec0c5dc, 0x01f8fcb784fe9e69, 0x43081d5c2d14a8fa,
+	0xcd2a5925d9681f90, 0x8fdab8ce70822903, 0x48cb9af28abc72b6, 0x0a3b7b1923564425,
+	0x70428b155b4eaf1e, 0x32b26afef2a4998d, 0xf5a348c2089ac238, 0xb753a929a170f4ab,
+	0x3971ed50550c43c1, 0x7b810cbbfce67552, 0xbc902e8706d82ee7, 0xfe60cf6caf321874,
+	0xe224479f47cb76a0, 0xa0d4a674ee214033, 0x67c58448141f1b86, 0x253565a3bdf52d15,
+	0xab1721da49899a7f, 0xe9e7c031e063acec, 0x2ef6e20d1a5df759, 0x6c0603e6b3b7c1ca,
+	0xf6fae5c07d3274cd, 0xb40a042bd4d8425e, 0x731b26172ee619eb, 0x31ebc7fc870c2f78,
+	0xbfc9838573709812, 0xfd39626eda9aae81, 0x3a28405220a4f534, 0x78d8a1b9894ec3a7,
+	0x649c294a61b7ad73, 0x266cc8a1c85d9be0, 0xe17dea9d3263c055, 0xa38d0b769b89f6c6,
+	0x2daf4f0f6ff541ac, 0x6f5faee4c61f773f, 0xa84e8cd83c212c8a, 0xeabe6d3395cb1a19,
+	0x90c79d3fedd3f122, 0xd2377cd44439c7b1, 0x15265ee8be079c04, 0x57d6bf0317edaa97,
+	0xd9f4fb7ae3911dfd, 0x9b041a914a7b2b6e, 0x5c1538adb04570db, 0x1ee5d94619af4648,
+	0x02a151b5f156289c, 0x4051b05e58bc1e0f, 0x87409262a28245ba, 0xc5b073890b687329,
+	0x4b9237f0ff14c443, 0x0962d61b56fef2d0, 0xce73f427acc0a965, 0x8c8315cc052a9ff6,
+	0x3a80143f5cf17f13, 0x7870f5d4f51b4980, 0xbf61d7e80f251235, 0xfd913603a6cf24a6,
+	0x73b3727a52b393cc, 0x31439391fb59a55f, 0xf652b1ad0167feea, 0xb4a25046a88dc879,
+	0xa8e6d8b54074a6ad, 0xea16395ee99e903e, 0x2d071b6213a0cb8b, 0x6ff7fa89ba4afd18,
+	0xe1d5bef04e364a72, 0xa3255f1be7dc7ce1, 0x64347d271de22754, 0x26c49cccb40811c7,
+	0x5cbd6cc0cc10fafc, 0x1e4d8d2b65facc6f, 0xd95caf179fc497da, 0x9bac4efc362ea149,
+	0x158e0a85c2521623, 0x577eeb6e6bb820b0, 0x906fc95291867b05, 0xd29f28b9386c4d96,
+	0xcedba04ad0952342, 0x8c2b41a1797f15d1, 0x4b3a639d83414e64, 0x09ca82762aab78f7,
+	0x87e8c60fded7cf9d, 0xc51827e4773df90e, 0x020905d88d03a2bb, 0x40f9e43324e99428,
+	0x2cffe7d5975e55e2, 0x6e0f063e3eb46371, 0xa91e2402c48a38c4, 0xebeec5e96d600e57,
+	0x65cc8190991cb93d, 0x273c607b30f68fae, 0xe02d4247cac8d41b, 0xa2dda3ac6322e288,
+	0xbe992b5f8bdb8c5c, 0xfc69cab42231bacf, 0x3b78e888d80fe17a, 0x7988096371e5d7e9,
+	0xf7aa4d1a85996083, 0xb55aacf12c735610, 0x724b8ecdd64d0da5, 0x30bb6f267fa73b36,
+	0x4ac29f2a07bfd00d, 0x08327ec1ae55e69e, 0xcf235cfd546bbd2b, 0x8dd3bd16fd818bb8,
+	0x03f1f96f09fd3cd2, 0x41011884a0170a41, 0x86103ab85a2951f4, 0xc4e0db53f3c36767,
+	0xd8a453a01b3a09b3, 0x9a54b24bb2d03f20, 0x5d45907748ee6495, 0x1fb5719ce1045206,
+	0x919735e51578e56c, 0xd367d40ebc92d3ff, 0x1476f63246ac884a, 0x568617d9ef46bed9,
+	0xe085162ab69d5e3c, 0xa275f7c11f7768af, 0x6564d5fde549331a, 0x279434164ca30589,
+	0xa9b6706fb8dfb2e3, 0xeb46918411358470, 0x2c57b3b8eb0bdfc5, 0x6ea7525342e1e956,
+	0x72e3daa0aa188782, 0x30133b4b03f2b111, 0xf7021977f9cceaa4, 0xb5f2f89c5026dc37,
+	0x3bd0bce5a45a6b5d, 0x79205d0e0db05dce, 0xbe317f32f78e067b, 0xfcc19ed95e6430e8,
+	0x86b86ed5267cdbd3, 0xc4488f3e8f96ed40, 0x0359ad0275a8b6f5, 0x41a94ce9dc428066,
+	0xcf8b0890283e370c, 0x8d7be97b81d4019f, 0x4a6acb477bea5a2a, 0x089a2aacd2006cb9,
+	0x14dea25f3af9026d, 0x562e43b4931334fe, 0x913f6188692d6f4b, 0xd3cf8063c0c759d8,
+	0x5dedc41a34bbeeb2, 0x1f1d25f19d51d821, 0xd80c07cd676f8394, 0x9afce626ce85b507,
 };

lib/std/hash/fnv32a.c3

@@ -0,0 +1,41 @@
+// Copyright (c) 2021 Christoffer Lerno. All rights reserved.
+// Use of this source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+module std::hash::fnv32a;
+
+distinct Fnv32a = uint;
+
+const FNV32A_START @private = 0x811c9dc5;
+const FNV32A_MUL @private = 0x01000193;
+
+macro void update(h, char x) @private => *h = (*h ^ ($typeof(*h))x) * FNV32A_MUL;
+
+fn void Fnv32a.init(&self)
+{
+	*self = FNV32A_START;
+}
+
+fn void Fnv32a.update(&self, char[] data)
+{
+	Fnv32a h = *self;
+	foreach (char x : data)
+	{
+		update(&h, x);
+	}
+	*self = h;
+}
+
+macro void Fnv32a.update_char(&self, char c)
+{
+	update(self, c);
+}
+
+fn uint encode(char[] data)
+{
+	uint h = FNV32A_START;
+	foreach (char x : data)
+	{
+		update(&h, x);
+	}
+	return h;
+}

lib/std/hash/fnv64a.c3

@@ -0,0 +1,41 @@
+// Copyright (c) 2021 Christoffer Lerno. All rights reserved.
+// Use of this source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+module std::hash::fnv64a;
+
+distinct Fnv64a = ulong;
+
+const FNV64A_START @private = 0xcbf29ce484222325;
+const FNV64A_MUL @private = 0x00000100000001b3;
+
+macro void update(h, char x) @private => *h = (*h ^ ($typeof(*h))x) * FNV64A_MUL;
+
+fn void Fnv64a.init(&self)
+{
+	*self = FNV64A_START;
+}
+
+fn void Fnv64a.update(&self, char[] data)
+{
+	Fnv64a h = *self;
+	foreach (char x : data)
+	{
+		update(&h, x);
+	}
+	*self = h;
+}
+
+macro void Fnv64a.update_char(&self, char c)
+{
+	update(self, c);
+}
+
+fn ulong encode(char[] data)
+{
+	ulong h = FNV64A_START;
+	foreach (char x : data)
+	{
+		update(&h, x);
+	}
+	return h;
+}

lib/std/hash/hmac.c3

@@ -0,0 +1,107 @@
+module std::hash::hmac(<HashAlg, HASH_BYTES, BLOCK_BYTES>);
+import std::crypto;
+
+struct Hmac
+{
+	HashAlg a, b;
+}
+
+fn char[HASH_BYTES] hash(char[] key, char[] message)
+{
+	Hmac hmac @noinit;
+	hmac.init(key);
+	hmac.update(message);
+	return hmac.final();
+}
+
+<*
+ @require output.len > 0 "Output must be greater than zero"
+ @require output.len < int.max / HASH_BYTES "Output is too large"
+*>
+fn void pbkdf2(char[] pw, char[] salt, uint iterations, char[] output)
+{
+	usz l = output.len / HASH_BYTES;
+	usz r = output.len % HASH_BYTES;
+
+	Hmac hmac;
+	hmac.init(pw);
+
+	char[] dst_curr = output;
+	for (usz i = 1; i <= l; i++)
+	{
+		@derive(&hmac, salt, iterations, i, dst_curr[:HASH_BYTES]);
+		dst_curr = dst_curr[HASH_BYTES..];
+	}
+
+	if (r > 0)
+	{
+		char[HASH_BYTES] tmp;
+		@derive(&hmac, salt, iterations, l + 1, &tmp);
+		dst_curr[..] = tmp[:dst_curr.len];
+		mem::zero_volatile(&tmp);
+	}
+}
+
+fn void Hmac.init(&self, char[] key)
+{
+	char[BLOCK_BYTES] buffer;
+	if (key.len > BLOCK_BYTES)
+	{
+		self.a.init();
+		self.a.update(key);
+		buffer[:HASH_BYTES] = self.a.final()[..];
+	}
+	else
+	{
+		buffer[:key.len] = key[..];
+	}
+
+	foreach (&b : buffer) *b ^= IPAD;
+
+	self.a.init();
+	self.a.update(&buffer);
+
+	foreach (&b : buffer) *b ^= IPAD ^ OPAD;
+
+	self.b.init();
+	self.b.update(&buffer);
+
+	mem::zero_volatile(&buffer);
+}
+
+fn void Hmac.update(&self, char[] data)
+{
+	self.a.update(data);
+}
+
+fn char[HASH_BYTES] Hmac.final(&self)
+{
+	self.b.update(&&self.a.final());
+	return self.b.final();
+}
+
+const IPAD @private = 0x36;
+const OPAD @private = 0x5C;
+
+macro @derive(Hmac *hmac_start, char[] salt, uint iterations, usz index, char[] out)
+{
+	assert(out.len == HASH_BYTES);
+	char[HASH_BYTES] tmp @noinit;
+	defer mem::zero_volatile(&tmp);
+	Hmac hmac  = *hmac_start;
+	hmac.update(salt);
+	UIntBE be = { (uint)index };
+	hmac.update(&&bitcast(be, char[4]));
+	tmp = hmac.final();
+	out[..] = tmp;
+	for (int it = 1; it < iterations; it++)
+	{
+		hmac = *hmac_start;
+		hmac.update(&tmp);
+		tmp = hmac.final();
+		foreach (i, v : tmp)
+		{
+			out[i] ^= v;
+		}
+	}
+}

lib/std/hash/md5.c3

@@ -0,0 +1,225 @@
+module std::hash::md5;
+import std::hash::hmac;
+import std::bits;
+
+const BLOCK_BYTES = 64;
+const HASH_BYTES = 16;
+
+struct Md5
+{
+	uint lo, hi;
+	uint a, b, c, d;
+	char[64] buffer;
+	uint[16] block;
+}
+
+def HmacMd5 = Hmac(<Md5, HASH_BYTES, BLOCK_BYTES>);
+def hmac = hmac::hash(<Md5, HASH_BYTES, BLOCK_BYTES>);
+def pbkdf2 = hmac::pbkdf2(<Md5, HASH_BYTES, BLOCK_BYTES>);
+
+fn char[HASH_BYTES] hash(char[] data)
+{
+	Md5 md5;
+	md5.init();
+	md5.update(data);
+	return md5.final();
+}
+
+fn void Md5.init(&self)
+{
+	self.a = 0x67452301;
+	self.b = 0xefcdab89;
+	self.c = 0x98badcfe;
+	self.d = 0x10325476;
+
+	self.lo = 0;
+	self.hi = 0;
+}
+
+
+fn void Md5.update(&ctx, char[] data)
+{
+	uint saved_lo = ctx.lo;
+	if ((ctx.lo = (saved_lo + data.len) & 0x1fffffff) < saved_lo) ctx.hi++;
+	ctx.hi += data.len >> 29;
+
+	usz used = (usz)saved_lo & 0x3f;
+
+	if (used)
+	{
+		usz available = 64 - used;
+
+		if (data.len < available)
+		{
+			ctx.buffer[used:data.len] = data[..];
+			return;
+		}
+		ctx.buffer[used:available] = data[:available];
+		data = data[available..];
+		body(ctx, &ctx.buffer, 64);
+	}
+
+	if (data.len >= 64)
+	{
+		data = body(ctx, data, data.len & ~(usz)0x3f)[:data.len & 0x3f];
+	}
+	ctx.buffer[:data.len] = data[..];
+}
+
+fn char[HASH_BYTES] Md5.final(&ctx)
+{
+	usz used = (usz)ctx.lo & 0x3f;
+	ctx.buffer[used++] = 0x80;
+
+	usz available = 64 - used;
+
+	if (available < 8)
+	{
+		ctx.buffer[used:available] = 0;
+		body(ctx, &ctx.buffer, 64);
+		used = 0;
+		available = 64;
+	}
+	ctx.buffer[used:available - 8] = 0;
+
+	ctx.lo <<= 3;
+	ctx.buffer[56:4] = bitcast(ctx.lo, char[4])[..];
+	ctx.buffer[60:4] = bitcast(ctx.hi, char[4])[..];
+
+	body(ctx, &ctx.buffer, 64);
+
+	char[16] res @noinit;
+	res[0:4] = bitcast(ctx.a, char[4]);
+	res[4:4] = bitcast(ctx.b, char[4]);
+	res[8:4] = bitcast(ctx.c, char[4]);
+	res[12:4] = bitcast(ctx.d, char[4]);
+	*ctx = {};
+	return res;
+}
+
+module std::hash::md5 @private;
+
+// Implementation
+macro @f(x, y, z) => z ^ (x & (y ^ z));
+macro @g(x, y, z) => y ^ (z & (x ^ y));
+macro @h(x, y, z) => (x ^ y) ^ z;
+macro @h2(x, y, z) => x ^ (y ^ z);
+macro @i(x, y, z) => y ^ (x | ~z);
+
+macro @step(#f, a, b, c, d, ptr, n, t, s)
+{
+	*a += #f(b, c, d) + @unaligned_load(*(uint *)&ptr[n * 4], 2) + t;
+	*a = (*a << s) | ((*a & 0xffffffff) >> (32 - s));
+	*a += b;
+}
+
+
+fn char* body(Md5* ctx, void* data, usz size)
+{
+	char* ptr;
+	uint a, b, c, d;
+	uint saved_a, saved_b, saved_c, saved_d;
+	ptr = data;
+	a = ctx.a;
+	b = ctx.b;
+	c = ctx.c;
+	d = ctx.d;
+
+	do
+	{
+		saved_a = a;
+		saved_b = b;
+		saved_c = c;
+		saved_d = d;
+
+/* Round 1 */
+		@step(@f, &a, b, c, d, ptr, 0, 0xd76aa478, 7)  ;
+		@step(@f, &d, a, b, c, ptr, 1, 0xe8c7b756, 12) ;
+		@step(@f, &c, d, a, b, ptr, 2, 0x242070db, 17) ;
+		@step(@f, &b, c, d, a, ptr, 3, 0xc1bdceee, 22) ;
+		@step(@f, &a, b, c, d, ptr, 4, 0xf57c0faf, 7)  ;
+		@step(@f, &d, a, b, c, ptr, 5, 0x4787c62a, 12) ;
+		@step(@f, &c, d, a, b, ptr, 6, 0xa8304613, 17) ;
+		@step(@f, &b, c, d, a, ptr, 7, 0xfd469501, 22) ;
+		@step(@f, &a, b, c, d, ptr, 8, 0x698098d8, 7)  ;
+		@step(@f, &d, a, b, c, ptr, 9, 0x8b44f7af, 12) ;
+		@step(@f, &c, d, a, b, ptr, 10, 0xffff5bb1, 17);
+		@step(@f, &b, c, d, a, ptr, 11, 0x895cd7be, 22);
+		@step(@f, &a, b, c, d, ptr, 12, 0x6b901122, 7) ;
+		@step(@f, &d, a, b, c, ptr, 13, 0xfd987193, 12);
+		@step(@f, &c, d, a, b, ptr, 14, 0xa679438e, 17);
+		@step(@f, &b, c, d, a, ptr, 15, 0x49b40821, 22);
+
+/* Round 2 */
+		@step(@g, &a, b, c, d, ptr, 1, 0xf61e2562, 5)  ;
+		@step(@g, &d, a, b, c, ptr, 6, 0xc040b340, 9)  ;
+		@step(@g, &c, d, a, b, ptr, 11, 0x265e5a51, 14);
+		@step(@g, &b, c, d, a, ptr, 0, 0xe9b6c7aa, 20) ;
+		@step(@g, &a, b, c, d, ptr, 5, 0xd62f105d, 5)  ;
+		@step(@g, &d, a, b, c, ptr, 10, 0x02441453, 9) ;
+		@step(@g, &c, d, a, b, ptr, 15, 0xd8a1e681, 14);
+		@step(@g, &b, c, d, a, ptr, 4, 0xe7d3fbc8, 20) ;
+		@step(@g, &a, b, c, d, ptr, 9, 0x21e1cde6, 5)  ;
+		@step(@g, &d, a, b, c, ptr, 14, 0xc33707d6, 9) ;
+		@step(@g, &c, d, a, b, ptr, 3, 0xf4d50d87, 14) ;
+		@step(@g, &b, c, d, a, ptr, 8, 0x455a14ed, 20) ;
+		@step(@g, &a, b, c, d, ptr, 13, 0xa9e3e905, 5) ;
+		@step(@g, &d, a, b, c, ptr, 2, 0xfcefa3f8, 9)  ;
+		@step(@g, &c, d, a, b, ptr, 7, 0x676f02d9, 14) ;
+		@step(@g, &b, c, d, a, ptr, 12, 0x8d2a4c8a, 20);
+
+/* Round 3 */
+		@step(@h,  &a, b, c, d, ptr, 5, 0xfffa3942, 4);
+		@step(@h2, &d, a, b, c, ptr, 8, 0x8771f681, 11);
+		@step(@h,  &c, d, a, b, ptr, 11, 0x6d9d6122, 16);
+		@step(@h2, &b, c, d, a, ptr, 14, 0xfde5380c, 23);
+		@step(@h,  &a, b, c, d, ptr, 1, 0xa4beea44, 4);
+		@step(@h2, &d, a, b, c, ptr, 4, 0x4bdecfa9, 11);
+		@step(@h,  &c, d, a, b, ptr, 7, 0xf6bb4b60, 16);
+		@step(@h2, &b, c, d, a, ptr, 10, 0xbebfbc70, 23);
+		@step(@h,  &a, b, c, d, ptr, 13, 0x289b7ec6, 4)   ;
+		@step(@h2, &d, a, b, c, ptr, 0, 0xeaa127fa, 11)  ;
+		@step(@h,  &c, d, a, b, ptr, 3, 0xd4ef3085, 16)   ;
+		@step(@h2, &b, c, d, a, ptr, 6, 0x04881d05, 23)  ;
+		@step(@h,  &a, b, c, d, ptr, 9, 0xd9d4d039, 4)   ;
+		@step(@h2, &d, a, b, c, ptr, 12, 0xe6db99e5, 11) ;
+		@step(@h,  &c, d, a, b, ptr, 15, 0x1fa27cf8, 16)  ;
+		@step(@h2, &b, c, d, a, ptr, 2, 0xc4ac5665, 23)  ;
+
+/* Round 4 */
+		@step(@i, &a, b, c, d, ptr, 0, 0xf4292244, 6)   ;
+		@step(@i, &d, a, b, c, ptr, 7, 0x432aff97, 10)  ;
+		@step(@i, &c, d, a, b, ptr, 14, 0xab9423a7, 15) ;
+		@step(@i, &b, c, d, a, ptr, 5, 0xfc93a039, 21)  ;
+		@step(@i, &a, b, c, d, ptr, 12, 0x655b59c3, 6)  ;
+		@step(@i, &d, a, b, c, ptr, 3, 0x8f0ccc92, 10)  ;
+		@step(@i, &c, d, a, b, ptr, 10, 0xffeff47d, 15) ;
+		@step(@i, &b, c, d, a, ptr, 1, 0x85845dd1, 21)  ;
+		@step(@i, &a, b, c, d, ptr, 8, 0x6fa87e4f, 6)   ;
+		@step(@i, &d, a, b, c, ptr, 15, 0xfe2ce6e0, 10) ;
+		@step(@i, &c, d, a, b, ptr, 6, 0xa3014314, 15)  ;
+		@step(@i, &b, c, d, a, ptr, 13, 0x4e0811a1, 21) ;
+		@step(@i, &a, b, c, d, ptr, 4, 0xf7537e82, 6)   ;
+		@step(@i, &d, a, b, c, ptr, 11, 0xbd3af235, 10) ;
+		@step(@i, &c, d, a, b, ptr, 2, 0x2ad7d2bb, 15)  ;
+		@step(@i, &b, c, d, a, ptr, 9, 0xeb86d391, 21)  ;
+
+		a += saved_a;
+		b += saved_b;
+		c += saved_c;
+		d += saved_d;
+
+		ptr += 64;
+
+	} while (size -= 64);
+
+	ctx.a = a;
+	ctx.b = b;
+	ctx.c = c;
+	ctx.d = d;
+
+	return ptr;
+}
+
+
+

lib/std/hash/sha1.c3

@@ -0,0 +1,257 @@
+// Copyright (c) 2021 Christoffer Lerno. All rights reserved.
+// Use of this source code is governed by the MIT license
+// a copy of which can be found in the LICENSE_STDLIB file.
+//
+// Implementation was off Steve Reid's SHA-1 C implementation
+
+module std::hash::sha1;
+import std::hash::hmac;
+import std::bits;
+
+const BLOCK_BYTES = 64;
+const HASH_BYTES = 20;
+
+struct Sha1
+{
+	uint[5] state;
+	uint[2] count;
+	char[BLOCK_BYTES] buffer;
+}
+
+def HmacSha1 = Hmac(<Sha1, HASH_BYTES, BLOCK_BYTES>);
+def hmac = hmac::hash(<Sha1, HASH_BYTES, BLOCK_BYTES>);
+def pbkdf2 = hmac::pbkdf2(<Sha1, HASH_BYTES, BLOCK_BYTES>);
+
+fn char[HASH_BYTES] hash(char[] data)
+{
+	Sha1 sha1 @noinit;
+	sha1.init();
+	sha1.update(data);
+	return sha1.final();
+}
+
+fn void Sha1.init(&self)
+{
+	// SHA1 initialization constants
+	*self = {
+		.state = {
+			0x67452301,
+			0xEFCDAB89,
+			0x98BADCFE,
+			0x10325476,
+			0xC3D2E1F0
+		}
+	};
+}
+
+<*
+ @param [in] data
+ @require data.len <= uint.max
+*>
+fn void Sha1.update(&self, char[] data)
+{
+	uint j = self.count[0];
+	uint len = data.len;
+	if ((self.count[0] += len << 3) < j) self.count[1]++;
+	self.count[1] += len >> 29;
+	j = (j >> 3) & 63;
+	uint i;
+	if (j + len > 63)
+	{
+		i = 64 - j;
+		self.buffer[j..] = data[:i];
+		sha1_transform(&self.state, &self.buffer);
+		for (; i + 63 < len; i += 64)
+		{
+			sha1_transform(&self.state, &data[i]);
+		}
+		j = 0;
+	}
+	self.buffer[j:len - i] = data[i..];
+}
+
+
+fn char[HASH_BYTES] Sha1.final(&self)
+{
+	char[8] finalcount;
+	for (uint i = 0; i < 8; i++)
+	{
+		finalcount[i] = (char)((self.count[(i >= 4 ? 0 : 1)] >> ((3 - (i & 3)) * 8)) & 0xFF);
+	}
+	self.update((char[]){ 0o200 });
+	while ((self.count[0] & 504) != 448)
+	{
+		self.update((char[]){ 0 });
+	}
+
+	self.update(&finalcount);
+	char[HASH_BYTES] digest;
+	for (uint i = 0; i < HASH_BYTES; i++)
+	{
+		digest[i] = (char)((self.state[i >> 2] >> ((3 - (i & 3)) * 8)) & 0xFF);
+	}
+
+	// Clear mem
+	*self = {};
+	finalcount = {};
+	return digest;
+}
+
+union Long16 @local
+{
+	char[BLOCK_BYTES] c;
+	uint[16] l;
+}
+
+macro blk(Long16* block, i) @local
+{
+	return (block.l[i & 15] = (block.l[(i + 13) & 15] ^ block.l[(i + 8) & 15]
+		^ block.l[(i + 2) & 15] ^ block.l[i & 15]).rotl(1));
+}
+
+macro blk0(Long16* block, i) @local
+{
+	$if env::BIG_ENDIAN:
+		return block.l[i];
+	$else
+		return block.l[i] = (block.l[i].rotl(24) & 0xFF00FF00)
+			| (block.l[i].rotl(8) & 0x00FF00FF);
+	$endif
+}
+
+macro r0(Long16* block, uint v, uint* wref, uint x, uint y, uint* z, uint i) @local
+{
+	var w = *wref;
+	*z += ((w & (x ^ y)) ^ y) + blk0(block, i) + 0x5A827999 + v.rotl(5);
+	*wref = w.rotl(30);
+}
+
+macro r1(Long16* block, uint v, uint* wref, uint x, uint y, uint* z, uint i) @local
+{
+	var w = *wref;
+	*z += ((w & (x ^ y)) ^ y) + blk(block, i) + 0x5A827999 + v.rotl(5);
+	*wref = w.rotl(30);
+}
+
+macro r2(Long16* block, uint v, uint* wref, uint x, uint y, uint* z, uint i) @local
+{
+	var w = *wref;
+	*z += (w ^ x ^ y) + blk(block, i) + 0x6ED9EBA1 + v.rotl(5);
+	*wref = w.rotl(30);
+}
+
+macro r3(Long16* block, uint v, uint* wref, uint x, uint y, uint* z, uint i) @local
+{
+	var w = *wref;
+	*z += (((w | x) & y) | (w & x)) + blk(block, i) + 0x8F1BBCDC + v.rotl(5);
+	*wref = w.rotl(30);
+}
+
+macro r4(Long16* block, uint v, uint* wref, uint x, uint y, uint* z, uint i) @local
+{
+	var w = *wref;
+	*z += (w ^ x ^ y) + blk(block, i) + 0xCA62C1D6 + v.rotl(5);
+	*wref = w.rotl(30);
+}
+
+<*
+ @param [&inout] state
+ @param [&in] buffer
+*>
+fn void sha1_transform(uint[5]* state, char* buffer) @local
+{
+	Long16 block;
+	block.c[..] = buffer[:64];
+	uint a = (*state)[0];
+	uint b = (*state)[1];
+	uint c = (*state)[2];
+	uint d = (*state)[3];
+	uint e = (*state)[4];
+	r0(&block, a, &b, c, d, &e, 0);
+	r0(&block, e, &a, b, c, &d, 1);
+	r0(&block, d, &e, a, b, &c, 2);
+	r0(&block, c, &d, e, a, &b, 3);
+	r0(&block, b, &c, d, e, &a, 4);
+	r0(&block, a, &b, c, d, &e, 5);
+	r0(&block, e, &a, b, c, &d, 6);
+	r0(&block, d, &e, a, b, &c, 7);
+	r0(&block, c, &d, e, a, &b, 8);
+	r0(&block, b, &c, d, e, &a, 9);
+	r0(&block, a, &b, c, d, &e, 10);
+	r0(&block, e, &a, b, c, &d, 11);
+	r0(&block, d, &e, a, b, &c, 12);
+	r0(&block, c, &d, e, a, &b, 13);
+	r0(&block, b, &c, d, e, &a, 14);
+	r0(&block, a, &b, c, d, &e, 15);
+	r1(&block, e, &a, b, c, &d, 16);
+	r1(&block, d, &e, a, b, &c, 17);
+	r1(&block, c, &d, e, a, &b, 18);
+	r1(&block, b, &c, d, e, &a, 19);
+	r2(&block, a, &b, c, d, &e, 20);
+	r2(&block, e, &a, b, c, &d, 21);
+	r2(&block, d, &e, a, b, &c, 22);
+	r2(&block, c, &d, e, a, &b, 23);
+	r2(&block, b, &c, d, e, &a, 24);
+	r2(&block, a, &b, c, d, &e, 25);
+	r2(&block, e, &a, b, c, &d, 26);
+	r2(&block, d, &e, a, b, &c, 27);
+	r2(&block, c, &d, e, a, &b, 28);
+	r2(&block, b, &c, d, e, &a, 29);
+	r2(&block, a, &b, c, d, &e, 30);
+	r2(&block, e, &a, b, c, &d, 31);
+	r2(&block, d, &e, a, b, &c, 32);
+	r2(&block, c, &d, e, a, &b, 33);
+	r2(&block, b, &c, d, e, &a, 34);
+	r2(&block, a, &b, c, d, &e, 35);
+	r2(&block, e, &a, b, c, &d, 36);
+	r2(&block, d, &e, a, b, &c, 37);
+	r2(&block, c, &d, e, a, &b, 38);
+	r2(&block, b, &c, d, e, &a, 39);
+	r3(&block, a, &b, c, d, &e, 40);
+	r3(&block, e, &a, b, c, &d, 41);
+	r3(&block, d, &e, a, b, &c, 42);
+	r3(&block, c, &d, e, a, &b, 43);
+	r3(&block, b, &c, d, e, &a, 44);
+	r3(&block, a, &b, c, d, &e, 45);
+	r3(&block, e, &a, b, c, &d, 46);
+	r3(&block, d, &e, a, b, &c, 47);
+	r3(&block, c, &d, e, a, &b, 48);
+	r3(&block, b, &c, d, e, &a, 49);
+	r3(&block, a, &b, c, d, &e, 50);
+	r3(&block, e, &a, b, c, &d, 51);
+	r3(&block, d, &e, a, b, &c, 52);
+	r3(&block, c, &d, e, a, &b, 53);
+	r3(&block, b, &c, d, e, &a, 54);
+	r3(&block, a, &b, c, d, &e, 55);
+	r3(&block, e, &a, b, c, &d, 56);
+	r3(&block, d, &e, a, b, &c, 57);
+	r3(&block, c, &d, e, a, &b, 58);
+	r3(&block, b, &c, d, e, &a, 59);
+	r4(&block, a, &b, c, d, &e, 60);
+	r4(&block, e, &a, b, c, &d, 61);
+	r4(&block, d, &e, a, b, &c, 62);
+	r4(&block, c, &d, e, a, &b, 63);
+	r4(&block, b, &c, d, e, &a, 64);
+	r4(&block, a, &b, c, d, &e, 65);
+	r4(&block, e, &a, b, c, &d, 66);
+	r4(&block, d, &e, a, b, &c, 67);
+	r4(&block, c, &d, e, a, &b, 68);
+	r4(&block, b, &c, d, e, &a, 69);
+	r4(&block, a, &b, c, d, &e, 70);
+	r4(&block, e, &a, b, c, &d, 71);
+	r4(&block, d, &e, a, b, &c, 72);
+	r4(&block, c, &d, e, a, &b, 73);
+	r4(&block, b, &c, d, e, &a, 74);
+	r4(&block, a, &b, c, d, &e, 75);
+	r4(&block, e, &a, b, c, &d, 76);
+	r4(&block, d, &e, a, b, &c, 77);
+	r4(&block, c, &d, e, a, &b, 78);
+	r4(&block, b, &c, d, e, &a, 79);
+	(*state)[0] += a;
+	(*state)[1] += b;
+	(*state)[2] += c;
+	(*state)[3] += d;
+	(*state)[4] += e;
+	a = b = c = d = e = 0;
+	block = {};
+}

lib/std/hash/sha256.c3

@@ -0,0 +1,176 @@
+module std::hash::sha256;
+
+import std::hash::hmac;
+
+const BLOCK_SIZE = 64;
+const HASH_SIZE = 32;
+
+const uint[64] K @local = {
+    0x428a2f98, 0x71374491, 0xb5c0fbcf, 0xe9b5dba5, 0x3956c25b, 0x59f111f1, 0x923f82a4, 0xab1c5ed5,
+    0xd807aa98, 0x12835b01, 0x243185be, 0x550c7dc3, 0x72be5d74, 0x80deb1fe, 0x9bdc06a7, 0xc19bf174,
+    0xe49b69c1, 0xefbe4786, 0x0fc19dc6, 0x240ca1cc, 0x2de92c6f, 0x4a7484aa, 0x5cb0a9dc, 0x76f988da,
+    0x983e5152, 0xa831c66d, 0xb00327c8, 0xbf597fc7, 0xc6e00bf3, 0xd5a79147, 0x06ca6351, 0x14292967,
+    0x27b70a85, 0x2e1b2138, 0x4d2c6dfc, 0x53380d13, 0x650a7354, 0x766a0abb, 0x81c2c92e, 0x92722c85,
+    0xa2bfe8a1, 0xa81a664b, 0xc24b8b70, 0xc76c51a3, 0xd192e819, 0xd6990624, 0xf40e3585, 0x106aa070,
+    0x19a4c116, 0x1e376c08, 0x2748774c, 0x34b0bcb5, 0x391c0cb3, 0x4ed8aa4a, 0x5b9cca4f, 0x682e6ff3,
+    0x748f82ee, 0x78a5636f, 0x84c87814, 0x8cc70208, 0x90befffa, 0xa4506ceb, 0xbef9a3f7, 0xc67178f2
+};
+
+// Right rotate function
+macro uint @rotr(uint x, uint n) @local => (((x) >> (n)) | ((x) << (32 - (n))));
+
+// SHA-256 functions
+macro uint @ch(uint x, uint y, uint z) @local => (x & y) ^ (~x & z);
+macro uint @maj(uint x, uint y, uint z) @local => (x & y) ^ (x & z) ^ (y & z);
+macro uint @_sigma0(uint x) @local => @rotr(x, 2) ^ @rotr(x, 13) ^ @rotr(x, 22);
+macro uint @_sigma1(uint x) @local => @rotr(x, 6) ^ @rotr(x, 11) ^ @rotr(x, 25);
+macro uint @sigma0(uint x) @local => @rotr(x, 7) ^ @rotr(x, 18) ^ (x >> 3);
+macro uint @sigma1(uint x) @local => @rotr(x, 17) ^ @rotr(x, 19) ^ (x >> 10);
+
+struct Sha256
+{
+    uint[8] state;
+    ulong bitcount;
+    char[BLOCK_SIZE] buffer;
+}
+
+def HmacSha256 = Hmac(<Sha256, HASH_SIZE, BLOCK_SIZE>);
+def hmac = hmac::hash(<Sha256, HASH_SIZE, BLOCK_SIZE>);
+def pbkdf2 = hmac::pbkdf2(<Sha256, HASH_SIZE, BLOCK_SIZE>);
+
+fn char[HASH_SIZE] hash(char[] data)
+{
+    Sha256 sha256 @noinit;
+    sha256.init();
+    sha256.update(data);
+    return sha256.final();
+}
+
+fn void Sha256.init(&self)
+{
+    // Sha256 initialization constants
+    *self = {
+        .state = {
+            0x6A09E667,
+            0xBB67AE85,
+            0x3C6EF372,
+            0xA54FF53A,
+            0x510E527F,
+            0x9B05688C,
+            0x1F83D9AB,
+            0x5BE0CD19
+        }
+    };
+}
+
+<*
+ @param [in] data
+ @require data.len <= uint.max
+*>
+fn void Sha256.update(&self, char[] data) {
+    uint i = 0;
+    uint len = data.len;
+    uint buffer_pos = (uint)(self.bitcount / 8) % BLOCK_SIZE;
+    self.bitcount += (ulong)(len * 8);
+
+    while (len--) {
+        self.buffer[buffer_pos++] = data[i++];
+        if (buffer_pos == BLOCK_SIZE) {
+            sha256_transform(&self.state, &self.buffer);
+            buffer_pos = 0;  // Reset buffer position
+        }
+    }
+}
+
+fn char[HASH_SIZE] Sha256.final(&self) {
+    char[HASH_SIZE] hash;
+    ulong i = (self.bitcount / 8) % BLOCK_SIZE;
+
+    // Append 0x80 to the buffer
+    self.buffer[i++] = 0x80;
+
+    // Pad the buffer with zeros
+    if (i > BLOCK_SIZE - 8) {
+        while (i < BLOCK_SIZE) {
+            self.buffer[i++] = 0x00;
+        }
+        sha256_transform(&self.state, &self.buffer);
+        i = 0;  // Reset buffer index after transformation
+    }
+    
+    while (i < BLOCK_SIZE - 8) {
+        self.buffer[i++] = 0x00;
+    }
+
+    // Append the bitcount in big-endian format
+    for (int j = 0; j < 8; ++j) {
+        self.buffer[BLOCK_SIZE - 8 + j] = (char)((self.bitcount >> (56 - j * 8)) & 0xFF);
+    }
+
+    sha256_transform(&self.state, &self.buffer);
+
+    // Convert state to the final hash
+    for (i = 0; i < 8; ++i) {
+        hash[i * 4] = (char)((self.state[i] >> 24) & 0xFF);
+        hash[i * 4 + 1] = (char)((self.state[i] >> 16) & 0xFF);
+        hash[i * 4 + 2] = (char)((self.state[i] >> 8) & 0xFF);
+        hash[i * 4 + 3] = (char)(self.state[i] & 0xFF);
+    }
+    
+    return hash;
+}
+
+<*
+ @param [&inout] state
+ @param [&in] buffer
+*>
+fn void sha256_transform(uint* state, char* buffer) @local {
+    uint a, b, c, d, e, f, g, h, t1, t2;
+    uint[64] m;
+    int i;
+
+    // Prepare the message schedule
+    for (i = 0; i < 16; ++i) {
+     m[i] = ((uint)buffer[i * 4] << 24) | ((uint)buffer[i * 4 + 1] << 16) |
+       ((uint)buffer[i * 4 + 2] << 8) | ((uint)buffer[i * 4 + 3]);  // Ensure values are cast to uint for correct shifts
+    }
+    for (i = 16; i < 64; ++i) {
+        m[i] = @sigma1(m[i - 2]) + m[i - 7] + @sigma0(m[i - 15]) + m[i - 16];
+    }
+
+    // Initialize working variables
+    a = state[0];
+    b = state[1];
+    c = state[2];
+    d = state[3];
+    e = state[4];
+    f = state[5];
+    g = state[6];
+    h = state[7];
+
+    // Perform the main SHA-256 compression function
+    for (i = 0; i < 64; ++i) {
+        t1 = h + @_sigma1(e) + @ch(e, f, g) + K[i] + m[i];
+        t2 = @_sigma0(a) + @maj(a, b, c);
+        h = g;
+        g = f;
+        f = e;
+        e = d + t1;
+        d = c;
+        c = b;
+        b = a;
+        a = t1 + t2;
+    }
+
+    // Update the state
+    state[0] += a;
+    state[1] += b;
+    state[2] += c;
+    state[3] += d;
+    state[4] += e;
+    state[5] += f;
+    state[6] += g;
+    state[7] += h;
+    a = b = c = d = e = f = g = h = t1 = t2 = i = 0;
+    m[:64] = buffer[:64] = 0;
+}

lib/std/io.c3

@@ -1,520 +0,0 @@
-// Copyright (c) 2021-2022 Christoffer Lerno. All rights reserved.
-// Use of this source code is governed by the MIT license
-// a copy of which can be found in the LICENSE_STDLIB file.
-module std::io;
-import std::mem;
-import libc;
-import std::env;
-
-struct File
-{
-    CFile file;
-}
-
-
-extern fn int _puts(char* message) @extname("puts");
-extern fn int printf(char* message, ...);
-extern fn int _putchar(char c) @extname("putchar");
-
-
-fn int putchar(char c) @inline
-{
-    return _putchar(c);
-}
-
-/**
- * @param [&in] message
- * @return `number of bytes printed.`
- */
-fn int print(char* message)
-{
-    char* pointer = message;
-    while (*pointer != '\0')
-    {
-        if (!putchar(*pointer)) return 0;
-        pointer++;
-    }
-    return 1;
-}
-
-/**
- * @param [&in] message
- * @return `number of bytes printed.`
- */
-fn int println(char *message = "") @inline
-{
-    return _puts(message);
-}
-
-fn void! File.open(File* file, char[] filename, char[] mode)
-{
-    char* filename_copy = mem::talloc(filename.len + 1)!!;
-
-    char* mode_copy = mem::talloc(mode.len + 1)!!;
-
-    mem::copy(filename_copy, (char*)(filename), filename.len);
-    mem::copy(mode_copy, (char*)(mode), mode.len);
-    filename_copy[filename.len] = 0;
-    mode_copy[filename.len] = 0;
-    file.file = libc::fopen(filename_copy, mode_copy);
-    if (!file.file) return IoError.FILE_NOT_FOUND!;
-}
-
-enum Seek
-{
-    SET = 0,
-    CURSOR = 1,
-    END = 2
-}
-
-fault IoError
-{
-    FILE_NOT_FOUND,
-    FILE_NOT_SEEKABLE,
-    FILE_NOT_VALID,
-    FILE_INVALID_POSITION,
-    FILE_OVERFLOW,
-    FILE_IS_PIPE,
-    FILE_EOF,
-    FILE_INCOMPLETE_WRITE,
-    INTERRUPTED,
-    UNKNOWN_ERROR,
-}
-
-/**
- * @require file.file != null
- **/
-fn void! File.seek(File *file, long offset, Seek seekMode = Seek.SET)
-{
-    if (libc::fseek(file.file, (SeekIndex)(offset), (int)(seekMode)))
-    {
-        switch (libc::errno())
-        {
-            case EBADF: return IoError.FILE_NOT_SEEKABLE!;
-            case EINVAL: return IoError.FILE_INVALID_POSITION!;
-            case EOVERFLOW: return IoError.FILE_OVERFLOW!;
-            case ESPIPE: return IoError.FILE_IS_PIPE!;
-            default: return IoError.UNKNOWN_ERROR!;
-        }
-    }
-}
-
-/**
- * @require file && file.file != null
- */
-fn void! File.putc(File *file, char c)
-{
-    if (!libc::fputc(c, file.file)) return IoError.FILE_EOF!;
-}
-
-/**
- * @require file != null
- */
-fn void! File.close(File *file) @inline
-{
-    if (file.file && libc::fclose(file.file))
-    {
-        switch (libc::errno())
-        {
-            case ECONNRESET:
-            case EBADF: return IoError.FILE_NOT_VALID!;
-            case EINTR: return IoError.INTERRUPTED!;
-            case EDQUOT:
-            case EFAULT:
-            case EAGAIN:
-            case EFBIG:
-            case ENETDOWN:
-            case ENETUNREACH:
-            case ENOSPC:
-            case EIO: return IoError.FILE_INCOMPLETE_WRITE!;
-            default: return IoError.UNKNOWN_ERROR!;
-        }
-    }
-    file.file = null;
-}
-
-/**
- * @require file && file.file
- */
-fn bool File.eof(File* file) @inline
-{
-    return libc::feof(file.file) != 0;
-}
-
-/**
- * @require file && file.file
- */
-fn usize File.read(File* file, void* buffer, usize items, usize element_size = 1)
-{
-    return libc::fread(buffer, element_size, items, file.file);
-}
-
-/**
- * @param [&in] file
- * @param [&out] buffer
- * @param items
- * @param element_size
- * @require file.file `File must be initialized`
- * @require element_size > 1
- */
-fn usize File.write(File* file, void* buffer, usize items, usize element_size = 1)
-{
-    return libc::fwrite(buffer, element_size, items, file.file);
-}
-
-/**
- * @param [&in] file
- * @require file.file `File must be initialized`
- */
-fn usize! File.println(File* file, char[] string)
-{
-    usize len = string.len;
-    if (len != libc::fwrite(string.ptr, 1, len, file.file)) return IoError.UNKNOWN_ERROR!;
-    if (!libc::putc('\n', file.file)) return IoError.UNKNOWN_ERROR!;
-    return len + 1;
-}
-
-/*
-
-
-error FileError
-{
-    ulong errno;
-}
-
-fn FileError errorFromErrno()
-{
-    return FileError { };
-}
-
-
-
-
-pubic fn void! File.clearerr(File *file) @inline
-{
-    clearerr(file->file);
-}
-
-
-
-fn void File.error(File *file) @inline
-{
-    int err = ferror
-}
-*/
-/*
-
-#define	__SLBF	0x0001		/* line buffered */
-#define	__SNBF	0x0002		/* unbuffered */
-#define	__SRD	0x0004		/* OK to read */
-#define	__SWR	0x0008		/* OK to write */
-	/* RD and WR are never simultaneously asserted */
-#define	__SRW	0x0010		/* open for reading & writing */
-#define	__SEOF	0x0020		/* found EOF */
-#define	__SERR	0x0040		/* found error */
-#define	__SMBF	0x0080		/* _buf is from malloc */
-#define	__SAPP	0x0100		/* fdopen()ed in append mode */
-#define	__SSTR	0x0200		/* this is an sprintf/snprintf string */
-#define	__SOPT	0x0400		/* do fseek() optimisation */
-#define	__SNPT	0x0800		/* do not do fseek() optimisation */
-#define	__SOFF	0x1000		/* set iff _offset is in fact correct */
-#define	__SMOD	0x2000		/* true => fgetln modified _p text */
-#define __SALC  0x4000		/* allocate string space dynamically */
-#define __SIGN  0x8000		/* ignore this file in _fwalk */
-
-/*
- * The following three definitions are for ANSI C, which took them
- * from System V, which brilliantly took internal interface macros and
- * made them official arguments to setvbuf(), without renaming them.
- * Hence, these ugly _IOxxx names are *supposed* to appear in user code.
- *
- * Although numbered as their counterparts above, the implementation
- * does not rely on this.
- */
-#define	_IOFBF	0		/* setvbuf should set fully buffered */
-#define	_IOLBF	1		/* setvbuf should set line buffered */
-#define	_IONBF	2		/* setvbuf should set unbuffered */
-
-#define	BUFSIZ	1024		/* size of buffer used by setbuf */
-#define	EOF	(-1)
-
-				/* must be == _POSIX_STREAM_MAX <limits.h> */
-#define	FOPEN_MAX	20	/* must be <= OPEN_MAX <sys/syslimits.h> */
-#define	FILENAME_MAX	1024	/* must be <= PATH_MAX <sys/syslimits.h> */
-
-/* System V/ANSI C; this is the wrong way to do this, do *not* use these. */
-#ifndef _ANSI_SOURCE
-#define	P_tmpdir	"/var/tmp/"
-#endif
-#define	L_tmpnam	1024	/* XXX must be == PATH_MAX */
-#define	TMP_MAX		308915776
-
-
-#define	stdin	__stdinp
-#define	stdout	__stdoutp
-#define	stderr	__stderrp
-
-
-/* ANSI-C */
-
-__BEGIN_DECLS
-char	*fgets(char * __restrict, int, FILE *);
-#if defined(_DARWIN_UNLIMITED_STREAMS) || defined(_DARWIN_C_SOURCE)
-#else /* !_DARWIN_UNLIMITED_STREAMS && !_DARWIN_C_SOURCE */
-FILE	*fopen(const char * __restrict __filename, const char * __restrict __mode) __DARWIN_ALIAS_STARTING(__MAC_10_6, __IPHONE_2_0, __DARWIN_ALIAS(fopen));
-#endif /* (DARWIN_UNLIMITED_STREAMS || _DARWIN_C_SOURCE) */
-int	 fprintf(FILE * __restrict, const char * __restrict, ...) __printflike(2, 3);
-int	 fputs(const char * __restrict, FILE * __restrict) __DARWIN_ALIAS(fputs);
-size_t	 fread(void * __restrict __ptr, size_t __size, size_t __nitems, FILE * __restrict __stream);
-FILE	*freopen(const char * __restrict, const char * __restrict,
-                 FILE * __restrict) __DARWIN_ALIAS(freopen);
-int	 fscanf(FILE * __restrict, const char * __restrict, ...) __scanflike(2, 3);
-int	 fsetpos(FILE *, const fpos_t *);
-long	 ftell(FILE *);
-size_t	 fwrite(const void * __restrict __ptr, size_t __size, size_t __nitems, FILE * __restrict __stream) __DARWIN_ALIAS(fwrite);
-int	 getc(FILE *);
-int	 getchar(void);
-char	*gets(char *);
-void	 perror(const char *) __cold;
-int	 printf(const char * __restrict, ...) __printflike(1, 2);
-int	 putc(int, FILE *);
-int	 putchar(int);
-int	 puts(const char *);
-int	 remove(const char *);
-int	 rename (const char *__old, const char *__new);
-void	 rewind(FILE *);
-int	 scanf(const char * __restrict, ...) __scanflike(1, 2);
-void	 setbuf(FILE * __restrict, char * __restrict);
-int	 setvbuf(FILE * __restrict, char * __restrict, int, size_t);
-int	 sprintf(char * __restrict, const char * __restrict, ...) __printflike(2, 3) __swift_unavailable("Use snprintf instead.");
-int	 sscanf(const char * __restrict, const char * __restrict, ...) __scanflike(2, 3);
-FILE	*tmpfile(void);
-
-__swift_unavailable("Use mkstemp(3) instead.")
-#if !defined(_POSIX_C_SOURCE)
-__deprecated_msg("This function is provided for compatibility reasons only.  Due to security concerns inherent in the design of tmpnam(3), it is highly recommended that you use mkstemp(3) instead.")
-#endif
-char	*tmpnam(char *);
-int	 ungetc(int, FILE *);
-int	 vfprintf(FILE * __restrict, const char * __restrict, va_list) __printflike(2, 0);
-int	 vprintf(const char * __restrict, va_list) __printflike(1, 0);
-int	 vsprintf(char * __restrict, const char * __restrict, va_list) __printflike(2, 0) __swift_unavailable("Use vsnprintf instead.");
-__END_DECLS
-
-
-
-/* Additional functionality provided by:
- * POSIX.1-1988
- */
-
-#if __DARWIN_C_LEVEL >= 198808L
-#define	L_ctermid	1024	/* size for ctermid(); PATH_MAX */
-
-__BEGIN_DECLS
-#include <_ctermid.h>
-
-#if defined(_DARWIN_UNLIMITED_STREAMS) || defined(_DARWIN_C_SOURCE)
-FILE	*fdopen(int, const char *) __DARWIN_ALIAS_STARTING(__MAC_10_6, __IPHONE_3_2, __DARWIN_EXTSN(fdopen));
-#else /* !_DARWIN_UNLIMITED_STREAMS && !_DARWIN_C_SOURCE */
-FILE	*fdopen(int, const char *) __DARWIN_ALIAS_STARTING(__MAC_10_6, __IPHONE_2_0, __DARWIN_ALIAS(fdopen));
-#endif /* (DARWIN_UNLIMITED_STREAMS || _DARWIN_C_SOURCE) */
-int	 fileno(FILE *);
-__END_DECLS
-#endif /* __DARWIN_C_LEVEL >= 198808L */
-
-
-/* Additional functionality provided by:
- * POSIX.2-1992 C Language Binding Option
- */
-#if TARGET_OS_EMBEDDED
-#define __swift_unavailable_on(osx_msg, ios_msg) __swift_unavailable(ios_msg)
-#else
-#define __swift_unavailable_on(osx_msg, ios_msg) __swift_unavailable(osx_msg)
-#endif
-
-#if __DARWIN_C_LEVEL >= 199209L
-__BEGIN_DECLS
-int	 pclose(FILE *) __swift_unavailable_on("Use posix_spawn APIs or NSTask instead.", "Process spawning is unavailable.");
-#if defined(_DARWIN_UNLIMITED_STREAMS) || defined(_DARWIN_C_SOURCE)
-FILE	*popen(const char *, const char *) __DARWIN_ALIAS_STARTING(__MAC_10_6, __IPHONE_3_2, __DARWIN_EXTSN(popen)) __swift_unavailable_on("Use posix_spawn APIs or NSTask instead.", "Process spawning is unavailable.");
-#else /* !_DARWIN_UNLIMITED_STREAMS && !_DARWIN_C_SOURCE */
-FILE	*popen(const char *, const char *) __DARWIN_ALIAS_STARTING(__MAC_10_6, __IPHONE_2_0, __DARWIN_ALIAS(popen)) __swift_unavailable_on("Use posix_spawn APIs or NSTask instead.", "Process spawning is unavailable.");
-#endif /* (DARWIN_UNLIMITED_STREAMS || _DARWIN_C_SOURCE) */
-__END_DECLS
-#endif /* __DARWIN_C_LEVEL >= 199209L */
-
-#undef __swift_unavailable_on
-
-/* Additional functionality provided by:
- * POSIX.1c-1995,
- * POSIX.1i-1995,
- * and the omnibus ISO/IEC 9945-1: 1996
- */
-
-#if __DARWIN_C_LEVEL >= 199506L
-
-/* Functions internal to the implementation. */
-__BEGIN_DECLS
-int	__srget(FILE *);
-int	__svfscanf(FILE *, const char *, va_list) __scanflike(2, 0);
-int	__swbuf(int, FILE *);
-__END_DECLS
-
-/*
- * The __sfoo macros are here so that we can
- * define function versions in the C library.
- */
-#define	__sgetc(p) (--(p)->_r < 0 ? __srget(p) : (int)(*(p)->_p++))
-#if defined(__GNUC__) && defined(__STDC__)
-__header_always_inline int __sputc(int _c, FILE *_p) {
-	if (--_p->_w >= 0 || (_p->_w >= _p->_lbfsize && (char)_c != '\n'))
-		return (*_p->_p++ = _c);
-	else
-		return (__swbuf(_c, _p));
-}
-#else
-/*
- * This has been tuned to generate reasonable code on the vax using pcc.
- */
-#define	__sputc(c, p) \
-	(--(p)->_w < 0 ? \
-		(p)->_w >= (p)->_lbfsize ? \
-			(*(p)->_p = (c)), *(p)->_p != '\n' ? \
-				(int)*(p)->_p++ : \
-				__swbuf('\n', p) : \
-			__swbuf((int)(c), p) : \
-		(*(p)->_p = (c), (int)*(p)->_p++))
-#endif
-
-#define	__sfeof(p)	(((p)->_flags & __SEOF) != 0)
-#define	__sferror(p)	(((p)->_flags & __SERR) != 0)
-#define	__sclearerr(p)	((void)((p)->_flags &= ~(__SERR|__SEOF)))
-#define	__sfileno(p)	((p)->_file)
-
-__BEGIN_DECLS
-void	 flockfile(FILE *);
-int	 ftrylockfile(FILE *);
-void	 funlockfile(FILE *);
-int	 getc_unlocked(FILE *);
-int	 getchar_unlocked(void);
-int	 putc_unlocked(int, FILE *);
-int	 putchar_unlocked(int);
-
-/* Removed in Issue 6 */
-#if !defined(_POSIX_C_SOURCE) || _POSIX_C_SOURCE < 200112L
-int	 getw(FILE *);
-int	 putw(int, FILE *);
-#endif
-
-__swift_unavailable("Use mkstemp(3) instead.")
-#if !defined(_POSIX_C_SOURCE)
-__deprecated_msg("This function is provided for compatibility reasons only.  Due to security concerns inherent in the design of tempnam(3), it is highly recommended that you use mkstemp(3) instead.")
-#endif
-char	*tempnam(const char *__dir, const char *__prefix) __DARWIN_ALIAS(tempnam);
-__END_DECLS
-
-#ifndef lint
-#define	getc_unlocked(fp)	__sgetc(fp)
-#define putc_unlocked(x, fp)	__sputc(x, fp)
-#endif /* lint */
-
-#define	getchar_unlocked()	getc_unlocked(stdin)
-#define	putchar_unlocked(x)	putc_unlocked(x, stdout)
-#endif /* __DARWIN_C_LEVEL >= 199506L */
-
-
-
-/* Additional functionality provided by:
- * POSIX.1-2001
- * ISO C99
- */
-
-#if __DARWIN_C_LEVEL >= 200112L
-#include <sys/_types/_off_t.h>
-
-__BEGIN_DECLS
-int	 fseeko(FILE * __stream, off_t __offset, int __whence);
-off_t	 ftello(FILE * __stream);
-__END_DECLS
-#endif /* __DARWIN_C_LEVEL >= 200112L */
-
-#if __DARWIN_C_LEVEL >= 200112L || defined(_C99_SOURCE) || defined(__cplusplus)
-__BEGIN_DECLS
-int	 snprintf(char * __restrict __str, size_t __size, const char * __restrict __format, ...) __printflike(3, 4);
-int	 vfscanf(FILE * __restrict __stream, const char * __restrict __format, va_list) __scanflike(2, 0);
-int	 vscanf(const char * __restrict __format, va_list) __scanflike(1, 0);
-int	 vsnprintf(char * __restrict __str, size_t __size, const char * __restrict __format, va_list) __printflike(3, 0);
-int	 vsscanf(const char * __restrict __str, const char * __restrict __format, va_list) __scanflike(2, 0);
-__END_DECLS
-#endif /* __DARWIN_C_LEVEL >= 200112L || defined(_C99_SOURCE) || defined(__cplusplus) */
-
-
-
-/* Additional functionality provided by:
- * POSIX.1-2008
- */
-
-#if __DARWIN_C_LEVEL >= 200809L
-#include <sys/_types/_ssize_t.h>
-
-__BEGIN_DECLS
-int	dprintf(int, const char * __restrict, ...) __printflike(2, 3) __OSX_AVAILABLE_STARTING(__MAC_10_7, __IPHONE_4_3);
-int	vdprintf(int, const char * __restrict, va_list) __printflike(2, 0) __OSX_AVAILABLE_STARTING(__MAC_10_7, __IPHONE_4_3);
-ssize_t getdelim(char ** __restrict __linep, size_t * __restrict __linecapp, int __delimiter, FILE * __restrict __stream) __OSX_AVAILABLE_STARTING(__MAC_10_7, __IPHONE_4_3);
-ssize_t getline(char ** __restrict __linep, size_t * __restrict __linecapp, FILE * __restrict __stream) __OSX_AVAILABLE_STARTING(__MAC_10_7, __IPHONE_4_3);
-FILE *fmemopen(void * __restrict __buf, size_t __size, const char * __restrict __mode) __API_AVAILABLE(macos(10.13), ios(11.0), tvos(11.0), watchos(4.0));
-FILE *open_memstream(char **__bufp, size_t *__sizep) __API_AVAILABLE(macos(10.13), ios(11.0), tvos(11.0), watchos(4.0));
-__END_DECLS
-#endif /* __DARWIN_C_LEVEL >= 200809L */
-
-
-
-/* Darwin extensions */
-
-#if __DARWIN_C_LEVEL >= __DARWIN_C_FULL
-__BEGIN_DECLS
-extern __const int sys_nerr;		/* perror(3) external variables */
-extern __const char *__const sys_errlist[];
-
-int	 asprintf(char ** __restrict, const char * __restrict, ...) __printflike(2, 3);
-char	*ctermid_r(char *);
-char	*fgetln(FILE *, size_t *);
-__const char *fmtcheck(const char *, const char *);
-int	 fpurge(FILE *);
-void	 setbuffer(FILE *, char *, int);
-int	 setlinebuf(FILE *);
-int	 vasprintf(char ** __restrict, const char * __restrict, va_list) __printflike(2, 0);
-FILE	*zopen(const char *, const char *, int);
-
-
-/*
- * Stdio function-access interface.
- */
-FILE	*funopen(const void *,
-                 int (* _Nullable)(void *, char *, int),
-                 int (* _Nullable)(void *, const char *, int),
-                 fpos_t (* _Nullable)(void *, fpos_t, int),
-                 int (* _Nullable)(void *));
-__END_DECLS
-#define	fropen(cookie, fn) funopen(cookie, fn, 0, 0, 0)
-#define	fwopen(cookie, fn) funopen(cookie, 0, fn, 0, 0)
-
-#define	feof_unlocked(p)	__sfeof(p)
-#define	ferror_unlocked(p)	__sferror(p)
-#define	clearerr_unlocked(p)	__sclearerr(p)
-#define	fileno_unlocked(p)	__sfileno(p)
-
-#endif /* __DARWIN_C_LEVEL >= __DARWIN_C_FULL */
-
-
-#ifdef _USE_EXTENDED_LOCALES_
-#include <xlocale/_stdio.h>
-#endif /* _USE_EXTENDED_LOCALES_ */
-
-#if defined (__GNUC__) && _FORTIFY_SOURCE > 0 && !defined (__cplusplus)
-/* Security checking functions.  */
-#include <secure/_stdio.h>
-#endif
-
-#endif /* _STDIO_H_ */
-*/

lib/std/io/bits.c3

@@ -0,0 +1,94 @@
+module std::io;
+
+struct BitReader
+{
+	InStream reader;
+	uint bits;
+	uint len;
+}
+
+fn void BitReader.init(&self, InStream byte_reader)
+{
+	*self = { .reader = byte_reader };
+}
+
+fn void BitReader.clear(&self) @inline
+{
+	self.len = 0;
+}
+
+<*
+ @require nbits <= 8
+ @require self.len + nbits <= uint.sizeof * 8
+*>
+fn char! BitReader.read_bits(&self, uint nbits)
+{
+	uint bits = self.bits;
+	if (self.len < nbits)
+	{
+		// New bits are pushed right.
+		char c = self.reader.read_byte()!;
+		bits <<= 8;
+		bits |= c;
+		self.bits = bits;
+		self.len += 8;
+	}
+	self.len -= nbits;
+	uint mask = (1 << nbits) - 1;
+	return (char)((bits >> self.len) & mask);
+}
+
+struct BitWriter
+{
+	OutStream writer;
+	uint bits;
+	uint len;
+}
+
+// c3 doesn't allow to shift more than bit width of a variable,
+// so use closest byte boundary of 24 instead of 32
+const int WRITER_BITS = 24;
+
+fn void BitWriter.init(&self, OutStream byte_writer)
+{
+	*self = { .writer = byte_writer };
+}
+
+fn void! BitWriter.flush(&self)
+{
+	if (self.len == 0) return;
+
+	int padding = ($sizeof(self.bits) * 8 - self.len);
+	uint bits = self.bits << padding;
+	uint n = (self.len + 7) / 8;
+	char[4] buffer;
+	bitorder::write(bits, &buffer, UIntBE);
+	io::write_all(self.writer, buffer[:n])!;
+	self.len = 0;
+}
+
+<*
+ @require nbits <= 32
+*>
+fn void! BitWriter.write_bits(&self, uint bits, uint nbits)
+{
+	if (nbits == 0) return;
+	while (self.len + nbits > WRITER_BITS)
+	{
+		uint to_push = WRITER_BITS - self.len;
+		uint bits_to_push = (bits >> (nbits - to_push)) & ((1 << to_push) - 1);
+
+		self.bits <<= to_push;
+		self.bits |= bits_to_push;
+		self.len += to_push;
+		nbits -= to_push;
+
+		self.flush()!;
+	}
+
+	if (nbits == 0) return;
+
+	self.bits <<= nbits;
+	self.bits |= bits & ((1 << nbits) - 1);
+	self.len += nbits;
+}

lib/std/io/file.c3

@@ -0,0 +1,222 @@
+module std::io;
+import libc;
+
+struct File (InStream, OutStream)
+{
+	CFile file;
+}
+
+module std::io::file;
+import libc, std::io::path, std::io::os;
+
+fn File! open(String filename, String mode)
+{
+	return from_handle(os::native_fopen(filename, mode));
+}
+
+fn File! open_path(Path path, String mode)
+{
+	return from_handle(os::native_fopen(path.str_view(), mode));
+}
+
+fn bool exists(String file) => @pool()
+{
+	return path::exists(path::temp_new(file)) ?? false;
+}
+
+fn File from_handle(CFile file)
+{
+	return { .file = file };
+}
+
+fn bool is_file(String path)
+{
+	return os::native_is_file(path);
+}
+
+fn usz! get_size(String path)
+{
+	return os::native_file_size(path);
+}
+
+fn void! delete(String filename) => os::native_remove(filename) @inline;
+
+
+<*
+ @require self.file != null
+*>
+fn void! File.reopen(&self, String filename, String mode)
+{
+	self.file = os::native_freopen(self.file, filename, mode)!;
+}
+
+<*
+ @require self.file != null
+*>
+fn usz! File.seek(&self, isz offset, Seek seek_mode = Seek.SET) @dynamic
+{
+	os::native_fseek(self.file, offset, seek_mode)!;
+	return os::native_ftell(self.file);
+}
+
+
+/*
+Implement later
+<*
+ @require self.file == null
+*>
+fn void! File.memopen(File* file, char[] data, String mode)
+{
+	@pool()
+	{
+		file.file = libc::memopen(data.ptr, data.len, mode.zstr_tcopy(), file.file);
+		// TODO errors
+	};
+}
+*/
+
+
+<*
+ @require self.file != null
+*>
+fn void! File.write_byte(&self, char c) @dynamic
+{
+	return os::native_fputc(c, self.file);
+}
+
+<*
+ @param [&inout] self
+*>
+fn void! File.close(&self) @inline @dynamic
+{
+	if (self.file && libc::fclose(self.file))
+	{
+		switch (libc::errno())
+		{
+			case errno::ECONNRESET:
+			case errno::EBADF: return IoError.FILE_NOT_VALID?;
+			case errno::EINTR: return IoError.INTERRUPTED?;
+			case errno::EDQUOT:
+			case errno::EFAULT:
+			case errno::EAGAIN:
+			case errno::EFBIG:
+			case errno::ENETDOWN:
+			case errno::ENETUNREACH:
+			case errno::ENOSPC:
+			case errno::EIO: return IoError.INCOMPLETE_WRITE?;
+			default: return IoError.UNKNOWN_ERROR?;
+		}
+	}
+	self.file = null;
+}
+
+<*
+ @require self.file != null
+*>
+fn bool File.eof(&self) @inline
+{
+	return libc::feof(self.file) != 0;
+}
+
+<*
+ @param [in] buffer
+*>
+fn usz! File.read(&self, char[] buffer) @dynamic
+{
+	return os::native_fread(self.file, buffer);
+}
+
+<*
+ @param [out] buffer
+ @require self.file != null `File must be initialized`
+*>
+fn usz! File.write(&self, char[] buffer) @dynamic
+{
+	return os::native_fwrite(self.file, buffer);
+}
+
+fn Fd File.fd(self) @if(env::LIBC)
+{
+	return libc::fileno(self.file);
+}
+
+fn bool File.isatty(self) @if(env::LIBC)
+{
+	return libc::isatty(self.fd()) > 0;
+}
+
+fn char! File.read_byte(&self) @dynamic
+{
+	int c = libc::fgetc(self.file);
+	if (c == -1) return IoError.EOF?;
+	return (char)c;
+}
+
+<*
+ Load up to buffer.len characters. Returns IoError.OVERFLOW if the file is longer
+ than the buffer.
+
+ @param filename "The path to the file to read"
+ @param [in] buffer "The buffer to read to"
+*>
+fn char[]! load_buffer(String filename, char[] buffer)
+{
+	File file = open(filename, "rb")!;
+	defer (void)file.close();
+	usz len = file.seek(0, END)!;
+	if (len > buffer.len) return IoError.OVERFLOW?;
+	file.seek(0, SET)!;
+	usz read = 0;
+	while (read < len)
+	{
+		read += file.read(buffer[read:len - read])!;
+	}
+	return buffer[:len];
+
+}
+
+fn char[]! load(Allocator allocator, String filename) => load_new(filename, allocator);
+
+fn char[]! load_new(String filename, Allocator allocator = allocator::heap())
+{
+	File file = open(filename, "rb")!;
+	defer (void)file.close();
+	usz len = file.seek(0, END)!;
+	file.seek(0, SET)!;
+	char* data = allocator::malloc_try(allocator, len)!;
+	defer catch allocator::free(allocator, data);
+	usz read = 0;
+	while (read < len)
+	{
+		read += file.read(data[read:len - read])!;
+	}
+	return data[:len];
+}
+
+fn char[]! load_path_new(Path path, Allocator allocator = allocator::heap()) => load_new(path.str_view(), allocator);
+
+fn char[]! load_temp(String filename)
+{
+	return load_new(filename, allocator::temp());
+}
+
+fn char[]! load_path_temp(Path path) => load_temp(path.str_view());
+
+fn void! save(String filename, char[] data)
+{
+	File file = open(filename, "wb")!;
+	defer (void)file.close();
+	while (data.len)
+	{
+		usz written = file.write(data)!;
+		data = data[written..];
+	}
+}
+
+<*
+ @require self.file != null `File must be initialized`
+*>
+fn void! File.flush(&self) @dynamic
+{
+	libc::fflush(self.file);
+}

lib/std/io/formatter.c3

@@ -0,0 +1,571 @@
+module std::io;
+import std::collections::map;
+import libc;
+
+const int PRINTF_NTOA_BUFFER_SIZE = 256;
+
+interface Printable
+{
+	fn String to_string(Allocator allocator) @optional;
+	fn String to_new_string(Allocator allocator) @optional @deprecated("Use to_string");
+	fn usz! to_format(Formatter* formatter) @optional;
+}
+
+fault PrintFault
+{
+	BUFFER_EXCEEDED,
+	INTERNAL_BUFFER_EXCEEDED,
+	INVALID_FORMAT,
+	NOT_ENOUGH_ARGUMENTS,
+	INVALID_ARGUMENT,
+}
+
+def OutputFn = fn void!(void* buffer, char c);
+def FloatType = double;
+
+
+macro bool is_struct_with_default_print($Type)
+{
+	return $Type.kindof == STRUCT
+		&&& !$defined($Type.to_format)
+		&&& !$defined($Type.to_new_string)
+		&&& !$defined($Type.to_string);
+}
+
+<*
+ Introspect a struct and print it to a formatter
+
+ @require @typekind(value) == STRUCT `This macro is only valid on macros`
+*>
+macro usz! struct_to_format(value, Formatter* f, bool $force_dump)
+{
+    var $Type = $typeof(value);
+    usz total = f.print("{ ")!;
+    $foreach ($i, $member : $Type.membersof)
+        $if $i > 0:
+            total += f.print(", ")!;
+        $endif
+        $if $member.nameof != "":
+            total += f.printf("%s: ", $member.nameof)!;
+		$endif
+		$if ($force_dump &&& $member.typeid.kindof == STRUCT) |||
+			is_struct_with_default_print($typefrom($member.typeid)):
+			total += struct_to_format($member.get(value), f, $force_dump)!;
+		$else
+			total += f.printf("%s", $member.get(value))!;
+		$endif
+	$endforeach
+	return total + f.print(" }");
+}
+
+fn usz! ReflectedParam.to_format(&self, Formatter* f) @dynamic
+{
+	return f.printf("[Parameter '%s']", self.name);
+}
+
+fn usz! Formatter.printf(&self, String format, args...)
+{
+	return self.vprintf(format, args) @inline;
+}
+
+struct Formatter
+{
+	void *data;
+	OutputFn out_fn;
+	struct
+	{
+		PrintFlags flags;
+		uint width;
+		uint prec;
+		usz idx;
+		anyfault first_fault;
+	}
+}
+
+bitstruct PrintFlags : uint
+{
+	bool zeropad;
+	bool left;
+	bool plus;
+	bool space;
+	bool hash;
+	bool uppercase;
+	bool precision;
+}
+
+fn void Formatter.init(&self, OutputFn out_fn, void* data = null)
+{
+	*self = { .data = data, .out_fn = out_fn};
+}
+
+fn usz! Formatter.out(&self, char c) @private
+{
+	if (catch err = self.out_fn(self.data, c))
+	{
+		if (self.first_fault) return self.first_fault?;
+		self.first_fault = err;
+		return err?;
+	}
+	return 1;
+}
+
+fn usz! Formatter.print_with_function(&self, Printable arg)
+{
+	if (&arg.to_format)
+	{
+		PrintFlags old = self.flags;
+		uint old_width = self.width;
+		uint old_prec = self.prec;
+		defer
+		{
+			self.flags = old;
+			self.width = old_width;
+			self.prec = old_prec;
+		}
+		if (!arg) return self.out_substr("(null)");
+		return arg.to_format(self);
+	}
+	if (&arg.to_string)
+	{
+		PrintFlags old = self.flags;
+		uint old_width = self.width;
+		uint old_prec = self.prec;
+		defer
+		{
+			self.flags = old;
+			self.width = old_width;
+			self.prec = old_prec;
+		}
+		if (!arg) return self.out_substr("(null)");
+		@stack_mem(1024; Allocator mem)
+		{
+			return self.out_substr(arg.to_string(mem));
+		};
+	}
+	return SearchResult.MISSING?;
+}
+
+fn usz! Formatter.out_unknown(&self, String category, any arg) @private
+{
+	return self.out_substr("[") + self.out_substr(category) + self.out_substr(" type:") + self.ntoa((iptr)arg.type, false, 16) + self.out_substr(", addr:") + self.ntoa((iptr)arg.ptr, false, 16) + self.out_substr("]");
+}
+fn usz! Formatter.out_str(&self, any arg) @private
+{
+	switch (arg.type.kindof)
+	{
+		case TYPEID:
+			return self.out_substr("typeid");
+		case VOID:
+			return self.out_substr("void");
+		case ANYFAULT:
+		case FAULT:
+			return self.out_substr((*(anyfault*)arg.ptr).nameof);
+		case INTERFACE:
+		case ANY:
+			return self.out_str(*(any*)arg);
+		case OPTIONAL:
+			unreachable();
+		case SIGNED_INT:
+		case UNSIGNED_INT:
+			PrintFlags flags = self.flags;
+			uint width = self.width;
+			defer
+			{
+				self.flags = flags;
+				self.width = width;
+			}
+			self.flags = {};
+			self.width = 0;
+			return self.ntoa_any(arg, 10) ?? self.out_substr("<INVALID>");
+		case FLOAT:
+			PrintFlags flags = self.flags;
+			uint width = self.width;
+			defer
+			{
+				self.flags = flags;
+				self.width = width;
+			}
+			self.flags = {};
+			self.width = 0;
+			return self.ftoa(float_from_any(arg)) ?? self.out_substr("ERR");
+		case BOOL:
+			return self.out_substr(*(bool*)arg.ptr ? "true" : "false");
+		default:
+	}
+	usz! n = self.print_with_function((Printable)arg);
+	if (try n) return n;
+	if (@catch(n) != SearchResult.MISSING) n!;
+	switch (arg.type.kindof)
+	{
+		case ENUM:
+			usz i = types::any_to_enum_ordinal(arg, usz)!!;
+			assert(i < arg.type.names.len, "Illegal enum value found, numerical value was %d.", i);
+			return self.out_substr(arg.type.names[i]);
+		case STRUCT:
+			return self.out_unknown("struct", arg);
+		case UNION:
+			return self.out_unknown("union", arg);
+		case BITSTRUCT:
+			return self.out_unknown("bitstruct", arg);
+		case FUNC:
+			return self.out_unknown("function", arg);
+		case DISTINCT:
+			if (arg.type == String.typeid)
+			{
+				return self.out_substr(*(String*)arg);
+			}
+			if (arg.type == ZString.typeid)
+			{
+				return self.out_substr(*(ZString*)arg ? ((ZString*)arg).str_view() : "(null)");
+			}
+			if (arg.type == DString.typeid)
+			{
+				return self.out_substr(*(DString*)arg ? ((DString*)arg).str_view() : "(null)");
+			}
+			return self.out_str(arg.as_inner());
+		case POINTER:
+			typeid inner = arg.type.inner;
+			void** pointer = arg.ptr;
+			if (arg.type.inner != void.typeid)
+			{
+				any deref = any_make(*pointer, inner);
+				n = self.print_with_function((Printable)deref);
+				if (try n) return n;
+				if (@catch(n) != SearchResult.MISSING) n!;
+			}
+			PrintFlags flags = self.flags;
+			uint width = self.width;
+			defer
+			{
+				self.flags = flags;
+				self.width = width;
+			}
+			self.width = 0;
+			return self.out_substr("0x")! + self.ntoa_any(arg, 16);
+		case ARRAY:
+			// this is SomeType[?] so grab the "SomeType"
+			PrintFlags flags = self.flags;
+			uint width = self.width;
+			defer
+			{
+				self.flags = flags;
+				self.width = width;
+			}
+			self.flags = {};
+			self.width = 0;
+			typeid inner = arg.type.inner;
+			usz size = inner.sizeof;
+			usz alen = arg.type.len;
+			// Pretend this is a String
+			void* ptr = (void*)arg.ptr;
+			usz len = self.out('[')!;
+			for (usz i = 0; i < alen; i++)
+			{
+				if (i != 0) len += self.out_substr(", ")!;
+				len += self.out_str(any_make(ptr, inner))!;
+				ptr += size;
+			}
+			len += self.out(']')!;
+			return len;
+		case VECTOR:
+			PrintFlags flags = self.flags;
+			uint width = self.width;
+			defer
+			{
+				self.flags = flags;
+				self.width = width;
+			}
+			self.flags = {};
+			self.width = 0;
+			// this is SomeType[?] so grab the "SomeType"
+			typeid inner = arg.type.inner;
+			usz size = inner.sizeof;
+			usz vlen = arg.type.len;
+			// Pretend this is a String
+			void* ptr = (void*)arg.ptr;
+			usz len = self.out_substr("[<")!;
+			for (usz i = 0; i < vlen; i++)
+			{
+				if (i != 0) len += self.out_substr(", ")!;
+				len += self.out_str(any_make(ptr, inner))!;
+				ptr += size;
+			}
+			len += self.out_substr(">]")!;
+			return len;
+		case SLICE:
+			// this is SomeType[] so grab the "SomeType"
+			typeid inner = arg.type.inner;
+			if (inner == void.typeid) inner = char.typeid;
+			PrintFlags flags = self.flags;
+			uint width = self.width;
+			defer
+			{
+				self.flags = flags;
+				self.width = width;
+			}
+			self.flags = {};
+			self.width = 0;
+			usz size = inner.sizeof;
+			// Pretend this is a String
+			String* temp = (void*)arg.ptr;
+			void* ptr = (void*)temp.ptr;
+			usz slen = temp.len;
+			usz len = self.out('[')!;
+			for (usz i = 0; i < slen; i++)
+			{
+				if (i != 0) len += self.out_substr(", ")!;
+				len += self.out_str(any_make(ptr, inner))!;
+				ptr += size;
+			}
+			len += self.out(']')!;
+			return len;
+		case ANY:
+		case INTERFACE:
+			unreachable("Already handled");
+		default:
+	}
+	return self.out_substr("Invalid type");
+}
+
+
+
+
+fn void! out_null_fn(void* data @unused, char c @unused) @private
+{
+}
+
+macro usz! @report_fault(Formatter* f, $fault)
+{
+	(void)f.out_substr($fault);
+	return PrintFault.INVALID_FORMAT?;
+}
+
+macro usz! @wrap_bad(Formatter* f, #action)
+{
+	usz! len = #action;
+	if (catch err = len)
+	{
+		case PrintFault.BUFFER_EXCEEDED:
+		case PrintFault.INTERNAL_BUFFER_EXCEEDED:
+			return f.first_err(err)?;
+		default:
+			err = f.first_err(PrintFault.INVALID_ARGUMENT);
+			f.out_substr("<INVALID>")!;
+			return err?;
+	}
+	return len;
+}
+
+fn usz! Formatter.vprintf(&self, String format, any[] anys)
+{
+	self.first_fault = {};
+	if (!self.out_fn)
+	{
+		// use null output function
+		self.out_fn = &out_null_fn;
+	}
+	usz total_len;
+	usz format_len = format.len;
+	usz variant_index = 0;
+	for (usz i = 0; i < format_len; i++)
+	{
+		// format specifier?  %[flags][width][.precision][length]
+		char c = format[i];
+		if (c != '%')
+		{
+			// no
+			total_len += self.out(c)!;
+			continue;
+		}
+		i++;
+		if (i >= format_len) return @report_fault(self, "%ERR");
+		c = format[i];
+		if (c == '%')
+		{
+			total_len += self.out(c)!;
+			continue;
+		}
+		// evaluate flags
+		self.flags = {};
+		while FLAG_EVAL: (true)
+		{
+			switch (c)
+			{
+				case '0': self.flags.zeropad = true;
+				case '-': self.flags.left = true;
+				case '+': self.flags.plus = true;
+				case ' ': self.flags.space = true;
+				case '#': self.flags.hash = true;
+				default: break FLAG_EVAL;
+			}
+			if (++i >= format_len) return @report_fault(self, "%ERR");
+			c = format[i];
+		}
+		// evaluate width field
+		int! w = printf_parse_format_field(anys.ptr, anys.len, &variant_index, format.ptr, format.len, &i);
+		if (catch w) return @report_fault(self, "%ERR");
+		c = format[i];
+		if (w < 0)
+		{
+			self.flags.left = true;
+			w = -w;
+		}
+		self.width = w;
+		// evaluate precision field
+		self.prec = 0;
+		if (c == '.')
+		{
+			self.flags.precision = true;
+			if (++i >= format_len) return @report_fault(self, "<BAD FORMAT>");
+			int! prec = printf_parse_format_field(anys.ptr, anys.len, &variant_index, format.ptr, format.len, &i);
+			if (catch prec) return @report_fault(self, "<BAD FORMAT>");
+			self.prec = prec < 0 ? 0 : prec;
+			c = format[i];
+		}
+
+		// evaluate specifier
+		uint base = 0;
+		if (variant_index >= anys.len)
+		{
+			self.first_err(PrintFault.NOT_ENOUGH_ARGUMENTS);
+			total_len += self.out_substr("<MISSING>")!;
+			continue;
+		}
+		any current = anys[variant_index++];
+		switch (c)
+		{
+			case 'd':
+				base = 10;
+				self.flags.hash = false;
+			case 'X' :
+				self.flags.uppercase = true;
+				nextcase;
+			case 'x' :
+				base = 16;
+			case 'O':
+				self.flags.uppercase = true;
+				nextcase;
+			case 'o' :
+				base = 8;
+			case 'B':
+				self.flags.uppercase = true;
+				nextcase;
+			case 'b' :
+				base = 2;
+			case 'A':
+				self.flags.uppercase = true;
+				nextcase;
+			case 'a':
+				total_len += @wrap_bad(self, self.atoa(float_from_any(current)))!;
+				continue;
+			case 'F' :
+				self.flags.uppercase = true;
+				nextcase;
+			case 'f':
+				total_len += @wrap_bad(self, self.ftoa(float_from_any(current)))!;
+				continue;
+			case 'E':
+				self.flags.uppercase = true;
+				nextcase;
+			case 'e':
+				total_len += @wrap_bad(self, self.etoa(float_from_any(current)))!;
+				continue;
+			case 'G':
+				self.flags.uppercase = true;
+				nextcase;
+			case 'g':
+				total_len += @wrap_bad(self, self.gtoa(float_from_any(current)))!;
+				continue;
+			case 'c':
+				total_len += self.out_char(current)!;
+				continue;
+			case 'H':
+				self.flags.uppercase = true;
+				nextcase;
+			case 'h':
+            	char[] out @noinit;
+				switch (current)
+				{
+					case char[]:
+						out = *current;
+					case ichar[]:
+						out = *(char[]*)current;
+					default:
+						if (current.type.kindof == ARRAY && (current.type.inner == char.typeid || current.type.inner == ichar.typeid))
+						{
+							out = ((char*)current.ptr)[:current.type.sizeof];
+							break;
+						}
+						total_len += self.out_substr("<INVALID>")!;
+						continue;
+				}
+				if (self.flags.left)
+				{
+					usz len = print_hex_chars(self, out, self.flags.uppercase)!;
+					total_len += len;
+					total_len += self.pad(' ', self.width, len)!;
+					continue;
+				}
+				if (self.width)
+				{
+					total_len += self.pad(' ', self.width, out.len * 2)!;
+				}
+				total_len += print_hex_chars(self, out, self.flags.uppercase)!;
+				continue;
+			case 's':
+				if (self.flags.left)
+				{
+					usz len = self.out_str(current)!;
+					total_len += len;
+					total_len += self.pad(' ', self.width, len)!;
+					continue;
+				}
+				if (self.width)
+				{
+					OutputFn out_fn = self.out_fn;
+					self.out_fn = (OutputFn)&out_null_fn;
+					usz len = self.out_str(current)!;
+					self.out_fn = out_fn;
+					total_len += self.pad(' ', self.width, len)!;
+				}
+				total_len += self.out_str(current)!;
+				continue;
+			case 'p':
+				self.flags.zeropad = true;
+				self.flags.hash = true;
+				base = 16;
+			default:
+				self.first_err(PrintFault.INVALID_FORMAT);
+				total_len += self.out_substr("<BAD FORMAT>")!;
+				continue;
+		}
+		if (base != 10)
+		{
+			self.flags.plus = false;
+			self.flags.space = false;
+		}
+		// ignore '0' flag when precision is given
+		if (self.flags.precision) self.flags.zeropad = false;
+
+		bool is_neg;
+		total_len += @wrap_bad(self, self.ntoa(int_from_any(current, &is_neg), is_neg, base))!;
+	}
+	// termination
+	//	out((char)0, buffer, idx < maxlen ? idx : maxlen - 1U, maxlen);
+
+	// return written chars without terminating \0
+	if (self.first_fault) return self.first_fault?;
+	return total_len;
+}
+
+
+fn usz! Formatter.print(&self, String str)
+{
+	if (!self.out_fn)
+	{
+		// use null output function
+		self.out_fn = &out_null_fn;
+	}
+	foreach (c : str) self.out(c)!;
+	return self.idx;
+}