Add deprecation silence notice.

- Multiple issues when rethrowing inside of expressions #2873

.clang-format

@@ -0,0 +1,15 @@
+IndentWidth: 4
+UseCRLF: false
+IndentCaseLabels: true
+UseTab: ForIndentation
+TabWidth: 4
+BreakBeforeBraces: Allman
+AllowShortBlocksOnASingleLine: Empty
+AllowShortIfStatementsOnASingleLine: WithoutElse
+SpaceBeforeAssignmentOperators: true
+SpaceBeforeCaseColon: false
+SpaceBeforeParens: ControlStatementsExceptControlMacros
+SpacesInCStyleCastParentheses: false
+SpacesInConditionalStatement: false
+SpacesInParentheses: false
+SpacesInSquareBrackets: false

.clang-tidy

@@ -1,21 +1,7 @@
 ---
 # Configure clang-tidy for this project.
 
-IndentWidth: 4
-UseCRLF: false
-IndentCaseLabels: true
-UseTab: UT_ForIndentation
-TabWidth: 4
-BreakBeforeBraces: Allman
-AllowShortBlocksOnASingleLine: SBS_Empty
-AllowShortIfStatementsOnASingleLine: SIS_WithoutElse
-SpaceBeforeAssignmentOperators: true
-SpaceBeforeCaseColon: false
-SpaceBeforeParens: SBPO_ControlStatementsExceptControlMacros
-SpacesInCStyleCastParentheses: false
-SpacesInConditionalStatement: false
-SpacesInParentheses: false
-SpacesInSquareBrackets: false
+
 
 # Disabled:
 #  -google-readability-namespace-comments the *_CLIENT_NS is a macro, and
@@ -40,17 +26,19 @@ Checks: >
 
 # Turn all the warnings from the checks above into errors.
 WarningsAsErrors: "*"
-
 CheckOptions:
-  - { key: readability-function-cognitive-complexity.Threshold,    value: 100 }
-  - { key: readability-identifier-naming.StructCase,             value: CamelCase  }
-  - { key: readability-identifier-naming.FunctionCase,           value: lower_case  }
-  - { key: readability-identifier-naming.VariableCase,           value: lower_case }
-  - { key: readability-identifier-naming.MacroDefinitionCase,    value: UPPER_CASE }
-  - { key: readability-identifier-naming.EnumConstantCase,         value: UPPER_CASE }
-  - { key: readability-identifier-naming.ConstexprVariableCase,    value: CamelCase }
-  - { key: readability-identifier-naming.ConstexprVariablePrefix,  value: k         }
-  - { key: readability-identifier-naming.GlobalConstantCase,       value: CamelCase }
-  - { key: readability-identifier-naming.GlobalConstantPrefix,     value: k         }
-  - { key: readability-identifier-naming.StaticConstantCase,       value: CamelCase }
-  - { key: readability-identifier-naming.StaticConstantPrefix,     value: k         }
+  - { key: readability-function-cognitive-complexity.Threshold,        value: 100 }
+  - { key: readability-identifier-naming.StructCase,                   value: CamelCase  }
+  - { key: readability-identifier-naming.FunctionCase,                 value: lower_case  }
+  - { key: readability-identifier-naming.VariableCase,                 value: lower_case }
+  - { key: readability-identifier-naming.MacroDefinitionCase,          value: UPPER_CASE }
+  - { key: readability-identifier-naming.EnumConstantCase,             value: UPPER_CASE }
+  - { key: readability-identifier-naming.ConstexprVariableCase,        value: CamelCase }
+  - { key: readability-identifier-naming.ConstexprVariablePrefix,      value: k         }
+  - { key: readability-identifier-naming.GlobalConstantCase,           value: CamelCase }
+  - { key: readability-identifier-naming.GlobalConstantPrefix,         value: k         }
+  - { key: readability-identifier-naming.StaticConstantCase,           value: CamelCase }
+  - { key: readability-identifier-naming.StaticConstantPrefix,         value: k         }
+  - { key: readability-identifier-naming.MinimumParameterNameLength,   value: 0 }
+
+MinimumParameterNameLength: 0

.gitignore

@@ -73,8 +73,12 @@ out/
 /cmake-build-debug/
 /cmake-build-release/
 
-# Emacs files
+# etags(Emacs), ctags, gtags
 TAGS
+GPATH
+GRTAGS
+GTAGS
+tags
 
 # Clangd LSP files
 /.cache/

CMakeLists.txt

@@ -2,7 +2,7 @@ cmake_minimum_required(VERSION 3.20)
 
 set(C3_LLVM_MIN_VERSION 17)
 set(C3_LLVM_MAX_VERSION 22)
-set(C3_LLVM_DEFAULT_VERSION 19)
+set(C3_LLVM_DEFAULT_VERSION 21)
 
 if (CMAKE_CURRENT_SOURCE_DIR STREQUAL CMAKE_CURRENT_BINARY_DIR)
     message(FATAL_ERROR "In-tree build detected, please build in a separate directory")
@@ -56,12 +56,30 @@ set(CMAKE_FIND_PACKAGE_SORT_DIRECTION DEC)
 set(CMAKE_C_STANDARD 11)
 set(CMAKE_CXX_STANDARD 17)
 
-# Use /MT or /MTd
-set(CMAKE_MSVC_RUNTIME_LIBRARY "MultiThreaded$<$<CONFIG:Debug>:Debug>")
-
 if(MSVC)
     message(STATUS "MSVC version ${MSVC_VERSION}")
     add_compile_options(/utf-8)
+
+    if(C3_WITH_LLVM)
+        FetchContent_GetProperties(LLVM_Windows)
+        if(NOT LLVM_Windows_URL MATCHES "msvcrt")
+            set(MSVC_CRT_SUFFIX "")
+            message(STATUS "Detected STATIC LLVM (libcmt)")
+        else()
+            set(MSVC_CRT_SUFFIX "DLL")
+            message(STATUS "Detected DYNAMIC LLVM (msvcrt)")
+        endif()
+    endif()
+
+    # Force the Runtime to Release (/MT or /MD) even in Debug
+    # This is required to match our RelWithDebInfo LLVM builds
+    set(CMAKE_MSVC_RUNTIME_LIBRARY "MultiThreaded${MSVC_CRT_SUFFIX}")
+    set_property(GLOBAL PROPERTY MSVC_RUNTIME_LIBRARY "MultiThreaded${MSVC_CRT_SUFFIX}")
+
+    add_compile_definitions(_ITERATOR_DEBUG_LEVEL=0)
+
+    # Suppresses the LNK4098 mismatch warning in Debug builds
+    set(CMAKE_EXE_LINKER_FLAGS_DEBUG "${CMAKE_EXE_LINKER_FLAGS_DEBUG} /NODEFAULTLIB:libcmtd /NODEFAULTLIB:msvcrtd")
 else()
     add_compile_options(-gdwarf-3 -fno-exceptions)
 
@@ -77,6 +95,7 @@ set(C3_USE_MIMALLOC         OFF                         CACHE BOOL   "Use built-
 set(C3_MIMALLOC_TAG         "v1.7.3"                    CACHE STRING "Used version of mimalloc")
 set(C3_USE_TB               OFF                         CACHE BOOL   "Use TB")
 set(C3_LLD_DIR              ""                          CACHE STRING "Use custom LLD directory")
+set(C3_LLD_INCLUDE_DIR      ""                          CACHE STRING "Use custom LLD include directory")
 set(C3_ENABLE_CLANGD_LSP    OFF                         CACHE BOOL   "Enable/Disable output of compile commands during generation")
 set(LLVM_CRT_LIBRARY_DIR    ""                          CACHE STRING "Use custom llvm's compiler-rt directory")
 set(TCC_LIB_PATH            "/usr/lib/tcc/libtcc1.a"    CACHE STRING "Use custom libtcc1.a path")
@@ -142,11 +161,11 @@ if(C3_WITH_LLVM)
         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
+                URL https://github.com/c3lang/win-llvm/releases/download/llvm_21_1_8/llvm-21.1.8-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
+                URL https://github.com/c3lang/win-llvm/releases/download/llvm_21_1_8/llvm-21.1.8-windows-amd64-msvc17-libcmt-dbg.7z
         )
         if(CMAKE_BUILD_TYPE STREQUAL "Debug")
             message("Loading Windows LLVM debug libraries, this may take a while...")
@@ -273,7 +292,12 @@ if(C3_WITH_LLVM)
     else()
         message(STATUS "Looking for shared lld libraries in ${LLVM_LIBRARY_DIRS}")
 
-        find_library(LLVM NAMES libLLVM.so PATHS ${LLVM_LIBRARY_DIRS} NO_DEFAULT_PATH REQUIRED)
+        #find_library(LLVM NAMES libLLVM.so PATHS ${LLVM_LIBRARY_DIRS} NO_DEFAULT_PATH REQUIRED)
+        if(UNIX AND NOT WIN32)
+            find_library(LLVM NAMES libLLVM.so PATHS ${LLVM_LIBRARY_DIRS} REQUIRED)
+        else()
+            find_library(LLVM NAMES libLLVM.a LLVM.lib PATHS ${LLVM_LIBRARY_DIRS} REQUIRED)
+        endif()        
         set(llvm_libs ${LLVM})
 
         # These don't seem to be reliable on windows.
@@ -435,10 +459,18 @@ if(C3_WITH_LLVM)
     target_compile_definitions(c3c PUBLIC LLVM_AVAILABLE=1)
     add_library(c3c_wrappers STATIC wrapper/src/wrapper.cpp)
     if (MSVC)
+        # Use the same detected CRT for the wrapper
+        set_target_properties(c3c_wrappers PROPERTIES MSVC_RUNTIME_LIBRARY "MultiThreaded${MSVC_CRT_SUFFIX}")
+        set_target_properties(miniz PROPERTIES MSVC_RUNTIME_LIBRARY "MultiThreaded${MSVC_CRT_SUFFIX}")
+
         target_compile_options(c3c PRIVATE
                 "$<$<CONFIG:Debug>:/EHa>"
                 "$<$<CONFIG:Release>:/EHsc>")
     endif()
+
+    if(C3_LLD_INCLUDE_DIR)
+        target_include_directories(c3c_wrappers PRIVATE ${C3_LLD_INCLUDE_DIR})
+    endif()
 else()
     target_sources(c3c PRIVATE src/utils/hostinfo.c)
     target_compile_definitions(c3c PUBLIC LLVM_AVAILABLE=0)
@@ -552,13 +584,21 @@ if(MSVC)
     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
-                ${clang_lib_dir}/clang_rt.asan_dynamic_runtime_thunk-x86_64.lib)
+        # The the sanitizer libs are in the folder "lib/clang/21/lib/windows/" but use the find anyway
+        file(GLOB_RECURSE FOUND_ASAN_LIB "${llvm_dir}/*clang_rt.asan_dynamic-x86_64.lib")
+        if(FOUND_ASAN_LIB)
+            list(GET FOUND_ASAN_LIB 0 _asan_path)
+            get_filename_component(_asan_dir "${_asan_path}" DIRECTORY)
+            set(sanitizer_runtime_libraries
+                    ${_asan_dir}/clang_rt.asan_dynamic-x86_64.lib
+                    ${_asan_dir}/clang_rt.asan_dynamic-x86_64.dll
+                    ${_asan_dir}/clang_rt.asan_dynamic_runtime_thunk-x86_64.lib)
+            message(STATUS "Found Sanitizer binaries at: ${_asan_dir}")
+        else()
+            message(WARNING "Could not find sanitizer runtime libraries in ${llvm_dir}")
+        endif()
     endif()
+
 else()
     if (C3_WITH_LLVM AND NOT LLVM_ENABLE_RTTI)
         target_compile_options(c3c_wrappers PRIVATE -fno-rtti)
@@ -572,6 +612,7 @@ else()
         -Wno-unused-function
         -Wno-unused-variable
         -Wno-unused-parameter
+        -Wno-char-subscripts
     )
     target_link_options(c3c PRIVATE -pthread)
 endif()

README.md

@@ -36,10 +36,10 @@ whole new language.
 
 ### Example code
 
-The following code shows [generic modules](https://c3-lang.org/generic-programming/generics/) (more examples can be found at https://c3-lang.org/language-overview/examples/). 
+The following code shows [generics](https://c3-lang.org/generic-programming/generics/) (more examples can be found at https://c3-lang.org/language-overview/examples/). 
 
-```cpp
-module stack {Type};
+```c3
+module stack <Type>;
 // Above: the parameterized type is applied to the entire module.
 
 struct Stack
@@ -78,7 +78,7 @@ fn bool Stack.empty(Stack* this)
 
 Testing it out:
 
-```cpp
+```c3
 import stack;
 
 // Define our new types, the first will implicitly create 
@@ -142,7 +142,7 @@ fn void main()
 
 ### Current status
 
-The current stable version of the compiler is **version 0.7.8**.
+The current stable version of the compiler is **version 0.7.9**.
 
 The upcoming 0.7.x releases will focus on expanding the standard library,
 fixing bugs and improving compile time analysis. 
@@ -384,7 +384,7 @@ scoop install c3
 #### Getting started with a "hello world"
 
 Create a `main.c3` file with:
-```c++
+```c3
 module hello_world;
 import std::io;
 
@@ -513,7 +513,7 @@ provide the link path to the LLVM CMake directories, e.g. `cmake -B build -S . -
 
 *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.*
+this functionality is non-essential and it is perfectly fine to use the compiler without it.*
 
 #### Licensing
 
@@ -523,7 +523,7 @@ which is licensed under LGPL 3.0.
 
 This means you are free to use all parts of standard library, 
 tests, benchmarks, grammar, examples and so on under the MIT license, including 
-using those libraries and tests if your build your own C3 compiler.
+using those libraries and tests if you build your own C3 compiler.
 
 #### Editor plugins
 

benchmarks/stdlib/crypto/chacha20.c3

@@ -0,0 +1,39 @@
+// Copyright (c) 2025 Zack Puhl <github@xmit.xyz>. 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 chacha20_benchmarks;
+
+import std::crypto::chacha20;
+
+
+fn void initialize_bench() @init
+{
+	set_benchmark_warmup_iterations(3);
+	set_benchmark_max_iterations(1024);
+}
+
+const char[] KEY = x'98bef1469be7269837a45bfbc92a5a6ac762507cf96443bf33b96b1bd4c6f8f6';
+const char[] NONCE = x'44e792d63335abb1582e9253';
+const uint COUNTER = 42;
+
+char[] one_mb @align(ulong.sizeof) = { [0..1024*1024] = 0xA5 };
+
+// This doesn't test both encryption + decryption, because it's a symmetric operation that shares
+//   a single common data transformation. Testing one limb is enough.
+fn void gogo_chacha20() @benchmark
+{
+	chacha20::encrypt_mut(one_mb[..], KEY, NONCE, COUNTER);
+}
+
+// Check what the speed of an unligned buffer looks like.
+fn void gogo_chacha20_unaligned() @benchmark => @pool()
+{
+    char[] copy = mem::talloc_array(char, one_mb.len + 3);
+    char[] im_off_slightly = copy[3..];
+    copy[3..] = one_mb[..];
+    assert((usz)im_off_slightly.ptr % usz.sizeof > 0);
+
+    runtime::@start_benchmark();
+	chacha20::encrypt_mut(im_off_slightly, KEY, NONCE, COUNTER);
+    runtime::@end_benchmark();
+}

benchmarks/stdlib/crypto/crypto_shootout.c3

@@ -0,0 +1,113 @@
+// Copyright (c) 2025 Zack Puhl <github@xmit.xyz>. 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 crypto_hash_benchmarks;
+
+import std::collections::pair;
+
+
+const usz COMMON_ITERATIONS = 1 << 17;
+
+char* common_1mib_ptr;
+char[] common_16;
+char[] common_256;
+char[] common_4kib;
+char[] common_1mib;
+
+fn void initialize_bench() @init
+{
+	set_benchmark_warmup_iterations(3);
+	set_benchmark_max_iterations(COMMON_ITERATIONS + 3);
+
+	common_1mib_ptr = mem::alloc_array(char, 1024*1024);
+	common_1mib = common_1mib_ptr[:1024*1024];
+	common_1mib[..] = 0xA5;
+
+	common_16 = common_1mib[:16];
+	common_256 = common_1mib[:256];
+	common_4kib = common_1mib[:4096];
+
+	static String[] function_prefixes = {
+		$qnameof(md5_16)[..^4],
+		$qnameof(sha1_16)[..^4],
+		$qnameof(sha2_256_16)[..^4],
+		$qnameof(sha2_512_16)[..^4],
+		$qnameof(blake2s_256_16)[..^4],
+		$qnameof(blake2b_256_16)[..^4],
+		$qnameof(blake3_16)[..^4],
+		$qnameof(ripemd_160_16)[..^4],
+		$qnameof(whirlpool_16)[..^4],
+		$qnameof(streebog_256_16)[..^4],
+		$qnameof(streebog_512_16)[..^4],
+	};
+
+	static Pair{ String, uint }[] to_iters = {
+		{ "_4kib", 1 << 15 },
+		{ "_1mib", 1024 },
+	};
+
+	foreach (p : to_iters)
+	{
+		foreach (name : function_prefixes) set_benchmark_func_iterations(name.tconcat(p.first), p.second);
+	}
+}
+
+fn void teardown_bench() @finalizer
+{
+	mem::free(common_1mib_ptr);
+}
+
+
+// =======================================================================================
+module crypto_hash_benchmarks @benchmark;
+
+import std::hash;
+
+
+fn void md5_16() => md5::hash(common_16);
+fn void sha1_16() => sha1::hash(common_16);
+fn void sha2_256_16() => sha256::hash(common_16);
+fn void sha2_512_16() => sha512::hash(common_16);
+fn void blake2s_256_16() => blake2::s(256, common_16);
+fn void blake2b_256_16() => blake2::b(256, common_16);
+fn void blake3_16() => blake3::hash(common_16);
+fn void ripemd_160_16() => ripemd::hash{160}(common_16);
+fn void whirlpool_16() => whirlpool::hash(common_16);
+fn void streebog_256_16() => streebog::hash_256(common_16);
+fn void streebog_512_16() => streebog::hash_512(common_16);
+
+fn void md5_256() => md5::hash(common_256);
+fn void sha1_256() => sha1::hash(common_256);
+fn void sha2_256_256() => sha256::hash(common_256);
+fn void sha2_512_256() => sha512::hash(common_256);
+fn void blake2s_256_256() => blake2::s(256, common_256);
+fn void blake2b_256_256() => blake2::b(256, common_256);
+fn void blake3_256() => blake3::hash(common_256);
+fn void ripemd_160_256() => ripemd::hash{160}(common_256);
+fn void whirlpool_256() => whirlpool::hash(common_256);
+fn void streebog_256_256() => streebog::hash_256(common_256);
+fn void streebog_512_256() => streebog::hash_512(common_256);
+
+fn void md5_4kib() => md5::hash(common_4kib);
+fn void sha1_4kib() => sha1::hash(common_4kib);
+fn void sha2_256_4kib() => sha256::hash(common_4kib);
+fn void sha2_512_4kib() => sha512::hash(common_4kib);
+fn void blake2s_256_4kib() => blake2::s(256, common_4kib);
+fn void blake2b_256_4kib() => blake2::b(256, common_4kib);
+fn void blake3_4kib() => blake3::hash(common_4kib);
+fn void ripemd_160_4kib() => ripemd::hash{160}(common_4kib);
+fn void whirlpool_4kib() => whirlpool::hash(common_4kib);
+fn void streebog_256_4kib() => streebog::hash_256(common_4kib);
+fn void streebog_512_4kib() => streebog::hash_512(common_4kib);
+
+fn void md5_1mib() => md5::hash(common_1mib);
+fn void sha1_1mib() => sha1::hash(common_1mib);
+fn void sha2_256_1mib() => sha256::hash(common_1mib);
+fn void sha2_512_1mib() => sha512::hash(common_1mib);
+fn void blake2s_256_1mib() => blake2::s(256, common_1mib);
+fn void blake2b_256_1mib() => blake2::b(256, common_1mib);
+fn void blake3_1mib() => blake3::hash(common_1mib);
+fn void ripemd_160_1mib() => ripemd::hash{160}(common_1mib);
+fn void whirlpool_1mib() => whirlpool::hash(common_1mib);
+fn void streebog_256_1mib() => streebog::hash_256(common_1mib);
+fn void streebog_512_1mib() => streebog::hash_512(common_1mib);

benchmarks/stdlib/hash/ripemd.c3

@@ -0,0 +1,76 @@
+// Copyright (c) 2025 Zack Puhl <github@xmit.xyz>. 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 ripemd_bench;
+
+fn void initialize_bench() @init
+{
+	set_benchmark_warmup_iterations(3);
+	set_benchmark_max_iterations(256);
+
+    input = mem::alloc_array(char, BUFSZ);
+    input[:BUFSZ] = (char[]){ [0..BUFSZ-1] = 0xA5 }[..];
+    input_slice = input[:BUFSZ];
+}
+
+fn void teardown_bench() @finalizer
+{
+    mem::free(input);
+    input = null;
+}
+
+char* input;
+char[] input_slice;
+const usz BUFSZ = 1024 * 1024;
+
+module ripemd_bench @benchmark;
+
+import std::hash;
+
+fn void ripemd_128()
+{
+    runtime::@start_benchmark();
+    char[*] myset = ripemd::hash{128}(input_slice);
+    runtime::@end_benchmark();
+    mem::zero_volatile(myset[..]);
+}
+
+fn void ripemd_160()
+{
+    runtime::@start_benchmark();
+    char[*] myset = ripemd::hash{160}(input_slice);
+    runtime::@end_benchmark();
+    mem::zero_volatile(myset[..]);
+}
+
+fn void ripemd_256()
+{
+    runtime::@start_benchmark();
+    char[*] myset = ripemd::hash{256}(input_slice);
+    runtime::@end_benchmark();
+    mem::zero_volatile(myset[..]);
+}
+
+fn void ripemd_320()
+{
+    runtime::@start_benchmark();
+    char[*] myset = ripemd::hash{320}(input_slice);
+    runtime::@end_benchmark();
+    mem::zero_volatile(myset[..]);
+}
+
+fn void compared_with_sha256()
+{
+    runtime::@start_benchmark();
+    char[*] myset = sha256::hash(input_slice);
+    runtime::@end_benchmark();
+    mem::zero_volatile(myset[..]);
+}
+
+fn void compared_with_whirlpool()
+{
+    runtime::@start_benchmark();
+    char[*] myset = whirlpool::hash(input_slice);
+    runtime::@end_benchmark();
+    mem::zero_volatile(myset[..]);
+}

benchmarks/stdlib/hash/streebog.c3

@@ -0,0 +1,51 @@
+// Copyright (c) 2025 Zack Puhl <github@xmit.xyz>. 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 streebog_bench;
+
+fn void initialize_bench() @init
+{
+	set_benchmark_warmup_iterations(3);
+	set_benchmark_max_iterations(256);
+}
+
+const char[] INPUT = { [0..1024*1024] = 0xA5 };
+const char[*] EXPECTED_256 = x'694676905b7cf099755db1cc186f741f0fd1877aaaa4badcbfb305537f986971';
+const char[*] EXPECTED_512 = x'fe21d08857ea97e79035d1e5c9ba5130786e8d1875bc74d628349560d94d6bdff0b0dcd2f6347eb8b3f0239b6cca76b5028c0ff45f631fcdf77b1d551dd079f3';
+
+
+module streebog_bench @benchmark;
+
+import std::hash;
+
+fn void get_in_the_bog_256()
+{
+    runtime::@start_benchmark();
+    char[*] myset = streebog::hash_256(INPUT);
+    runtime::@end_benchmark();
+    mem::zero_volatile(myset[..]);
+}
+
+fn void get_in_the_bog_512()
+{
+    runtime::@start_benchmark();
+    char[*] myset = streebog::hash_512(INPUT);
+    runtime::@end_benchmark();
+    mem::zero_volatile(myset[..]);
+}
+
+fn void compared_with_sha256()
+{
+    runtime::@start_benchmark();
+    char[*] myset = sha256::hash(INPUT);
+    runtime::@end_benchmark();
+    mem::zero_volatile(myset[..]);
+}
+
+fn void compared_with_whirlpool()
+{
+    runtime::@start_benchmark();
+    char[*] myset = whirlpool::hash(INPUT);
+    runtime::@end_benchmark();
+    mem::zero_volatile(myset[..]);
+}

lib/std/atomic.c3

@@ -1,7 +1,7 @@
 // Copyright (c) 2023-2025 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::types{Type};
+module std::atomic::types <Type>;
 
 struct Atomic
 {

lib/std/collections/anylist.c3

@@ -70,7 +70,7 @@ fn any? AnyList.last_any(&self) @inline => InterfaceList {any}.last(self);
 *>
 macro AnyList.pop(&self, $Type)
 {
-	if (!self.size) return NO_MORE_ELEMENT?;
+	if (!self.size) return NO_MORE_ELEMENT~;
 	defer self.free_element(self.entries[self.size]);
 	return *anycast(self.entries[--self.size], $Type);
 }
@@ -85,7 +85,7 @@ macro AnyList.pop(&self, $Type)
 *>
 macro AnyList.pop_first(&self, $Type)
 {
-	if (!self.size) return NO_MORE_ELEMENT?;
+	if (!self.size) return NO_MORE_ELEMENT~;
 	defer self.remove_at(0);
 	return *anycast(self.entries[0], $Type);
 }

lib/std/collections/bitset.c3

@@ -4,7 +4,7 @@
 <*
  @require SIZE > 0 : "The size of the bitset in bits must be at least 1"
 *>
-module std::collections::bitset {SIZE};
+module std::collections::bitset <SIZE>;
 
 const BITS = uint.sizeof * 8;
 const SZ = (SIZE + BITS - 1) / BITS;
@@ -172,7 +172,7 @@ fn void BitSet.set_bool(&self, usz i, bool value) @operator([]=) @inline
 <*
  @require Type.kindof == UNSIGNED_INT
 *>
-module std::collections::growablebitset{Type};
+module std::collections::growablebitset <Type>;
 import std::collections::list;
 
 const BITS = Type.sizeof * 8;

lib/std/collections/elastic_array.c3

@@ -4,7 +4,7 @@
 <*
  @require MAX_SIZE >= 1 : `The size must be at least 1 element big.`
 *>
-module std::collections::elastic_array {Type, MAX_SIZE};
+module std::collections::elastic_array <Type, MAX_SIZE>;
 import std::io, std::math, std::collections::list_common;
 
 alias ElementPredicate = fn bool(Type *type);
@@ -46,7 +46,7 @@ fn String ElasticArray.to_tstring(&self)
 
 fn void? ElasticArray.push_try(&self, Type element) @inline
 {
-	if (self.size == MAX_SIZE) return mem::OUT_OF_MEMORY?;
+	if (self.size == MAX_SIZE) return mem::OUT_OF_MEMORY~;
 	self.entries[self.size++] = element;
 }
 
@@ -60,7 +60,7 @@ fn void ElasticArray.push(&self, Type element) @inline
 
 fn Type? ElasticArray.pop(&self)
 {
-	if (!self.size) return NO_MORE_ELEMENT?;
+	if (!self.size) return NO_MORE_ELEMENT~;
 	return self.entries[--self.size];
 }
 
@@ -74,7 +74,7 @@ fn void ElasticArray.clear(&self)
 *>
 fn Type? ElasticArray.pop_first(&self)
 {
-	if (!self.size) return NO_MORE_ELEMENT?;
+	if (!self.size) return NO_MORE_ELEMENT~;
 	defer self.remove_at(0);
 	return self.entries[0];
 }
@@ -241,7 +241,7 @@ fn void? ElasticArray.push_front_try(&self, Type type) @inline
 *>
 fn void? ElasticArray.insert_at_try(&self, usz index, Type value)
 {
-	if (self.size == MAX_SIZE) return mem::OUT_OF_MEMORY?;
+	if (self.size == MAX_SIZE) return mem::OUT_OF_MEMORY~;
 	self.insert_at(index, value);
 }
 
@@ -269,25 +269,25 @@ fn void ElasticArray.set_at(&self, usz index, Type type)
 
 fn void? ElasticArray.remove_last(&self) @maydiscard
 {
-	if (!self.size) return NO_MORE_ELEMENT?;
+	if (!self.size) return NO_MORE_ELEMENT~;
 	self.size--;
 }
 
 fn void? ElasticArray.remove_first(&self) @maydiscard
 {
-	if (!self.size) return NO_MORE_ELEMENT?;
+	if (!self.size) return NO_MORE_ELEMENT~;
 	self.remove_at(0);
 }
 
 fn Type? ElasticArray.first(&self)
 {
-	if (!self.size) return NO_MORE_ELEMENT?;
+	if (!self.size) return NO_MORE_ELEMENT~;
 	return self.entries[0];
 }
 
 fn Type? ElasticArray.last(&self)
 {
-	if (!self.size) return NO_MORE_ELEMENT?;
+	if (!self.size) return NO_MORE_ELEMENT~;
 	return self.entries[self.size - 1];
 }
 
@@ -368,7 +368,7 @@ fn usz? ElasticArray.index_of(&self, Type type) @if(ELEMENT_IS_EQUATABLE)
 	{
 		if (equals(v, type)) return i;
 	}
-	return NOT_FOUND?;
+	return NOT_FOUND~;
 }
 
 fn usz? ElasticArray.rindex_of(&self, Type type) @if(ELEMENT_IS_EQUATABLE)
@@ -377,7 +377,7 @@ fn usz? ElasticArray.rindex_of(&self, Type type) @if(ELEMENT_IS_EQUATABLE)
 	{
 		if (equals(v, type)) return i;
 	}
-	return NOT_FOUND?;
+	return NOT_FOUND~;
 }
 
 fn bool ElasticArray.equals(&self, ElasticArray other_list) @if(ELEMENT_IS_EQUATABLE)

lib/std/collections/enummap.c3

@@ -1,7 +1,8 @@
 <*
  @require Enum.kindof == TypeKind.ENUM : "Only enums may be used with an enummap"
+ @require Enum.values.len > 0 : "Only non-empty enums may be used with enummap"
 *>
-module std::collections::enummap{Enum, ValueType};
+module std::collections::enummap <Enum, ValueType>;
 import std::io;
 
 struct EnumMap (Printable)

lib/std/collections/enumset.c3

@@ -5,7 +5,7 @@
 <*
  @require Enum.kindof == TypeKind.ENUM : "Only enums may be used with an enumset"
 *>
-module std::collections::enumset{Enum};
+module std::collections::enumset <Enum>;
 import std::io;
 
 const ENUM_COUNT @private = Enum.values.len;

lib/std/collections/hashmap.c3

@@ -4,7 +4,7 @@
 <*
  @require $defined((Key){}.hash()) : `No .hash function found on the key`
 *>
-module std::collections::map{Key, Value};
+module std::collections::map <Key, Value>;
 import std::math;
 import std::io @norecurse;
 
@@ -175,13 +175,13 @@ fn usz HashMap.len(&map) @inline
 
 fn Value*? HashMap.get_ref(&map, Key key)
 {
-	if (!map.count) return NOT_FOUND?;
+	if (!map.count) return NOT_FOUND~;
 	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 NOT_FOUND?;
+	return NOT_FOUND~;
 }
 
 fn Value* HashMap.get_or_create_ref(&map, Key key) @operator(&[])
@@ -204,13 +204,13 @@ fn Value* HashMap.get_or_create_ref(&map, Key key) @operator(&[])
 
 fn Entry*? HashMap.get_entry(&map, Key key)
 {
-	if (!map.count) return NOT_FOUND?;
+	if (!map.count) return NOT_FOUND~;
 	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 NOT_FOUND?;
+	return NOT_FOUND~;
 }
 
 <*
@@ -275,7 +275,7 @@ fn bool HashMap.set(&map, Key key, Value value) @operator([]=)
 
 fn void? HashMap.remove(&map, Key key) @maydiscard
 {
-	if (!map.remove_entry_for_key(key)) return NOT_FOUND?;
+	if (!map.remove_entry_for_key(key)) return NOT_FOUND~;
 }
 
 fn void HashMap.clear(&map)

lib/std/collections/hashset.c3

@@ -1,7 +1,7 @@
 <*
  @require $defined((Value){}.hash()) : `No .hash function found on the value`
 *>
-module std::collections::set {Value};
+module std::collections::set <Value>;
 import std::math;
 import std::io @norecurse;
 
@@ -32,7 +32,7 @@ struct HashSet (Printable)
 	float load_factor;
 }
 
-fn int HashSet.len(&self) @operator(len) => (int) self.count;
+fn usz HashSet.len(&self) @operator(len) => self.count;
 
 <*
  @param [&inout] allocator : "The allocator to use"
@@ -258,7 +258,7 @@ fn bool HashSet.contains(&set, Value value)
 *>
 fn void? HashSet.remove(&set, Value value) @maydiscard
 {
-	if (!set.remove_entry_for_value(value)) return NOT_FOUND?;
+	if (!set.remove_entry_for_value(value)) return NOT_FOUND~;
 }
 
 fn usz HashSet.remove_all(&set, Value[] values)
@@ -634,7 +634,7 @@ fn Value? HashSetIterator.next(&self)
 		}
 	}
 
-	return NOT_FOUND?;
+	return NOT_FOUND~;
 }
 
 fn usz HashSetIterator.len(&self) @operator(len)

lib/std/collections/interfacelist.c3

@@ -4,7 +4,7 @@
 <*
  @require Type.kindof == INTERFACE || Type.kindof == ANY : "The kind of an interfacelist must be an interface or `any`"
 *>
-module std::collections::interfacelist {Type};
+module std::collections::interfacelist <Type>;
 import std::io,std::math;
 
 alias InterfacePredicate = fn bool(Type value);
@@ -93,7 +93,7 @@ fn void InterfaceList.free_element(&self, Type element) @inline
 *>
 fn Type? InterfaceList.copy_pop(&self, Allocator allocator)
 {
-	if (!self.size) return NO_MORE_ELEMENT?;
+	if (!self.size) return NO_MORE_ELEMENT~;
 	defer self.free_element(self.entries[self.size]);
 	return (Type)allocator::clone_any(allocator, self.entries[--self.size]);
 }
@@ -114,7 +114,7 @@ fn Type? InterfaceList.tcopy_pop(&self) => self.copy_pop(tmem);
 *>
 fn Type? InterfaceList.pop_retained(&self)
 {
-	if (!self.size) return NO_MORE_ELEMENT?;
+	if (!self.size) return NO_MORE_ELEMENT~;
 	return self.entries[--self.size];
 }
 
@@ -138,7 +138,7 @@ fn void InterfaceList.clear(&self)
 *>
 fn Type? InterfaceList.pop_first_retained(&self)
 {
-	if (!self.size) return NO_MORE_ELEMENT?;
+	if (!self.size) return NO_MORE_ELEMENT~;
 	defer self.remove_at(0);
 	return self.entries[0];
 }
@@ -152,7 +152,7 @@ fn Type? InterfaceList.pop_first_retained(&self)
 *>
 fn Type? InterfaceList.copy_pop_first(&self, Allocator allocator)
 {
-	if (!self.size) return NO_MORE_ELEMENT?;
+	if (!self.size) return NO_MORE_ELEMENT~;
 	defer self.free_element(self.entries[self.size]);
 	defer self.remove_at(0);
 	return (Type)allocator::clone_any(allocator, self.entries[0]);
@@ -276,7 +276,7 @@ fn void InterfaceList.remove_first(&self)
 *>
 fn Type? InterfaceList.first(&self) @inline
 {
-	return self.size ? self.entries[0] : NO_MORE_ELEMENT?;
+	return self.size ? self.entries[0] : NO_MORE_ELEMENT~;
 }
 
 <*
@@ -287,7 +287,7 @@ fn Type? InterfaceList.first(&self) @inline
 *>
 fn Type? InterfaceList.last(&self) @inline
 {
-	return self.size ? self.entries[self.size - 1] : NO_MORE_ELEMENT?;
+	return self.size ? self.entries[self.size - 1] : NO_MORE_ELEMENT~;
 }
 
 <*

lib/std/collections/linked_blockingqueue.c3

@@ -1,4 +1,4 @@
-module std::collections::blockingqueue { Value };
+module std::collections::blockingqueue <Value>;
 import std::thread, std::time;
 
 
@@ -70,9 +70,9 @@ fn void LinkedBlockingQueue.free(&self)
 		}
 	};
 
-	(void)self.lock.destroy();
-	(void)self.not_empty.destroy();
-	(void)self.not_full.destroy();
+	self.lock.destroy();
+	self.not_empty.destroy();
+	self.not_full.destroy();
 }
 
 fn void LinkedBlockingQueue.link_entry(&self, QueueEntry* entry) @private 
@@ -126,7 +126,7 @@ fn void LinkedBlockingQueue.push(&self, Value value)
 	{
 		while (self.capacity > 0 && self.count >= self.capacity)
 		{
-			self.not_full.wait(&self.lock)!!;
+			self.not_full.wait(&self.lock);
 		}
 
 		QueueEntry* entry = allocator::new(self.allocator, QueueEntry, {
@@ -137,7 +137,7 @@ fn void LinkedBlockingQueue.push(&self, Value value)
 		self.link_entry(entry);
 
 		// Signal that queue is no longer empty
-		self.not_empty.signal()!!;
+		self.not_empty.signal();
 	};
 }
 
@@ -153,7 +153,7 @@ fn Value LinkedBlockingQueue.poll(&self)
 	{
 		while (self.count == 0)
 		{
-			self.not_empty.wait(&self.lock)!!;
+			self.not_empty.wait(&self.lock);
 		}
 
 		QueueEntry* entry = self.unlink_head();
@@ -161,7 +161,7 @@ fn Value LinkedBlockingQueue.poll(&self)
 		allocator::free(self.allocator, entry);
 		if (self.capacity > 0)
 		{
-			self.not_full.signal()!!;
+			self.not_full.signal();
 		}
 		return value;
 	};
@@ -178,7 +178,7 @@ fn Value? LinkedBlockingQueue.pop(&self)
 {
 	self.lock.@in_lock()
 	{
-		if (self.count == 0) return NO_MORE_ELEMENT?;
+		if (self.count == 0) return NO_MORE_ELEMENT~;
 
 		QueueEntry* entry = self.unlink_head();
 		Value value = entry.value;
@@ -186,7 +186,7 @@ fn Value? LinkedBlockingQueue.pop(&self)
 
 		if (self.capacity > 0)
 		{
-			self.not_full.signal()!!;
+			self.not_full.signal();
 		}
 		return value;
 	};
@@ -214,7 +214,7 @@ fn Value? LinkedBlockingQueue.poll_timeout(&self, Duration timeout)
         		if (end <= time::now()) break;
         		if (catch self.not_empty.wait_until(&self.lock, end)) break;
 			}
-			if (!self.count) return NO_MORE_ELEMENT?;
+			if (!self.count) return NO_MORE_ELEMENT~;
 		}
 
 		QueueEntry* entry = self.unlink_head();
@@ -224,7 +224,7 @@ fn Value? LinkedBlockingQueue.poll_timeout(&self, Duration timeout)
 		// Must signal not_full after removing an item
 		if (self.capacity > 0)
 		{
-			self.not_full.signal()!!;
+			self.not_full.signal();
 		}
 		return value;
 	};
@@ -266,7 +266,7 @@ fn void? LinkedBlockingQueue.try_push(&self, Value value)
 {
 	self.lock.@in_lock()
 	{
-		if (self.capacity > 0 && self.count >= self.capacity) return CAPACITY_EXCEEDED?;
+		if (self.capacity > 0 && self.count >= self.capacity) return CAPACITY_EXCEEDED~;
 
 		QueueEntry* entry = allocator::new(self.allocator, QueueEntry, {
 			.value = value,
@@ -274,7 +274,7 @@ fn void? LinkedBlockingQueue.try_push(&self, Value value)
 			.prev = null
 		});
 		self.link_entry(entry);
-		self.not_empty.signal()!!;
+		self.not_empty.signal();
 	};
 }
 
@@ -299,7 +299,7 @@ fn void? LinkedBlockingQueue.push_timeout(&self, Value value, Duration timeout)
         		if (end <= time::now()) break;
 				if (catch self.not_empty.wait_until(&self.lock, end)) break;
 			}
-			if (self.capacity > 0 && self.count >= self.capacity) return CAPACITY_EXCEEDED?;
+			if (self.capacity > 0 && self.count >= self.capacity) return CAPACITY_EXCEEDED~;
 		}
 
 		QueueEntry* entry = allocator::new(self.allocator, QueueEntry, {
@@ -308,19 +308,19 @@ fn void? LinkedBlockingQueue.push_timeout(&self, Value value, Duration timeout)
 			.prev = null
 		});
 		self.link_entry(entry);
-		self.not_empty.signal()!!;
+		self.not_empty.signal();
 	};
 }
 
 <*
  @require self.is_initialized() : "Queue must be initialized"
- @return "The head value or NO_MORE_ELEMENT? if queue is empty"
+ @return "The head value or NO_MORE_ELEMENT~ if queue is empty"
 *>
 fn Value? LinkedBlockingQueue.peek(&self)
 {
 	self.lock.@in_lock()
 	{
-		return (self.head != null) ? self.head.value : NO_MORE_ELEMENT?;
+		return (self.head != null) ? self.head.value : NO_MORE_ELEMENT~;
 	};
 }
 

lib/std/collections/linked_hashmap.c3

@@ -4,7 +4,7 @@
 <*
  @require $defined((Key){}.hash()) : `No .hash function found on the key`
 *>
-module std::collections::map{Key, Value};
+module std::collections::map <Key, Value>;
 import std::math;
 import std::io @norecurse;
 
@@ -172,24 +172,24 @@ fn usz LinkedHashMap.len(&map) @inline => map.count;
 
 fn Value*? LinkedHashMap.get_ref(&map, Key key)
 {
-	if (!map.count) return NOT_FOUND?;
+	if (!map.count) return NOT_FOUND~;
 	uint hash = rehash(key.hash());
 	for (LinkedEntry *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 NOT_FOUND?;
+	return NOT_FOUND~;
 }
 
 fn LinkedEntry*? LinkedHashMap.get_entry(&map, Key key)
 {
-	if (!map.count) return NOT_FOUND?;
+	if (!map.count) return NOT_FOUND~;
 	uint hash = rehash(key.hash());
 	for (LinkedEntry *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 NOT_FOUND?;
+	return NOT_FOUND~;
 }
 
 <*
@@ -248,7 +248,7 @@ fn bool LinkedHashMap.set(&map, Key key, Value value) @operator([]=)
 
 fn void? LinkedHashMap.remove(&map, Key key) @maydiscard
 {
-	if (!map.remove_entry_for_key(key)) return NOT_FOUND?;
+	if (!map.remove_entry_for_key(key)) return NOT_FOUND~;
 }
 
 fn void LinkedHashMap.clear(&map)
@@ -375,17 +375,17 @@ fn bool LinkedHashMapIterator.next(&self)
 
 fn LinkedEntry*? LinkedHashMapIterator.get(&self)
 {
-	return self.current ? self.current : NOT_FOUND?;
+	return self.current ? self.current : NOT_FOUND~;
 }
 
 fn Value*? LinkedHashMapValueIterator.get(&self)
 {
-	return self.current ? &self.current.value : NOT_FOUND?;
+	return self.current ? &self.current.value : NOT_FOUND~;
 }
 
 fn Key*? LinkedHashMapKeyIterator.get(&self)
 {
-	return self.current ? &self.current.key : NOT_FOUND?;
+	return self.current ? &self.current.key : NOT_FOUND~;
 }
 
 fn bool LinkedHashMapIterator.has_next(&self)

lib/std/collections/linked_hashset.c3

@@ -1,7 +1,7 @@
 <*
  @require $defined((Value){}.hash()) : `No .hash function found on the value`
 *>
-module std::collections::set {Value};
+module std::collections::set <Value>;
 import std::math;
 import std::io @norecurse;
 
@@ -34,7 +34,7 @@ struct LinkedHashSet (Printable)
 	LinkedEntry* tail;
 }
 
-fn int LinkedHashSet.len(&self) @operator(len) => (int) self.count;
+fn usz LinkedHashSet.len(&self) @operator(len) => self.count;
 
 <*
  @param [&inout] allocator : "The allocator to use"
@@ -266,7 +266,7 @@ fn bool LinkedHashSet.contains(&set, Value value)
 *>
 fn void? LinkedHashSet.remove(&set, Value value) @maydiscard
 {
-	if (!set.remove_entry_for_value(value)) return NOT_FOUND?;
+	if (!set.remove_entry_for_value(value)) return NOT_FOUND~;
 }
 
 fn usz LinkedHashSet.remove_all(&set, Value[] values)
@@ -713,7 +713,7 @@ fn bool LinkedHashSetIterator.next(&self)
 
 fn Value*? LinkedHashSetIterator.get(&self)
 {
-	return self.current ? &self.current.value : NOT_FOUND?;
+	return self.current ? &self.current.value : NOT_FOUND~;
 }
 
 fn bool LinkedHashSetIterator.has_next(&self)

lib/std/collections/linkedlist.c3

@@ -1,7 +1,7 @@
 // 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};
+module std::collections::linkedlist <Type>;
 import std::io;
 
 const ELEMENT_IS_EQUATABLE = types::is_equatable_type(Type);
@@ -24,9 +24,9 @@ struct LinkedList
 fn usz? LinkedList.to_format(&self, Formatter* f) @dynamic
 {
 	usz len = f.print("{ ")!;
-	for (Node* node = self._first; node != null; node.next)
+	for (Node* node = self._first; node != null; node = node.next)
 	{
-		len += f.printf(node.next ? "%s, " : "s", node.value)!;
+		len += f.printf(node.next ? "%s, " : "%s", node.value)!;
 	}
 	return len + f.print(" }");
 }
@@ -124,13 +124,13 @@ fn Type? LinkedList.peek_last(&self) => self.last() @inline;
 
 fn Type? LinkedList.first(&self)
 {
-	if (!self._first) return NO_MORE_ELEMENT?;
+	if (!self._first) return NO_MORE_ELEMENT~;
 	return self._first.value;
 }
 
 fn Type? LinkedList.last(&self)
 {
-	if (!self._last) return NO_MORE_ELEMENT?;
+	if (!self._last) return NO_MORE_ELEMENT~;
 	return self._last.value;
 }
 
@@ -198,7 +198,7 @@ fn usz? LinkedList.index_of(&self, Type t) @if(ELEMENT_IS_EQUATABLE)
 	{
 		if (node.value == t) return i;
 	}
-	return NOT_FOUND?;
+	return NOT_FOUND~;
 }
 
 fn usz? LinkedList.rindex_of(&self, Type t) @if(ELEMENT_IS_EQUATABLE)
@@ -208,7 +208,7 @@ fn usz? LinkedList.rindex_of(&self, Type t) @if(ELEMENT_IS_EQUATABLE)
 		if (node.value == t) return i;
 		if (i == 0) break;
 	}
-	return NOT_FOUND?;
+	return NOT_FOUND~;
 }
 
 
@@ -296,7 +296,7 @@ fn usz LinkedList.remove(&self, Type t) @if(ELEMENT_IS_EQUATABLE)
 
 fn Type? LinkedList.pop(&self)
 {
-	if (!self._last) return NO_MORE_ELEMENT?;
+	if (!self._last) return NO_MORE_ELEMENT~;
 	defer self.unlink_last();
 	return self._last.value;
 }
@@ -308,20 +308,20 @@ fn bool LinkedList.is_empty(&self)
 
 fn Type? LinkedList.pop_front(&self)
 {
-	if (!self._first) return NO_MORE_ELEMENT?;
+	if (!self._first) return NO_MORE_ELEMENT~;
 	defer self.unlink_first();
 	return self._first.value;
 }
 
 fn void? LinkedList.remove_last(&self) @maydiscard
 {
-	if (!self._first) return NO_MORE_ELEMENT?;
+	if (!self._first) return NO_MORE_ELEMENT~;
 	self.unlink_last();
 }
 
 fn void? LinkedList.remove_first(&self) @maydiscard
 {
-	if (!self._first) return NO_MORE_ELEMENT?;
+	if (!self._first) return NO_MORE_ELEMENT~;
 	self.unlink_first();
 }
 

lib/std/collections/list.c3

@@ -1,7 +1,7 @@
 // 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};
+module std::collections::list <Type>;
 import std::io, std::math, std::collections::list_common;
 
 alias ElementPredicate = fn bool(Type *type);
@@ -115,7 +115,7 @@ fn void List.push(&self, Type element) @inline
 
 fn Type? List.pop(&self)
 {
-	if (!self.size) return NO_MORE_ELEMENT?;
+	if (!self.size) return NO_MORE_ELEMENT~;
 	defer self.set_size(self.size - 1);
 	return self.entries[self.size - 1];
 }
@@ -127,7 +127,7 @@ fn void List.clear(&self)
 
 fn Type? List.pop_first(&self)
 {
-	if (!self.size) return NO_MORE_ELEMENT?;
+	if (!self.size) return NO_MORE_ELEMENT~;
 	defer self.remove_at(0);
 	return self.entries[0];
 }
@@ -250,25 +250,25 @@ fn void List.set_at(&self, usz index, Type type)
 
 fn void? List.remove_last(&self) @maydiscard
 {
-	if (!self.size) return NO_MORE_ELEMENT?;
+	if (!self.size) return NO_MORE_ELEMENT~;
 	self.set_size(self.size - 1);
 }
 
 fn void? List.remove_first(&self) @maydiscard
 {
-	if (!self.size) return NO_MORE_ELEMENT?;
+	if (!self.size) return NO_MORE_ELEMENT~;
 	self.remove_at(0);
 }
 
 fn Type? List.first(&self)
 {
-	if (!self.size) return NO_MORE_ELEMENT?;
+	if (!self.size) return NO_MORE_ELEMENT~;
 	return self.entries[0];
 }
 
 fn Type? List.last(&self)
 {
-	if (!self.size) return NO_MORE_ELEMENT?;
+	if (!self.size) return NO_MORE_ELEMENT~;
 	return self.entries[self.size - 1];
 }
 
@@ -461,22 +461,22 @@ macro void List.post_alloc(&self) @private
 // Functions for equatable types
 
 
-fn usz? List.index_of(&self, Type type) @if(ELEMENT_IS_EQUATABLE)
+fn usz? List.index_of(&self, Type type) @if (ELEMENT_IS_EQUATABLE)
 {
 	foreach (i, v : self)
 	{
 		if (equals(v, type)) return i;
 	}
-	return NOT_FOUND?;
+	return NOT_FOUND~;
 }
 
-fn usz? List.rindex_of(&self, Type type) @if(ELEMENT_IS_EQUATABLE)
+fn usz? List.rindex_of(&self, Type type) @if (ELEMENT_IS_EQUATABLE)
 {
 	foreach_r (i, v : self)
 	{
 		if (equals(v, type)) return i;
 	}
-	return NOT_FOUND?;
+	return NOT_FOUND~;
 }
 
 fn bool List.equals(&self, List other_list) @if(ELEMENT_IS_EQUATABLE)

lib/std/collections/maybe.c3

@@ -1,4 +1,4 @@
-module std::collections::maybe{Type};
+module std::collections::maybe <Type>;
 import std::io;
 
 struct Maybe (Printable)
@@ -32,7 +32,7 @@ const Maybe EMPTY = { };
 
 macro Type? Maybe.get(self)
 {
-	return self.has_value ? self.value : NOT_FOUND?;
+	return self.has_value ? self.value : NOT_FOUND~;
 }
 
 fn bool Maybe.equals(self, Maybe other) @operator(==) @if(types::is_equatable_type(Type))

lib/std/collections/object.c3

@@ -242,8 +242,7 @@ macro Object* Object.push(&self, value)
 <*
  @require self.is_keyable()
 *>
-fn Object*? Object.get(&self, String key) => self.is_empty() ? NOT_FOUND? : self.map.get(key);
-
+fn Object*? Object.get(&self, String key) => self.is_empty() ? NOT_FOUND~ : self.map.get(key);
 
 fn bool Object.has_key(&self, String key) => self.is_map() && self.map.has_key(key);
 
@@ -308,7 +307,7 @@ macro get_integer_value(Object* value, $Type)
 			return ($Type)value.s.to_uint128();
 		$endif
 	}
-	if (!value.is_int()) return string::MALFORMED_INTEGER?;
+	if (!value.is_int()) return string::MALFORMED_INTEGER~;
 	return ($Type)value.i;
 }
 
@@ -361,7 +360,7 @@ fn uint128? Object.get_uint128_at(&self, usz index) => self.get_integer_at(uint1
 fn String? Object.get_string(&self, String key)
 {
 	Object* value = self.get(key)!;
-	if (!value.is_string()) return TYPE_MISMATCH?;
+	if (!value.is_string()) return TYPE_MISMATCH~;
 	return value.s;
 }
 
@@ -371,7 +370,7 @@ fn String? Object.get_string(&self, String key)
 fn String? Object.get_string_at(&self, usz index)
 {
 	Object* value = self.get_at(index);
-	if (!value.is_string()) return TYPE_MISMATCH?;
+	if (!value.is_string()) return TYPE_MISMATCH~;
 	return value.s;
 }
 
@@ -381,7 +380,7 @@ fn String? Object.get_string_at(&self, usz index)
 macro String? Object.get_enum(&self, $EnumType, String key)
 {
 	Object value = self.get(key)!;
-	if ($EnumType.typeid != value.type) return TYPE_MISMATCH?;
+	if ($EnumType.typeid != value.type) return TYPE_MISMATCH~;
 	return ($EnumType)value.i;
 }
 
@@ -391,7 +390,7 @@ macro String? Object.get_enum(&self, $EnumType, String key)
 macro String? Object.get_enum_at(&self, $EnumType, usz index)
 {
 	Object value = self.get_at(index);
-	if ($EnumType.typeid != value.type) return TYPE_MISMATCH?;
+	if ($EnumType.typeid != value.type) return TYPE_MISMATCH~;
 	return ($EnumType)value.i;
 }
 
@@ -401,7 +400,7 @@ macro String? Object.get_enum_at(&self, $EnumType, usz index)
 fn bool? Object.get_bool(&self, String key)
 {
 	Object* value = self.get(key)!;
-	if (!value.is_bool()) return TYPE_MISMATCH?;
+	if (!value.is_bool()) return TYPE_MISMATCH~;
 	return value.b;
 }
 
@@ -412,7 +411,7 @@ fn bool? Object.get_bool(&self, String key)
 fn bool? Object.get_bool_at(&self, usz index)
 {
 	Object* value = self.get_at(index);
-	if (!value.is_bool()) return TYPE_MISMATCH?;
+	if (!value.is_bool()) return TYPE_MISMATCH~;
 	return value.b;
 }
 
@@ -431,7 +430,7 @@ fn double? Object.get_float(&self, String key)
 		case FLOAT:
 			return value.f;
 		default:
-			return TYPE_MISMATCH?;
+			return TYPE_MISMATCH~;
 	}
 }
 
@@ -450,7 +449,7 @@ fn double? Object.get_float_at(&self, usz index)
 		case FLOAT:
 			return value.f;
 		default:
-			return TYPE_MISMATCH?;
+			return TYPE_MISMATCH~;
 	}
 }
 

lib/std/collections/priorityqueue.c3

@@ -20,16 +20,13 @@
 //  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;
-
-typedef PriorityQueue = inline PrivatePriorityQueue{Type, false};
-typedef PriorityQueueMax = inline PrivatePriorityQueue{Type, true};
-
-module std::collections::priorityqueue::private{Type, MAX};
+module std::collections::priorityqueue;
 import std::collections::list, std::io;
 
-struct PrivatePriorityQueue (Printable)
+typedef PriorityQueue <Type> = inline PrivatePriorityQueue{Type, false};
+typedef PriorityQueueMax <Type> = inline PrivatePriorityQueue{Type, true};
+
+struct PrivatePriorityQueue (Printable) <Type, MAX>
 {
 	List{Type} heap;
 }
@@ -86,7 +83,7 @@ fn Type? PrivatePriorityQueue.pop(&self)
 {
 	usz i = 0;
 	usz len = self.heap.len();
-	if (!len) return NO_MORE_ELEMENT?;
+	if (!len) return NO_MORE_ELEMENT~;
 	usz new_count = len - 1;
 	self.heap.swap(0, new_count);
 	while OUTER: ((2 * i + 1) < new_count)

lib/std/collections/range.c3

@@ -1,7 +1,7 @@
 <*
  @require Type.is_ordered : "The type must be ordered"
 *>
-module std::collections::range{Type};
+module std::collections::range <Type>;
 import std::io;
 
 struct Range (Printable)

lib/std/collections/ringbuffer.c3

@@ -1,7 +1,7 @@
 <*
  @require Type.kindof == ARRAY : "Required an array type"
 *>
-module std::collections::ringbuffer{Type};
+module std::collections::ringbuffer <Type>;
 import std::io;
 
 alias Element = $typeof((Type){}[0]);
@@ -48,7 +48,7 @@ fn Element? RingBuffer.pop(&self)
 	switch
 	{
 		case self.written == 0:
-			return NO_MORE_ELEMENT?;
+			return NO_MORE_ELEMENT~;
 		case self.written < self.buf.len:
 			self.written--;
 			return self.buf[self.written];

lib/std/collections/tuple.c3

@@ -1,4 +1,4 @@
-module std::collections::pair{Type1, Type2};
+module std::collections::pair <Type1, Type2>;
 import std::io;
 
 struct Pair (Printable)
@@ -29,9 +29,7 @@ fn bool Pair.equal(self, Pair other) @operator(==) @if (types::has_equals(Type1)
 	return self.first == other.first && self.second == other.second;
 }
 
-
-
-module std::collections::triple{Type1, Type2, Type3};
+module std::collections::triple <Type1, Type2, Type3>;
 import std::io;
 
 struct Triple (Printable)
@@ -65,11 +63,10 @@ fn bool Triple.equal(self, Triple other) @operator(==) @if (types::has_equals(Ty
 	return self.first == other.first && self.second == other.second && self.third == other.third;
 }
 
-
-module std::collections::tuple{Type1, Type2};
+module std::collections::tuple <Type1, Type2>;
 
 struct Tuple @deprecated("Use 'Pair' instead")
 {
 	Type1 first;
 	Type2 second;
-}
+}

lib/std/compression/qoi.c3

@@ -100,7 +100,7 @@ fn usz? write(String filename, char[] input, QOIDesc* desc) => @pool()
 fn char[]? read(Allocator allocator, String filename, QOIDesc* desc, QOIChannels channels = AUTO) => @pool()
 {
 	// read file
-	char[] data = file::load_temp(filename) ?? FILE_OPEN_FAILED?!;
+	char[] data = file::load_temp(filename) ?? FILE_OPEN_FAILED~!;
 	// pass data to decode function
 	return decode(allocator, data, desc, channels);
 }
@@ -128,14 +128,14 @@ import std::bits;
 fn char[]? encode(Allocator allocator, char[] input, QOIDesc* desc) @nodiscard
 {
 	// check info in desc
-	if (desc.width == 0 || desc.height == 0) return INVALID_PARAMETERS?;
-	if (desc.channels == AUTO) return INVALID_PARAMETERS?;
+	if (desc.width == 0 || desc.height == 0) return INVALID_PARAMETERS~;
+	if (desc.channels == AUTO) return INVALID_PARAMETERS~;
 	uint pixels = desc.width * desc.height;
-	if (pixels > PIXELS_MAX) return TOO_MANY_PIXELS?;
+	if (pixels > PIXELS_MAX) return TOO_MANY_PIXELS~;
 	
 	// check input data size
 	uint image_size = pixels * desc.channels;
-	if (image_size != input.len) return INVALID_DATA?;
+	if (image_size != input.len) return INVALID_DATA~;
 
 	// allocate memory for encoded data (output)
 	// header + chunk tag and RGB(A) data for each pixel + end of stream
@@ -283,27 +283,27 @@ fn char[]? encode(Allocator allocator, char[] input, QOIDesc* desc) @nodiscard
 fn char[]? decode(Allocator allocator, char[] data, QOIDesc* desc, QOIChannels channels = AUTO) @nodiscard
 {
 	// check input data
-	if (data.len < Header.sizeof + END_OF_STREAM.len) return INVALID_DATA?;
+	if (data.len < Header.sizeof + END_OF_STREAM.len) return INVALID_DATA~;
 
 	// get header
 	Header* header = (Header*)data.ptr;
 
 	// check magic bytes (FourCC)
-	if (bswap(header.be_magic) != 'qoif') return INVALID_DATA?;
+	if (bswap(header.be_magic) != 'qoif') return INVALID_DATA~;
 
 	// copy header data to desc
-	desc.width = bswap(header.be_width);
-	desc.height = bswap(header.be_height);
-	desc.channels = header.channels;
+	uint width = desc.width = bswap(header.be_width);
+	uint height = desc.height = bswap(header.be_height);
+	QOIChannels desc_channels = desc.channels = header.channels;
 	desc.colorspace = header.colorspace;
-	if (desc.channels == AUTO) return INVALID_DATA?; // Channels must be specified in the header
+	if (desc_channels == AUTO) return INVALID_DATA~; // Channels must be specified in the header
 
 	// check width and height
-	if (desc.width == 0 || desc.height == 0) return INVALID_DATA?;
+	if (width == 0 || height == 0) return INVALID_DATA~;
 
 	// check pixel count
-	ulong pixels = (ulong)desc.width * (ulong)desc.height;
-	if (pixels > PIXELS_MAX) return TOO_MANY_PIXELS?;
+	ulong pixels = (ulong)width * (ulong)height;
+	if (pixels > PIXELS_MAX) return TOO_MANY_PIXELS~;
 
 	uint pos = Header.sizeof; 	// Current position in data
 	uint loc;					// Current position in image (top-left corner)
@@ -313,7 +313,7 @@ fn char[]? decode(Allocator allocator, char[] data, QOIDesc* desc, QOIChannels c
 	Pixel[64] palette; // Zero-initialized by default
 	Pixel p = { 0, 0, 0, 255 };
 
-	if (channels == AUTO) channels = desc.channels;
+	if (channels == AUTO) channels = desc_channels;
 	
 	// allocate memory for image data
 	usz image_size = (usz)pixels * channels;

lib/std/core/allocators/arena_allocator.c3

@@ -90,12 +90,12 @@ fn void*? ArenaAllocator.acquire(&self, usz size, AllocInitType init_type, usz a
 {
 	alignment = alignment_for_allocation(alignment);
 	usz total_len = self.data.len;
-	if (size > total_len) return mem::INVALID_ALLOC_SIZE?;
+	if (size > total_len) return mem::INVALID_ALLOC_SIZE~;
 	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 mem::OUT_OF_MEMORY?;
+	if (end > total_len) return mem::OUT_OF_MEMORY~;
 	self.used = end;
 	ArenaAllocatorHeader* header = mem - ArenaAllocatorHeader.sizeof;
 	header.size = size;
@@ -117,7 +117,7 @@ fn void*? ArenaAllocator.resize(&self, void *old_pointer, usz size, usz alignmen
 	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 mem::INVALID_ALLOC_SIZE?;
+	if (size > total_len) return mem::INVALID_ALLOC_SIZE~;
 	ArenaAllocatorHeader* header = old_pointer - ArenaAllocatorHeader.sizeof;
 	usz old_size = header.size;
 	// Do last allocation and alignment match?
@@ -130,7 +130,7 @@ fn void*? ArenaAllocator.resize(&self, void *old_pointer, usz size, usz alignmen
 		else
 		{
 			usz new_used = self.used + size - old_size;
-			if (new_used > total_len) return mem::OUT_OF_MEMORY?;
+			if (new_used > total_len) return mem::OUT_OF_MEMORY~;
 			self.used = new_used;
 		}
 		header.size = size;

lib/std/core/allocators/dynamic_arena.c3

@@ -153,7 +153,7 @@ fn void*? DynamicArenaAllocator._alloc_new(&self, usz size, usz alignment) @loca
 	if (catch err = page)
 	{
 		allocator::free(self.backing_allocator, mem);
-		return err?;
+		return err~;
 	}
 	page.memory = mem;
 	void* mem_start = mem::aligned_pointer(mem + DynamicArenaChunk.sizeof, alignment);

lib/std/core/allocators/libc_allocator.c3

@@ -23,22 +23,22 @@ fn void*? LibcAllocator.acquire(&self, usz bytes, AllocInitType init_type, usz a
 		void* data @noinit;
 		if (alignment > mem::DEFAULT_MEM_ALIGNMENT)
 		{
-			if (posix::posix_memalign(&data, alignment, bytes)) return mem::OUT_OF_MEMORY?;
+			if (posix::posix_memalign(&data, alignment, bytes)) return mem::OUT_OF_MEMORY~;
 			mem::clear(data, bytes, mem::DEFAULT_MEM_ALIGNMENT);
 			return data;
 		}
-		return libc::calloc(1, bytes) ?: mem::OUT_OF_MEMORY?;
+		return libc::calloc(1, bytes) ?: mem::OUT_OF_MEMORY~;
 	}
 	else
 	{
 		void* data @noinit;
 		if (alignment > mem::DEFAULT_MEM_ALIGNMENT)
 		{
-			if (posix::posix_memalign(&data, alignment, bytes)) return mem::OUT_OF_MEMORY?;
+			if (posix::posix_memalign(&data, alignment, bytes)) return mem::OUT_OF_MEMORY~;
 		}
 		else
 		{
-			if (!(data = libc::malloc(bytes))) return mem::OUT_OF_MEMORY?;
+			if (!(data = libc::malloc(bytes))) return mem::OUT_OF_MEMORY~;
 		}
 		$if env::TESTING:
 			for (usz i = 0; i < bytes; i++) ((char*)data)[i] = 0xAA;
@@ -49,9 +49,9 @@ fn void*? LibcAllocator.acquire(&self, usz bytes, AllocInitType init_type, usz a
 
 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) ?: mem::OUT_OF_MEMORY?;
+	if (alignment <= mem::DEFAULT_MEM_ALIGNMENT) return libc::realloc(old_ptr, new_bytes) ?: mem::OUT_OF_MEMORY~;
 	void* new_ptr;
-	if (posix::posix_memalign(&new_ptr, alignment, new_bytes)) return mem::OUT_OF_MEMORY?;
+	if (posix::posix_memalign(&new_ptr, alignment, new_bytes)) return mem::OUT_OF_MEMORY~;
 
 	$switch:
 	$case env::DARWIN:
@@ -83,12 +83,12 @@ fn void*? LibcAllocator.acquire(&self, usz bytes, AllocInitType init_type, usz a
 	{
 		if (alignment > 0)
 		{
-			return win32::_aligned_recalloc(null, 1, bytes, alignment) ?: mem::OUT_OF_MEMORY?;
+			return win32::_aligned_recalloc(null, 1, bytes, alignment) ?: mem::OUT_OF_MEMORY~;
 		}
-		return libc::calloc(1, bytes) ?: mem::OUT_OF_MEMORY?;
+		return libc::calloc(1, bytes) ?: mem::OUT_OF_MEMORY~;
 	}
 	void* data = alignment > 0 ? win32::_aligned_malloc(bytes, alignment) : libc::malloc(bytes);
-	if (!data) return mem::OUT_OF_MEMORY?;
+	if (!data) return mem::OUT_OF_MEMORY~;
 	$if env::TESTING:
 		for (usz i = 0; i < bytes; i++) ((char*)data)[i] = 0xAA;
 	$endif
@@ -99,9 +99,9 @@ fn void*? LibcAllocator.resize(&self, void* old_ptr, usz new_bytes, usz alignmen
 {
 	if (alignment)
 	{
-		return win32::_aligned_realloc(old_ptr, new_bytes, alignment) ?: mem::OUT_OF_MEMORY?;
+		return win32::_aligned_realloc(old_ptr, new_bytes, alignment) ?: mem::OUT_OF_MEMORY~;
 	}
-	return libc::realloc(old_ptr, new_bytes) ?: mem::OUT_OF_MEMORY?;
+	return libc::realloc(old_ptr, new_bytes) ?: mem::OUT_OF_MEMORY~;
 }
 
 fn void LibcAllocator.release(&self, void* old_ptr, bool aligned) @dynamic
@@ -122,12 +122,12 @@ fn void*? LibcAllocator.acquire(&self, usz bytes, AllocInitType init_type, usz a
 	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 ?: mem::OUT_OF_MEMORY?;
+		return data ?: mem::OUT_OF_MEMORY~;
 	}
 	else
 	{
 		void* data = alignment ? @aligned_alloc(libc::malloc, bytes, alignment)!! : libc::malloc(bytes);
-		if (!data) return mem::OUT_OF_MEMORY?;
+		if (!data) return mem::OUT_OF_MEMORY~;
 		$if env::TESTING:
 			for (usz i = 0; i < bytes; i++) ((char*)data)[i] = 0xAA;
 		$endif
@@ -141,9 +141,9 @@ fn void*? LibcAllocator.resize(&self, void* old_ptr, usz new_bytes, usz alignmen
 	if (alignment)
 	{
 		void* data = @aligned_realloc(fn void*(usz bytes) => libc::malloc(bytes), libc::free, old_ptr, new_bytes, alignment)!!;
-		return data ?: mem::OUT_OF_MEMORY?;
+		return data ?: mem::OUT_OF_MEMORY~;
 	}
-	return libc::realloc(old_ptr, new_bytes) ?: mem::OUT_OF_MEMORY?;
+	return libc::realloc(old_ptr, new_bytes) ?: mem::OUT_OF_MEMORY~;
 }
 
 

lib/std/core/allocators/tracking_allocator.c3

@@ -28,6 +28,8 @@ struct TrackingAllocator (Allocator)
 	AllocMap map;
 	usz mem_total;
 	usz allocs_total;
+	usz usage;
+	usz max_usage;
 }
 
 <*
@@ -70,6 +72,11 @@ fn usz TrackingAllocator.total_allocated(&self) => self.mem_total;
 *>
 fn usz TrackingAllocator.total_allocation_count(&self) => self.allocs_total;
 
+<*
+ @return "the maximum amount of memory allocated"
+*>
+fn usz TrackingAllocator.max_allocated(&self) => self.max_usage;
+
 fn Allocation[] TrackingAllocator.allocations_tlist(&self, Allocator allocator)
 {
 	return self.map.tvalues();
@@ -88,27 +95,34 @@ fn void*? TrackingAllocator.acquire(&self, usz size, AllocInitType init_type, us
 	backtrace::capture_current(&bt);
 	self.map.set((uptr)data, { data, size, bt });
 	self.mem_total += size;
+	self.usage += size;
+	if (self.usage > self.max_usage) self.max_usage = self.usage;
 	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.usage -= self.map[(uptr)old_pointer]!!.size;
 	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.usage += size;
+	if (self.usage > self.max_usage) self.max_usage = self.usage;
 	self.allocs_total++;
 	return data;
 }
 
 fn void TrackingAllocator.release(&self, void* old_pointer, bool is_aligned) @dynamic
 {
+	usz? old_size = self.map[(uptr)old_pointer].size;
 	if (catch self.map.remove((uptr)old_pointer))
 	{
 		unreachable("Attempt to release untracked pointer %p, this is likely a bug.", old_pointer);
 	}
+	self.usage -= old_size!!;
 	self.inner_allocator.release(old_pointer, is_aligned);
 }
 
@@ -177,6 +191,7 @@ fn void? TrackingAllocator.fprint_report(&self, OutStream out) => @pool()
 	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)!;
+	io::fprintfn(out, "- Maximum memory usage:                        %d", self.max_usage)!;
 	if (leaks)
 	{
 		io::fprintn(out)!;

lib/std/core/allocators/vmem.c3

@@ -45,7 +45,7 @@ fn void? Vmem.init(&self, usz preferred_size, usz reserve_page_size = 0, VmemOpt
 	if (page_size < reserve_page_size) page_size = reserve_page_size;
 	preferred_size = mem::aligned_offset(preferred_size, page_size);
 	if (!min_size) min_size = max(preferred_size / 1024, 1);
-	VirtualMemory? memory = mem::OUT_OF_MEMORY?;
+	VirtualMemory? memory = mem::OUT_OF_MEMORY~;
 	while (preferred_size >= min_size)
 	{
 		memory = vm::virtual_alloc(preferred_size, PROTECTED);
@@ -62,7 +62,7 @@ fn void? Vmem.init(&self, usz preferred_size, usz reserve_page_size = 0, VmemOpt
 				break;
 		}
 	}
-	if (catch memory) return VMEM_RESERVE_FAILED?;
+	if (catch memory) return VMEM_RESERVE_FAILED~;
 	if (page_size > preferred_size) page_size = preferred_size;
 	$if env::ADDRESS_SANITIZER:
 		asan::poison_memory_region(memory.ptr, memory.size);
@@ -87,12 +87,12 @@ fn void*? Vmem.acquire(&self, usz size, AllocInitType init_type, usz alignment)
 {
 	alignment = alignment_for_allocation(alignment);
 	usz total_len = self.memory.size;
-	if (size > total_len) return mem::INVALID_ALLOC_SIZE?;
+	if (size > total_len) return mem::INVALID_ALLOC_SIZE~;
 	void* start_mem = self.memory.ptr;
 	void* unaligned_pointer_to_offset = start_mem + self.allocated + VmemHeader.sizeof;
 	void* mem = mem::aligned_pointer(unaligned_pointer_to_offset, alignment);
 	usz after = (usz)(mem - start_mem) + size;
-	if (after > total_len) return mem::OUT_OF_MEMORY?;
+	if (after > total_len) return mem::OUT_OF_MEMORY~;
 	if (init_type == ZERO && self.high_water <= self.allocated)
 	{
 		init_type = NO_ZERO;
@@ -119,7 +119,7 @@ fn bool Vmem.owns_pointer(&self, void* ptr) @inline
 *>
 fn void*? Vmem.resize(&self, void *old_pointer, usz size, usz alignment) @dynamic
 {
-	if (size > self.memory.size) return mem::INVALID_ALLOC_SIZE?;
+	if (size > self.memory.size) return mem::INVALID_ALLOC_SIZE~;
 	alignment = alignment_for_allocation(alignment);
 	assert(self.owns_pointer(old_pointer), "Pointer originates from a different allocator: %p, not in %p - %p", old_pointer, self.memory.ptr, self.memory.ptr + self.allocated);
 	VmemHeader* header = old_pointer - VmemHeader.sizeof;
@@ -135,7 +135,7 @@ fn void*? Vmem.resize(&self, void *old_pointer, usz size, usz alignment) @dynami
     	else
     	{
     		usz allocated = self.allocated + size - old_size;
-    		if (allocated > self.memory.size) return mem::OUT_OF_MEMORY?;
+    		if (allocated > self.memory.size) return mem::OUT_OF_MEMORY~;
     		protect(self, allocated)!;
     	}
     	header.size = size;

lib/std/core/ansi.c3

@@ -1,4 +1,5 @@
 module std::core::string::ansi;
+import std::io;
 
 enum Ansi : const inline String
 {
@@ -56,6 +57,37 @@ enum Ansi : const inline String
 	BG_BRIGHT_WHITE    = "\e[107m",
 }
 
+struct AnsiColor (Printable)
+{
+	char r, g, b;
+	bool bg;
+}
+
+fn usz? AnsiColor.to_format(&self, Formatter* fmt) @dynamic
+{
+	return fmt.printf("\e[%s8;2;%s;%s;%sm", self.bg ? 4 : 3, self.r, self.g, self.b);
+}
+
+<*
+ 24-bit color code
+
+ @return `A struct that, when formatted with '%s', sets the foreground or background colour to the specified rgb value`
+*>
+fn AnsiColor get_color_rgb(char r, char g, char b, bool bg = false)
+{
+	return {r, g, b, bg};
+}
+
+<*
+ 24-bit color code
+
+ @return `A struct that, when formatted with '%s', sets the foreground or background colour of printed text to the specified rgb value`
+*>
+fn AnsiColor get_color(uint rgb, bool bg = false)
+{
+	return {(char)(rgb >> 16), (char)((rgb & 0x00FF00) >> 8), (char)rgb, bg};
+}
+
 <*
  8-bit color code
 
@@ -96,7 +128,7 @@ macro String color(uint $rgb, bool $bg = false) @const
  @require rgb <= 0xFF_FF_FF : `Expected a 24 bit RGB value`
  @return `the string char for the given foreground color`
 *>
-fn String make_color(Allocator mem, uint rgb, bool bg = false)
+fn String make_color(Allocator mem, uint rgb, bool bg = false) @deprecated("use get_color instead")
 {
 	return make_color_rgb(mem, (char)(rgb >> 16), (char)((rgb & 0xFF00) >> 8), (char)rgb, bg);
 }
@@ -107,7 +139,7 @@ fn String make_color(Allocator mem, uint rgb, bool bg = false)
  @require rgb <= 0xFF_FF_FF : `Expected a 24 bit RGB value`
  @return `the string char for the given foreground color`
 *>
-fn String make_tcolor(uint rgb, bool bg = false)
+fn String make_tcolor(uint rgb, bool bg = false) @deprecated("use get_color instead")
 {
 	return make_color_rgb(tmem, (char)(rgb >> 16), (char)((rgb & 0xFF00) >> 8), (char)rgb, bg);
 }
@@ -117,7 +149,7 @@ fn String make_tcolor(uint rgb, bool bg = false)
 
  @return `the string char for the given foreground color`
 *>
-fn String make_color_rgb(Allocator mem, char r, char g, char b, bool bg = false)
+fn String make_color_rgb(Allocator mem, char r, char g, char b, bool bg = false) @deprecated("use get_color_rgb instead")
 {
 	return string::format(mem, "\e[%s8;2;%s;%s;%sm", bg ? 4 : 3, r, g, b);
 }
@@ -127,7 +159,7 @@ fn String make_color_rgb(Allocator mem, char r, char g, char b, bool bg = false)
 
  @return `the string char for the given foreground color`
 *>
-fn String make_tcolor_rgb(char r, char g, char b, bool bg = false)
+fn String make_tcolor_rgb(char r, char g, char b, bool bg = false) @deprecated("use get_color_rgb instead")
 {
 	return string::format(tmem, "\e[%s8;2;%s;%s;%sm", bg ? 4 : 3, r, g, b);
 }

lib/std/core/array.c3

@@ -35,7 +35,7 @@ macro usz? index_of(array, element)
 	{
 		if (*e == element) return i;
 	}
-	return NOT_FOUND?;
+	return NOT_FOUND~;
 }
 
 
@@ -75,7 +75,7 @@ macro usz? rindex_of(array, element)
 	{
 		if (*e == element) return i;
 	}
-	return NOT_FOUND?;
+	return NOT_FOUND~;
 }
 
 

lib/std/core/ascii.c3

@@ -130,9 +130,27 @@ fn AsciiCharset create_set(String string)
 	return set;
 }
 
+macro bool AsciiCharset.@contains($set, char $c) @const => !!($c < 128) & !!($set & (AsciiCharset)(1ULL << $c));
+
+macro AsciiCharset @combine_sets(AsciiCharset $first, AsciiCharset... $sets) @const
+{
+	var $res = $first;
+	$foreach $c : $sets:
+		$res |= $c;
+	$endforeach
+	return $res;
+}
+fn AsciiCharset combine_sets(AsciiCharset first, AsciiCharset... sets)
+{
+	foreach (c : sets) first |= c;
+	return first;
+}
+
 macro bool AsciiCharset.contains(set, char c) => !!(c < 128) & !!(set & (AsciiCharset)(1ULL << c));
 
 const AsciiCharset WHITESPACE_SET = @create_set("\t\n\v\f\r ");
 const AsciiCharset NUMBER_SET = @create_set("0123456789");
-
-
+const AsciiCharset ALPHA_UPPER_SET = @create_set("ABCDEFGHIJKLMNOPQRSTUVWXYZ");
+const AsciiCharset ALPHA_LOWER_SET = @create_set("abcdefghijklmnopqrstuvwxyz");
+const AsciiCharset ALPHA_SET = @combine_sets(ALPHA_UPPER_SET, ALPHA_LOWER_SET);
+const AsciiCharset ALPHANUMERIC_SET = @combine_sets(ALPHA_SET, NUMBER_SET);

lib/std/core/bitorder.c3

@@ -90,33 +90,35 @@ bitstruct UInt128LE : uint128 @littleendian
 <*
  @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"
+ @require $defined(*bytes) ||| $defined(bytes[:$Type.sizeof]) : "Data is too short to contain value"
 *>
 macro read(bytes, $Type)
 {
-	char[] s;
+	char *ptr;
 	$switch $kindof(bytes):
 		$case POINTER:
-			s = (*bytes)[:$Type.sizeof];
+			ptr = bytes;
 		$default:
-			s = bytes[:$Type.sizeof];
+			ptr = bytes[..].ptr;
 	$endswitch
-	return bitcast(*(char[$Type.sizeof]*)s.ptr, $Type).val;
+	return bitcast(mem::load((char[$Type.sizeof]*)ptr, $align: 1), $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"
+ @require $defined(*bytes) ||| $defined(bytes[:$Type.sizeof]) : "Data is not sufficent to hold value"
 *>
 macro write(x, bytes, $Type)
 {
-	char[] s;
+	char *ptr;
 	$switch $kindof(bytes):
 		$case POINTER:
-			s = (*bytes)[:$Type.sizeof];
+			ptr = bytes;
 		$default:
-			s = bytes[:$Type.sizeof];
+			ptr = bytes[..].ptr;
 	$endswitch
-	*($typeof(x)*)s.ptr = bitcast(x, $Type).val;
+	mem::store(($typeof(x)*)ptr, bitcast(x, $Type).val, 1);
 }
 
 macro bool is_bitorder($Type)

lib/std/core/builtin.c3

@@ -24,11 +24,15 @@ macro foo(a, #b = EMPTY_MACRO_SLOT)
 	$endif
 }
 *>
-const EmptySlot EMPTY_MACRO_SLOT @builtin = null;
+const EmptySlot EMPTY_MACRO_SLOT @builtin @deprecated("Use `#arg = ...` instead.") = null;
 
 typedef EmptySlot = void*;
-macro bool @is_empty_macro_slot(#arg) @const @builtin => $typeof(#arg) == EmptySlot;
-macro bool @is_valid_macro_slot(#arg) @const @builtin => $typeof(#arg) != EmptySlot;
+macro bool @is_empty_macro_slot(#arg) @const @builtin
+		@deprecated("Use `#arg = ...` to define an optional macro slot, and `$defined(#arg)` to detect whether the argument is set.")
+	=> $typeof(#arg) == EmptySlot;
+macro bool @is_valid_macro_slot(#arg) @const @builtin
+		@deprecated("Use `#arg = ...` to define an optional macro slot, and `$defined(#arg)` to detect whether the argument is set.")
+	=> $typeof(#arg) != EmptySlot;
 
 <*
  Returns a random value at compile time.
@@ -93,6 +97,21 @@ macro usz bitsizeof($Type) @builtin @const => $Type.sizeof * 8u;
 
 macro usz @bitsizeof(#expr) @builtin @const => $sizeof(#expr) * 8u;
 
+<*
+ Compile-time check for whether a set of constants contains a certain expression.
+
+ @param #needle : "The expression whose value should be located."
+*>
+macro bool @in(#needle, ...) @builtin @const
+{
+    $for var $x = 0; $x < $vacount; $x++:
+        $assert $defined(#needle == $vaconst[$x])
+            : "Index %s: types '%s' (needle) and '%s' are not equatable", $x, $typeof(#needle), $typeof($vaconst[$x]);
+        $if #needle == $vaconst[$x]: return true; $endif
+    $endfor
+    return false;
+}
+
 <*
  Convert an `any` type to a type, returning an failure if there is a type mismatch.
 
@@ -104,10 +123,29 @@ macro usz @bitsizeof(#expr) @builtin @const => $sizeof(#expr) * 8u;
 *>
 macro anycast(any v, $Type) @builtin
 {
-	if (v.type != $Type.typeid) return TYPE_MISMATCH?;
+	if (v.type != $Type.typeid) return TYPE_MISMATCH~;
 	return ($Type*)v.ptr;
 }
 
+<*
+  @return "The value in the pointer"
+  @return? TYPE_MISMATCH
+*>
+macro any.to(self, $Type)
+{
+	if (self.type != $Type.typeid) return TYPE_MISMATCH~;
+	return *($Type*)self.ptr;
+}
+
+<*
+ @require self.type == $Type : "The 'any' contained an unexpected type."
+ @return "The value in the pointer"
+*>
+macro any.as(self, $Type)
+{
+	return *($Type*)self.ptr;
+}
+
 macro bool @assignable_to(#foo, $Type) @const @builtin @deprecated("use '$defined($Type x = #foo)'") => $defined(*&&($Type){} = #foo);
 
 macro @addr(#val) @builtin
@@ -143,7 +181,7 @@ fn bool print_backtrace(String message, int backtraces_to_ignore, void *added_ba
 	{
 		backtraces[++backtraces_to_ignore] = added_backtrace;
 	}
-	@stack_mem(2048; Allocator mem)
+	@stack_mem(4096; Allocator mem)
 	{
 		BacktraceList? backtrace = backtrace::symbolize_backtrace(mem, backtraces);
 		if (catch backtrace) return false;
@@ -174,6 +212,7 @@ fn bool print_backtrace(String message, int backtraces_to_ignore, void *added_ba
 
 fn void default_panic(String message, String file, String function, uint line) @if(env::NATIVE_STACKTRACE)
 {
+	in_panic = true;
 	$if $defined(io::stderr) && env::PANIC_MSG:
 		if (!print_backtrace(message, 2))
 		{
@@ -190,7 +229,7 @@ macro void abort(String string = "Unrecoverable error reached", ...) @format(0)
 	$$trap();
 }
 
-bool in_panic @local = false;
+bool in_panic @private = false;
 
 fn void default_panic(String message, String file, String function, uint line) @if (!env::NATIVE_STACKTRACE)
 {
@@ -315,7 +354,7 @@ macro enum_by_name($Type, String enum_name) @builtin
 	{
 		if (name == enum_name) return $Type.from_ordinal(i);
 	}
-	return NOT_FOUND?;
+	return NOT_FOUND~;
 }
 
 <*
@@ -332,7 +371,7 @@ macro @enum_from_value($Type, #value, value) @builtin @deprecated("Use Enum.look
 	{
 		if (e.#value == value) return e;
 	}
-	return NOT_FOUND?;
+	return NOT_FOUND~;
 }
 
 <*
@@ -416,11 +455,25 @@ macro @prefetch(void* ptr, PrefetchLocality $locality = VERY_NEAR, bool $write =
 	$endif
 }
 
+<*
+ Shuffle a vector by its index
+
+ int[<4>] a = { 1, 2, 3, 4 };
+ assert(swizzle(a, 0, 1, 1, 3) == (int[<4>]) { 1, 2, 2, 4 });
+*>
 macro swizzle(v, ...) @builtin
 {
 	return $$swizzle(v, $vasplat);
 }
 
+
+<*
+ Shuffle two vectors by a common index from arranging the vectors sequentially in memory
+
+ int[<4>] a = { 1, 2, 3, 4 };
+ int[<4>] b = { 100, 1000, 10000, 100000 };
+ assert(swizzle2(a, b, 0, 1, 4, 6, 2) == (int[<5>]) { 1, 2, 100, 10000, 3 });
+*>
 macro swizzle2(v, v2, ...) @builtin
 {
 	return $$swizzle2(v, v2, $vasplat);
@@ -488,7 +541,7 @@ macro bool @ok(#expr) @builtin
 macro void? @try(#v, #expr) @builtin @maydiscard
 {
 	var res = #expr;
-	if (catch err = res) return err?;
+	if (catch err = res) return err~;
 	#v = res;
 }
 
@@ -532,7 +585,7 @@ macro bool? @try_catch(#v, #expr, fault expected_fault) @builtin
 	var res = #expr;
 	if (catch err = res)
 	{
-		return err == expected_fault ? true : err?;
+		return err == expected_fault ? true : err~;
 	}
 	#v = res;
 	return false;
@@ -651,7 +704,7 @@ macro uint hash_array(array_ptr) @local
 *>
 macro uint hash_vec(vec) @local
 {
-	var $len = $sizeof(vec.len * $typeof(vec).inner.sizeof);
+	var $len = $sizeof(vec);
 
 	$if $len > 16:
 		return (uint)komi::hash(((char*)&&vec)[:$len]);
@@ -963,7 +1016,7 @@ fn void sig_bus_error(CInt i, void* info, void* context)
 			}
 		$endif
 	$endif
-	$$trap();
+	os::fastexit(128 + i);
 }
 
 fn void sig_segmentation_fault(CInt i, void* p1, void* context)
@@ -978,15 +1031,39 @@ fn void sig_segmentation_fault(CInt i, void* p1, void* context)
 			}
 		$endif
 	$endif
-	$$trap();
+	os::fastexit(128 + i);
 }
 
+fn void sig_illegal_instruction(CInt i, void* p1, void* context)
+{
+	if (in_panic) os::fastexit(128 + i);
+	$if !env::NATIVE_STACKTRACE:
+		sig_panic("Illegal instruction.");
+	$else
+		$if $defined(io::stderr):
+			if (!print_backtrace("Illegal instruction.", 2, posix::stack_instruction(context)))
+			{
+				io::eprintn("\nERROR: Illegal instruction.");
+			}
+		$endif
+	$endif
+	os::fastexit(128 + i);
+}
 
+char[64 * 1024] sig_stack @local @if(env::BACKTRACE && env::LINUX);
 
 // Clean this up
 fn void install_signal_handlers() @init(101) @local @if(env::BACKTRACE)
 {
+	$if env::LINUX:
+		Stack_t ss = {
+			.ss_sp = &sig_stack,
+			.ss_size = sig_stack.len
+		};
+		libc::sigaltstack(&ss, null);
+	$endif
+
 	posix::install_signal_handler(libc::SIGBUS, &sig_bus_error);
 	posix::install_signal_handler(libc::SIGSEGV, &sig_segmentation_fault);
+	posix::install_signal_handler(libc::SIGILL,  &sig_illegal_instruction);
 }
-

lib/std/core/conv.c3

@@ -16,25 +16,25 @@ const uint UTF16_SURROGATE_HIGH_VALUE @private = 0xD800;
 *>
 fn usz? char32_to_utf8(Char32 c, char[] output)
 {
-	if (!output.len) return string::CONVERSION_FAILED?;
+	if (!output.len) return string::CONVERSION_FAILED~;
 	switch (true)
 	{
 		case c <= 0x7f:
 			output[0] = (char)c;
 			return 1;
 		case c <= 0x7ff:
-			if (output.len < 2) return string::CONVERSION_FAILED?;
+			if (output.len < 2) return string::CONVERSION_FAILED~;
 			output[0] = (char)(0xC0 | c >> 6);
 			output[1] = (char)(0x80 | (c & 0x3F));
 			return 2;
 		case c <= 0xffff:
-			if (output.len < 3) return string::CONVERSION_FAILED?;
+			if (output.len < 3) return string::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 string::CONVERSION_FAILED?;
+			if (output.len < 4) return string::CONVERSION_FAILED~;
 			output[0] = (char)(0xF0 | c >> 18);
 			output[1] = (char)(0x80 | (c >> 12 & 0x3F));
 			output[2] = (char)(0x80 | (c >> 6 & 0x3F));
@@ -42,7 +42,7 @@ fn usz? char32_to_utf8(Char32 c, char[] output)
 			return 4;
 		default:
 			// 0x10FFFF and above is not defined.
-			return string::CONVERSION_FAILED?;
+			return string::CONVERSION_FAILED~;
 	}
 }
 
@@ -84,15 +84,15 @@ fn void? char16_to_utf8_unsafe(Char16 *ptr, usz *available, char** output)
 		return;
 	}
 	// Low surrogate first is an error
-	if (high & UTF16_SURROGATE_MASK != UTF16_SURROGATE_HIGH_VALUE) return string::INVALID_UTF16?;
+	if (high & UTF16_SURROGATE_MASK != UTF16_SURROGATE_HIGH_VALUE) return string::INVALID_UTF16~;
 
 	// Unmatched high surrogate is an error
-	if (*available == 1) return string::INVALID_UTF16?;
+	if (*available == 1) return string::INVALID_UTF16~;
 
 	Char16 low = ptr[1];
 
 	// Unmatched high surrogate, invalid
-	if (low & UTF16_SURROGATE_MASK != UTF16_SURROGATE_LOW_VALUE) return string::INVALID_UTF16?;
+	if (low & UTF16_SURROGATE_MASK != UTF16_SURROGATE_LOW_VALUE) return string::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
@@ -138,7 +138,7 @@ fn usz char32_to_utf8_unsafe(Char32 c, char** output)
 fn Char32? utf8_to_char32(char* ptr, usz* size)
 {
 	usz max_size = *size;
-	if (max_size < 1) return string::INVALID_UTF8?;
+	if (max_size < 1) return string::INVALID_UTF8~;
 	char c = (ptr++)[0];
 
 	if ((c & 0x80) == 0)
@@ -148,40 +148,40 @@ fn Char32? utf8_to_char32(char* ptr, usz* size)
 	}
 	if ((c & 0xE0) == 0xC0)
 	{
-		if (max_size < 2) return string::INVALID_UTF8?;
+		if (max_size < 2) return string::INVALID_UTF8~;
 		*size = 2;
 		Char32 uc = (c & 0x1F) << 6;
 		c = *ptr;
 		// Overlong sequence or invalid second.
-		if (!uc || c & 0xC0 != 0x80) return string::INVALID_UTF8?;
+		if (!uc || c & 0xC0 != 0x80) return string::INVALID_UTF8~;
 		return uc + c & 0x3F;
 	}
 	if ((c & 0xF0) == 0xE0)
 	{
-		if (max_size < 3) return string::INVALID_UTF8?;
+		if (max_size < 3) return string::INVALID_UTF8~;
 		*size = 3;
 		Char32 uc = (c & 0x0F) << 12;
 		c = ptr++[0];
-		if (c & 0xC0 != 0x80) return string::INVALID_UTF8?;
+		if (c & 0xC0 != 0x80) return string::INVALID_UTF8~;
 		uc += (c & 0x3F) << 6;
 		c = ptr++[0];
 		// Overlong sequence or invalid last
-		if (!uc || c & 0xC0 != 0x80) return string::INVALID_UTF8?;
+		if (!uc || c & 0xC0 != 0x80) return string::INVALID_UTF8~;
 		return uc + c & 0x3F;
 	}
-	if (max_size < 4) return string::INVALID_UTF8?;
-	if ((c & 0xF8) != 0xF0) return string::INVALID_UTF8?;
+	if (max_size < 4) return string::INVALID_UTF8~;
+	if ((c & 0xF8) != 0xF0) return string::INVALID_UTF8~;
 	*size = 4;
 	Char32 uc = (c & 0x07) << 18;
 	c = ptr++[0];
-	if (c & 0xC0 != 0x80) return string::INVALID_UTF8?;
+	if (c & 0xC0 != 0x80) return string::INVALID_UTF8~;
 	uc += (c & 0x3F) << 12;
 	c = ptr++[0];
-	if (c & 0xC0 != 0x80) return string::INVALID_UTF8?;
+	if (c & 0xC0 != 0x80) return string::INVALID_UTF8~;
 	uc += (c & 0x3F) << 6;
 	c = ptr++[0];
 	// Overlong sequence or invalid last
-	if (!uc || c & 0xC0 != 0x80) return string::INVALID_UTF8?;
+	if (!uc || c & 0xC0 != 0x80) return string::INVALID_UTF8~;
 	return uc + c & 0x3F;
 }
 
@@ -329,7 +329,7 @@ fn usz? utf8to32(String utf8, Char32[] utf32_buffer)
 	usz buf_len = utf32_buffer.len;
 	for (usz i = 0; i < len;)
 	{
-		if (len32 == buf_len) return string::CONVERSION_FAILED?;
+		if (len32 == buf_len) return string::CONVERSION_FAILED~;
 		usz width = len - i;
 		Char32 uc = utf8_to_char32(&utf8[i], &width) @inline!;
 		i += width;

lib/std/core/dstring.c3

@@ -95,7 +95,7 @@ fn void DString.replace(&self, String needle, String replacement)
 				match++;
 				if (match == needle_len)
 				{
-					self.append_chars(replacement);
+					self.append_string(replacement);
 					match = 0;
 					continue;
 				}
@@ -103,12 +103,12 @@ fn void DString.replace(&self, String needle, String replacement)
 			}
 			if (match > 0)
 			{
-				self.append_chars(str[i - match:match]);
+				self.append_string(str[i - match:match]);
 				match = 0;
 			}
 			self.append_char(c);
 		}
-		if (match > 0) self.append_chars(str[^match:match]);
+		if (match > 0) self.append_string(str[^match:match]);
 	};
 }
 
@@ -305,18 +305,23 @@ fn bool DString.less(self, DString other_string)
 	return true;
 }
 
-fn void DString.append_chars(&self, String str)
+fn void DString.append_chars(&self, String str) @deprecated("Use append_string")
 {
-	usz other_len = str.len;
+	self.append_bytes(str);
+}
+
+fn void DString.append_bytes(&self, char[] bytes)
+{
+	usz other_len = bytes.len;
 	if (!other_len) return;
 	if (!*self)
 	{
-		*self = temp(str);
+		*self = temp((String)bytes);
 		return;
 	}
 	self.reserve(other_len);
 	StringData* data = self.data();
-	mem::copy(&data.chars[data.len], str.ptr, other_len);
+	mem::copy(&data.chars[data.len], bytes.ptr, other_len);
 	data.len += other_len;
 }
 
@@ -325,7 +330,24 @@ fn Char32[] DString.copy_utf32(&self, Allocator allocator)
 	return self.str_view().to_utf32(allocator) @inline!!;
 }
 
-fn void DString.append_string(&self, DString str)
+<*
+ @require $typeof(str) == String || $typeof(str) == DString : "Expected string or DString"
+*>
+macro void DString.append_string(&self, str)
+{
+	$if $typeof(str) == String:
+		self.append_bytes(str);
+	$else
+		self.append_string_deprecated(str);
+	$endif
+}
+
+macro void DString.append_string_deprecated(&self, DString str) @deprecated("Use .append_dstring()")
+{
+	self.append_dstring(str);
+}
+
+fn void DString.append_dstring(&self, DString str)
 {
 	StringData* other = str.data();
 	if (!other) return;
@@ -340,7 +362,7 @@ fn void DString.clear(self)
 
 fn usz? DString.write(&self, char[] buffer) @dynamic
 {
-	self.append_chars((String)buffer);
+	self.append_bytes(buffer);
 	return buffer.len;
 }
 
@@ -400,9 +422,9 @@ macro void DString.append(&self, value)
 		$case ichar:
 			self.append_char(value);
 		$case DString:
-			self.append_string(value);
+			self.append_dstring(value);
 		$case String:
-			self.append_chars(value);
+			self.append_string(value);
 		$case Char32:
 			self.append_char32(value);
 		$default:
@@ -410,7 +432,7 @@ macro void DString.append(&self, value)
 				$case $defined((Char32)value):
 					self.append_char32((Char32)value);
 				$case $defined((String)value):
-					self.append_chars((String)value);
+					self.append_string((String)value);
 				$default:
 					$error "Unsupported type for append – use appendf instead.";
 			$endswitch

lib/std/core/env.c3

@@ -124,7 +124,8 @@ const usz MAX_VECTOR_SIZE = $$MAX_VECTOR_SIZE;
 const bool ARCH_32_BIT = $$REGISTER_SIZE == 32;
 const bool ARCH_64_BIT = $$REGISTER_SIZE == 64;
 const bool LIBC = $$COMPILER_LIBC_AVAILABLE;
-const bool NO_LIBC = !$$COMPILER_LIBC_AVAILABLE;
+const bool NO_LIBC = !LIBC && !CUSTOM_LIBC;
+const bool CUSTOM_LIBC = $$CUSTOM_LIBC;
 const CompilerOptLevel COMPILER_OPT_LEVEL = CompilerOptLevel.from_ordinal($$COMPILER_OPT_LEVEL);
 const bool BIG_ENDIAN = $$PLATFORM_BIG_ENDIAN;
 const bool I128_NATIVE_SUPPORT = $$PLATFORM_I128_SUPPORTED;
@@ -153,19 +154,20 @@ const bool FREEBSD = LIBC && OS_TYPE == FREEBSD;
 const bool NETBSD = LIBC && OS_TYPE == NETBSD;
 const bool BSD_FAMILY = env::FREEBSD || env::OPENBSD || env::NETBSD;
 const bool WASI = LIBC && OS_TYPE == WASI;
+const bool WASM = ARCH_TYPE == ArchType.WASM32 || ARCH_TYPE == ArchType.WASM64;
 const bool ANDROID = LIBC && OS_TYPE == ANDROID;
 const bool WASM_NOLIBC @builtin @deprecated("Use 'FREESTANDING_WASM' instead") = !LIBC && ARCH_TYPE == ArchType.WASM32 || ARCH_TYPE == ArchType.WASM64;
 const bool FREESTANDING_PE32 = NO_LIBC && OS_TYPE == WIN32;
 const bool FREESTANDING_MACHO = NO_LIBC && OS_TYPE == MACOS;
 const bool FREESTANDING_ELF = NO_LIBC && !env::FREESTANDING_PE32 && !env::FREESTANDING_MACHO && !env::FREESTANDING_WASM;
-const bool FREESTANDING_WASM = NO_LIBC && (ARCH_TYPE == ArchType.WASM32 || ARCH_TYPE == ArchType.WASM64);
+const bool FREESTANDING_WASM = NO_LIBC && WASM;
 const bool FREESTANDING = env::FREESTANDING_PE32 || env::FREESTANDING_MACHO || env::FREESTANDING_ELF || env::FREESTANDING_WASM;
 const bool ADDRESS_SANITIZER = $$ADDRESS_SANITIZER;
 const bool MEMORY_SANITIZER = $$MEMORY_SANITIZER;
 const bool THREAD_SANITIZER = $$THREAD_SANITIZER;
 const bool ANY_SANITIZER = ADDRESS_SANITIZER || MEMORY_SANITIZER || THREAD_SANITIZER;
 const int LANGUAGE_DEV_VERSION = $$LANGUAGE_DEV_VERSION;
-const bool HAS_NATIVE_ERRNO = env::LINUX || env::ANDROID || env::OPENBSD || env::DARWIN || env::WIN32;
+const bool HAS_NATIVE_ERRNO = env::LINUX || env::ANDROID || env::OPENBSD || env::DARWIN || env::WIN32 || env::NETBSD;
 
 macro bool os_is_darwin() @const
 {

lib/std/core/logging.c3

@@ -148,17 +148,18 @@ fn void call_log_internal(LogPriority prio, LogCategory category, String file, S
 	LogPriority priority = mem::@atomic_load(config_priorities[category], UNORDERED);
 	if (priority > prio) return;
 	init();
-	bool locked = logger_mutex.is_initialized() && @ok(logger_mutex.lock());
+	bool locked = logger_mutex.is_initialized();
+	if (locked) logger_mutex.lock();
 	Logger logger = current_logger;
 	LogFn logfn = current_logfn;
-	defer if (locked) (void)logger_mutex.unlock();
+	defer if (locked) logger_mutex.unlock();
 	logfn(logger.ptr, prio, category, current_tag, file, func, line, fmt, args);
 }
 
 fn String? get_category_name(LogCategory category)
 {
 	String val = category_names[category];
-	return val ?: NOT_FOUND?;
+	return val ?: NOT_FOUND~;
 }
 
 fn void set_category_name(LogCategory category, String name)

lib/std/core/mem.c3

@@ -7,7 +7,7 @@ import std::os::posix, std::os::win32;
 import std::math;
 
 const MAX_MEMORY_ALIGNMENT = 0x1000_0000;
-const DEFAULT_MEM_ALIGNMENT = (void*.alignof) * 2;
+const DEFAULT_MEM_ALIGNMENT = env::WASM ? 16 : (void*.alignof) * 2;
 const ulong KB = 1024;
 const ulong MB = KB * 1024;
 const ulong GB = MB * 1024;
@@ -200,7 +200,7 @@ macro @scatter_aligned(ptrvec, value, bool[<*>] mask, usz $alignment)
 *>
 macro @unaligned_load(#x, usz $alignment) @builtin
 {
-	return $$unaligned_load(&#x, $alignment);
+	return $$unaligned_load(&#x, $alignment, false);
 }
 
 <*
@@ -215,9 +215,10 @@ macro @unaligned_load(#x, usz $alignment) @builtin
 *>
 macro @unaligned_store(#x, value, usz $alignment) @builtin
 {
-	return $$unaligned_store(&#x, ($typeof(#x))value, $alignment);
+	return $$unaligned_store(&#x, ($typeof(#x))value, $alignment, false);
 }
 
+
 <*
  @param #x : "The variable or dereferenced pointer to load."
  @return "The value of the variable"
@@ -242,6 +243,36 @@ macro @volatile_store(#x, value) @builtin
 	return $$volatile_store(&#x, ($typeof(#x))value);
 }
 
+<*
+ @param ptr : "The pointer to load from"
+ @param $align : "The alignment to assume for the load"
+ @param $volatile : "Whether the load is volatile or not, defaults to false"
+ @return "The value of the variable"
+
+ @require $defined(*ptr) : "This must be a typed pointer"
+ @require @constant_is_power_of_2($align) : "The alignment must be a power of two"
+*>
+macro load(ptr, usz $align, bool $volatile = false)
+{
+	return $$unaligned_load(ptr, $align, $volatile);
+}
+
+<*
+ @param ptr : "The pointer to store to."
+ @param value : "The value to store."
+ @param $align : "The alignment to assume for the store"
+ @param $volatile : "Whether the store is volatile, defaults to false"
+ @return "The value stored"
+
+ @require $defined(*ptr) : "This must be a typed pointer"
+ @require $defined(*ptr = value) : "The value doesn't match the variable"
+ @require @constant_is_power_of_2($align) : "The alignment must be a power of two"
+*>
+macro store(ptr, value, usz $align, bool $volatile = false)
+{
+	return $$unaligned_store(ptr, ($typeof(*ptr))value, $align, $volatile);
+}
+
 <*
  All possible atomic orderings
 *>
@@ -807,6 +838,7 @@ macro new_aligned($Type, #init = ...) @nodiscard @safemacro
 <*
  @param $Type : "The type to allocate"
 
+ @require $Type.kindof != OPTIONAL : "Expected a non-optional type"
  @require $Type.alignof <= DEFAULT_MEM_ALIGNMENT : "Types with alignment exceeding the default must use 'alloc_aligned' instead"
  @return "A pointer to uninitialized data for the type $Type"
 *>
@@ -819,7 +851,8 @@ macro alloc($Type) @nodiscard
  @param $Type : "The type to allocate"
  @param padding : "The padding to add after the allocation"
 
- @require $Type.alignof <= DEFAULT_MEM_ALIGNMENT : "Types with alignment exceeding the default must use 'alloc_aligned' instead"
+ @require $Type.kindof != OPTIONAL : "Expected a non-optional type"
+  @require $Type.alignof <= DEFAULT_MEM_ALIGNMENT : "Types with alignment exceeding the default must use 'alloc_aligned' instead"
  @return "A pointer to uninitialized data for the type $Type"
 *>
 macro alloc_with_padding($Type, usz padding) @nodiscard
@@ -833,6 +866,7 @@ macro alloc_with_padding($Type, usz padding) @nodiscard
  exceeding DEFAULT_MEM_ALIGNMENT. IMPORTANT! It must be freed using free_aligned.
 
  @param $Type : "The type to allocate"
+ @require $Type.kindof != OPTIONAL : "Expected a non-optional type"
  @return "A pointer to uninitialized data for the type $Type with the proper alignment"
 *>
 macro alloc_aligned($Type) @nodiscard
@@ -844,6 +878,7 @@ macro alloc_aligned($Type) @nodiscard
  @param $Type : "The type to allocate"
  @param #init : "The optional initializer"
 
+ @require $Type.kindof != OPTIONAL : "Expected a non-optional type"
  @require !$defined(#init) ||| $defined($Type a = #init) : "#init must be an initializer for the type"
  @return "A pointer to temporary data of type $Type."
 *>
@@ -863,6 +898,7 @@ macro tnew($Type, #init = ...) @nodiscard @safemacro
  @param padding : "The padding to add after the allocation"
  @param #init : "The optional initializer"
 
+ @require $Type.kindof != OPTIONAL : "Expected a non-optional type"
  @require !$defined(#init) ||| $defined($Type a = #init) : "#init must be an initializer for the type"
  @return "A pointer to temporary data of type $Type with added padding at the end."
 *>
@@ -877,17 +913,24 @@ macro temp_with_padding($Type, usz padding, #init = ...) @nodiscard @safemacro
 	$endif
 }
 
+<*
+ @require $Type.kindof != OPTIONAL : "Expected a non-optional type"
+*>
 macro talloc($Type) @nodiscard
 {
 	return tmalloc($Type.sizeof, $Type.alignof);
 }
 
+<*
+ @require $Type.kindof != OPTIONAL : "Expected a non-optional type"
+*>
 macro talloc_with_padding($Type, usz padding) @nodiscard
 {
 	return tmalloc($Type.sizeof + padding, $Type.alignof);
 }
 
 <*
+ @require $Type.kindof != OPTIONAL : "Expected a non-optional type"
  @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
@@ -898,6 +941,8 @@ macro new_array($Type, usz elements) @nodiscard
 <*
  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 $Type.kindof != OPTIONAL : "Expected a non-optional type"
 *>
 macro new_array_aligned($Type, usz elements) @nodiscard
 {
@@ -905,6 +950,7 @@ macro new_array_aligned($Type, usz elements) @nodiscard
 }
 
 <*
+ @require $Type.kindof != OPTIONAL : "Expected a non-optional type"
  @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
@@ -915,6 +961,8 @@ macro alloc_array($Type, usz elements) @nodiscard
 <*
  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 $Type.kindof != OPTIONAL : "Expected a non-optional type"
 *>
 macro alloc_array_aligned($Type, usz elements) @nodiscard
 {
@@ -1029,7 +1077,7 @@ fn void* __memcpy(void* dst, void* src, usz n) @weak @export("memcpy")
 }
 
 
-module std::core::mem::volatile { Type };
+module std::core::mem::volatile <Type>;
 
 typedef Volatile @structlike = Type;
 
@@ -1046,7 +1094,7 @@ macro Type Volatile.set(&self, Type val)
 <*
  @require mem::@constant_is_power_of_2(ALIGNMENT) : "The alignment must be a power of 2"
 *>
-module std::core::mem::alignment { Type, ALIGNMENT };
+module std::core::mem::alignment <Type, ALIGNMENT>;
 import std::core::mem @public;
 
 <*
@@ -1056,10 +1104,10 @@ typedef UnalignedRef = Type*;
 
 macro Type UnalignedRef.get(self)
 {
-	return @unaligned_load(*(Type*)self, ALIGNMENT);
+	return @unaligned_load(*(Type*)self, ALIGNMENT, false);
 }
 
 macro Type UnalignedRef.set(&self, Type val)
 {
-	return @unaligned_store(*(Type*)self, val, ALIGNMENT);
+	return @unaligned_store(*(Type*)self, val, ALIGNMENT, false);
 }

lib/std/core/mem_allocator.c3

@@ -304,6 +304,31 @@ macro alloc_array_try(Allocator allocator, $Type, usz elements) @nodiscard
 	return (($Type*)malloc_try(allocator, $Type.sizeof * elements))[:elements];
 }
 
+<*
+ @require $Type.alignof <= mem::DEFAULT_MEM_ALIGNMENT : "Types with alignment exceeding the default must use 'alloc_array_aligned' instead"
+*>
+macro realloc_array(Allocator allocator, void* ptr, $Type, usz elements) @nodiscard
+{
+	return realloc_array_try(allocator, ptr, $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 realloc_array_aligned(Allocator allocator, void* ptr, $Type, usz elements) @nodiscard
+{
+	return (($Type*)realloc_aligned(allocator, ptr, $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 realloc_array_try(Allocator allocator, void* ptr, $Type, usz elements) @nodiscard
+{
+	return (($Type*)realloc_try(allocator, ptr, $Type.sizeof * elements))[:elements];
+}
+
 <*
  Clone a value.
 
@@ -326,6 +351,8 @@ macro clone(Allocator allocator, value) @nodiscard
 *>
 macro clone_slice(Allocator allocator, slice) @nodiscard
 {
+	if (!lengthof(slice)) return {};
+
 	var $Type = $typeof(slice[0]);
 
 	$Type[] new_arr = new_array(allocator, $Type, slice.len);
@@ -556,12 +583,12 @@ typedef NullAllocator (Allocator) = uptr;
 
 fn void*? NullAllocator.acquire(&self, usz bytes, AllocInitType init_type, usz alignment) @dynamic
 {
-	return mem::OUT_OF_MEMORY?;
+	return mem::OUT_OF_MEMORY~;
 }
 
 fn void*? NullAllocator.resize(&self, void* old_ptr, usz new_bytes, usz alignment) @dynamic
 {
-	return mem::OUT_OF_MEMORY?;
+	return mem::OUT_OF_MEMORY~;
 }
 
 fn void NullAllocator.release(&self, void* old_ptr, bool aligned) @dynamic

lib/std/core/os/mem_vm.c3

@@ -54,11 +54,11 @@ fn void*? alloc(usz size, VirtualMemoryAccess access)
 			if (ptr != posix::MAP_FAILED) return ptr;
 			switch (libc::errno())
 			{
-				case errno::ENOMEM:    return mem::OUT_OF_MEMORY?;
-				case errno::EOVERFLOW: return RANGE_OVERFLOW?;
-				case errno::EPERM:     return ACCESS_DENIED?;
-				case errno::EINVAL:    return INVALID_ARGS?;
-				default:               return UNKNOWN_ERROR?;
+				case errno::ENOMEM:    return mem::OUT_OF_MEMORY~;
+				case errno::EOVERFLOW: return RANGE_OVERFLOW~;
+				case errno::EPERM:     return ACCESS_DENIED~;
+				case errno::EINVAL:    return INVALID_ARGS~;
+				default:               return UNKNOWN_ERROR~;
 			}
 		$case env::WIN32:
 			void* ptr = win32::virtualAlloc(null, aligned_alloc_size(size), MEM_RESERVE, access.to_win32());
@@ -66,8 +66,8 @@ fn void*? alloc(usz size, VirtualMemoryAccess access)
 			switch (win32::getLastError())
 			{
 				case win32::ERROR_NOT_ENOUGH_MEMORY:
-				case win32::ERROR_COMMITMENT_LIMIT:  return mem::OUT_OF_MEMORY?;
-				default:                             return UNKNOWN_ERROR?;
+				case win32::ERROR_COMMITMENT_LIMIT:  return mem::OUT_OF_MEMORY~;
+				default:                             return UNKNOWN_ERROR~;
 			}
 		$default:
 			unsupported("Virtual alloc only available on Win32 and Posix");
@@ -89,18 +89,18 @@ fn void? release(void* ptr, usz size)
 			{
 				switch (libc::errno())
 				{
-					case errno::EINVAL: return INVALID_ARGS?;    // Not a valid mapping or size
-					case errno::ENOMEM: return UNMAPPED_ACCESS?; // Address not mapped
-					default:            return RELEASE_FAILED?;
+					case errno::EINVAL: return INVALID_ARGS~;    // Not a valid mapping or size
+					case errno::ENOMEM: return UNMAPPED_ACCESS~; // Address not mapped
+					default:            return RELEASE_FAILED~;
 				}
 			}
 		$case env::WIN32:
 			if (win32::virtualFree(ptr, 0, MEM_RELEASE)) return;
 			switch (win32::getLastError())
 			{
-				case win32::ERROR_INVALID_ADDRESS:   return INVALID_ARGS?;
-				case win32::ERROR_NOT_ENOUGH_MEMORY: return mem::OUT_OF_MEMORY?;
-				default:                             return RELEASE_FAILED?;
+				case win32::ERROR_INVALID_ADDRESS:   return INVALID_ARGS~;
+				case win32::ERROR_NOT_ENOUGH_MEMORY: return mem::OUT_OF_MEMORY~;
+				default:                             return RELEASE_FAILED~;
 			}
 		$default:
 			unsupported("Virtual free only available on Win32 and Posix");
@@ -124,21 +124,21 @@ fn void? protect(void* ptr, usz len, VirtualMemoryAccess access)
 			if (!posix::mprotect(ptr, len, access.to_posix())) return;
 			switch (libc::errno())
 			{
-				case errno::EACCES:    return ACCESS_DENIED?;
-				case errno::EINVAL:    return UNALIGNED_ADDRESS?;
-				case errno::EOVERFLOW: return RANGE_OVERFLOW?;
-				case errno::ENOMEM:    return UNMAPPED_ACCESS?;
-				default:               return UPDATE_FAILED?;
+				case errno::EACCES:    return ACCESS_DENIED~;
+				case errno::EINVAL:    return UNALIGNED_ADDRESS~;
+				case errno::EOVERFLOW: return RANGE_OVERFLOW~;
+				case errno::ENOMEM:    return UNMAPPED_ACCESS~;
+				default:               return UPDATE_FAILED~;
 			}
 		$case env::WIN32:
 			Win32_Protect old;
 			if (win32::virtualProtect(ptr, len, access.to_win32(), &old)) return;
 			switch (win32::getLastError())
 			{
-				case win32::ERROR_INVALID_ADDRESS:   return UNALIGNED_ADDRESS?;
-				case win32::ERROR_ACCESS_DENIED:     return ACCESS_DENIED?;
-				case win32::ERROR_INVALID_PARAMETER: return INVALID_ARGS?;
-				default:                             return UPDATE_FAILED?;
+				case win32::ERROR_INVALID_ADDRESS:   return UNALIGNED_ADDRESS~;
+				case win32::ERROR_ACCESS_DENIED:     return ACCESS_DENIED~;
+				case win32::ERROR_INVALID_PARAMETER: return INVALID_ARGS~;
+				default:                             return UPDATE_FAILED~;
 			}
 		$default:
 			unsupported("'virtual_protect' is only available on Win32 and Posix.");
@@ -165,12 +165,12 @@ fn void? commit(void* ptr, usz len, VirtualMemoryAccess access = READWRITE)
 			if (result) return;
 			switch (win32::getLastError())
 			{
-				case win32::ERROR_INVALID_ADDRESS:   return UNALIGNED_ADDRESS?;
-				case win32::ERROR_ACCESS_DENIED:     return ACCESS_DENIED?;
+				case win32::ERROR_INVALID_ADDRESS:   return UNALIGNED_ADDRESS~;
+				case win32::ERROR_ACCESS_DENIED:     return ACCESS_DENIED~;
 				case win32::ERROR_COMMITMENT_LIMIT:
-				case win32::ERROR_NOT_ENOUGH_MEMORY: return mem::OUT_OF_MEMORY?;
-				case win32::ERROR_INVALID_PARAMETER: return INVALID_ARGS?;
-				default:                             return UNKNOWN_ERROR?;
+				case win32::ERROR_NOT_ENOUGH_MEMORY: return mem::OUT_OF_MEMORY~;
+				case win32::ERROR_INVALID_PARAMETER: return INVALID_ARGS~;
+				default:                             return UNKNOWN_ERROR~;
 			}
 		$default:
 			unsupported("'virtual_commit' is only available on Win32 and Posix.");
@@ -197,9 +197,9 @@ fn void? decommit(void* ptr, usz len, bool block = true)
 			{
 				switch (libc::errno())
 				{
-					case errno::EINVAL: return UNALIGNED_ADDRESS?;
-					case errno::ENOMEM: return UNMAPPED_ACCESS?;
-					default:            return UPDATE_FAILED?;
+					case errno::EINVAL: return UNALIGNED_ADDRESS~;
+					case errno::ENOMEM: return UNMAPPED_ACCESS~;
+					default:            return UPDATE_FAILED~;
 				}
 			}
 			if (block) (void)protect(ptr, len, PROTECTED) @inline;
@@ -208,10 +208,10 @@ fn void? decommit(void* ptr, usz len, bool block = true)
 			{
 				switch (win32::getLastError())
 				{
-					case win32::ERROR_INVALID_ADDRESS:   return UNALIGNED_ADDRESS?;
-					case win32::ERROR_INVALID_PARAMETER: return INVALID_ARGS?;
-					case win32::ERROR_ACCESS_DENIED:     return ACCESS_DENIED?;
-					default:                             return UPDATE_FAILED?;
+					case win32::ERROR_INVALID_ADDRESS:   return UNALIGNED_ADDRESS~;
+					case win32::ERROR_INVALID_PARAMETER: return INVALID_ARGS~;
+					case win32::ERROR_ACCESS_DENIED:     return ACCESS_DENIED~;
+					default:                             return UPDATE_FAILED~;
 				}
 			}
 		$default:
@@ -237,15 +237,15 @@ fn void*? mmap_file(Fd fd, usz size, usz offset = 0, VirtualMemoryAccess access
 	if (ptr != posix::MAP_FAILED) return ptr;
 	switch (libc::errno())
 	{
-		case errno::ENOMEM:    return mem::OUT_OF_MEMORY?;
-		case errno::EOVERFLOW: return RANGE_OVERFLOW?;
-		case errno::EPERM:     return ACCESS_DENIED?;
-		case errno::EINVAL:    return INVALID_ARGS?;
-		case errno::EACCES:    return io::NO_PERMISSION?;
-		case errno::EBADF:     return io::FILE_NOT_VALID?;
-		case errno::EAGAIN:    return io::WOULD_BLOCK?;
-		case errno::ENXIO:     return io::FILE_NOT_FOUND?;
-		default:               return UNKNOWN_ERROR?;
+		case errno::ENOMEM:    return mem::OUT_OF_MEMORY~;
+		case errno::EOVERFLOW: return RANGE_OVERFLOW~;
+		case errno::EPERM:     return ACCESS_DENIED~;
+		case errno::EINVAL:    return INVALID_ARGS~;
+		case errno::EACCES:    return io::NO_PERMISSION~;
+		case errno::EBADF:     return io::FILE_NOT_VALID~;
+		case errno::EAGAIN:    return io::WOULD_BLOCK~;
+		case errno::ENXIO:     return io::FILE_NOT_FOUND~;
+		default:               return UNKNOWN_ERROR~;
 	}
 }
 

lib/std/core/os/wasm/wasm_memory.c3

@@ -25,7 +25,7 @@ fn char[]? WasmMemory.allocate_block(&self, usz bytes)
 	}
 
 	usz blocks_required = (bytes_required + WASM_BLOCK_SIZE + 1) / WASM_BLOCK_SIZE;
-	if ($$wasm_memory_grow(0, blocks_required) == -1) return mem::OUT_OF_MEMORY?;
+	if ($$wasm_memory_grow(0, blocks_required) == -1) return mem::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/macho_runtime.c3

@@ -79,7 +79,7 @@ fn SegmentCommand64*? find_segment(MachHeader* header, char* segname)
 		}
 		command = (void*)command + command.cmdsize;
 	}
-	return NOT_FOUND?;
+	return NOT_FOUND~;
 }
 fn Section64*? find_section(SegmentCommand64* command, char* sectname)
 {
@@ -89,7 +89,7 @@ fn Section64*? find_section(SegmentCommand64* command, char* sectname)
 		if (name_cmp(sectname, &section.sectname)) return section;
 		section++;
 	}
-	return NOT_FOUND?;
+	return NOT_FOUND~;
 }
 
 macro find_segment_section_body(MachHeader* header, char* segname, char* sectname, $Type)

lib/std/core/private/main_stub.c3

@@ -12,7 +12,10 @@ 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_int_main(#m, int, char**)
+{
+	return #m();
+}
 macro int @main_to_void_main(#m, int, char**)
 {
 	#m();
@@ -63,6 +66,17 @@ macro int @main_to_void_main_args(#m, int argc, char** argv)
 }
 
 module std::core::main_stub @if(env::WIN32);
+import std::os::win32;
+
+macro win32_set_utf8_codepage() @local
+{
+	// By default windows uses an OEM codepage that differs based on locale
+	// and does not support printing utf-8 characters. This allows both
+	// printing utf-8 characters from strings and reading them from stdin.
+	win32::setConsoleCP(UTF8);
+	win32::setConsoleOutputCP(UTF8);
+}
+
 
 extern fn Char16** _win_command_line_to_argv_w(ushort* cmd_line, int* argc_ptr) @cname("CommandLineToArgvW");
 
@@ -93,18 +107,26 @@ macro void release_wargs(String[] list) @private
 
 macro int @win_to_err_main_noargs(#m, void* handle, void* prev_handle, Char16* cmd_line, int show_cmd)
 {
+	win32_set_utf8_codepage();
 	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_int_main_noargs(#m, void* handle, void* prev_handle, Char16* cmd_line, int show_cmd)
+{
+	win32_set_utf8_codepage();
+	return #m();
+}
+
 macro int @win_to_void_main_noargs(#m, void* handle, void* prev_handle, Char16* cmd_line, int show_cmd)
 {
+	win32_set_utf8_codepage();
 	#m();
 	return 0;
 }
 
 macro int @win_to_err_main_args(#m, void* handle, void* prev_handle, Char16* cmd_line, int show_cmd)
 {
+	win32_set_utf8_codepage();
 	String[] args = win_command_line_to_strings(cmd_line);
 	defer release_wargs(args);
 	if (catch #m(args)) return 1;
@@ -113,6 +135,7 @@ macro int @win_to_err_main_args(#m, void* handle, void* prev_handle, Char16* cmd
 
 macro int @win_to_int_main_args(#m, void* handle, void* prev_handle, Char16* cmd_line, int show_cmd)
 {
+	win32_set_utf8_codepage();
 	String[] args = win_command_line_to_strings(cmd_line);
 	defer release_wargs(args);
 	return #m(args);
@@ -120,6 +143,7 @@ macro int @win_to_int_main_args(#m, void* handle, void* prev_handle, Char16* cmd
 
 macro int @win_to_void_main_args(#m, void* handle, void* prev_handle, Char16* cmd_line, int show_cmd)
 {
+	win32_set_utf8_codepage();
 	String[] args = win_command_line_to_strings(cmd_line);
 	defer release_wargs(args);
 	#m(args);
@@ -128,6 +152,7 @@ macro int @win_to_void_main_args(#m, void* handle, void* prev_handle, Char16* cm
 
 macro int @win_to_err_main(#m, void* handle, void* prev_handle, Char16* cmd_line, int show_cmd)
 {
+	win32_set_utf8_codepage();
 	String[] args = win_command_line_to_strings(cmd_line);
 	defer release_wargs(args);
 	if (catch #m(handle, prev_handle, args, show_cmd)) return 1;
@@ -136,6 +161,7 @@ macro int @win_to_err_main(#m, void* handle, void* prev_handle, Char16* cmd_line
 
 macro int @win_to_int_main(#m, void* handle, void* prev_handle, Char16* cmd_line, int show_cmd)
 {
+	win32_set_utf8_codepage();
 	String[] args = win_command_line_to_strings(cmd_line);
 	defer release_wargs(args);
 	return #m(handle, prev_handle, args, show_cmd);
@@ -143,6 +169,7 @@ macro int @win_to_int_main(#m, void* handle, void* prev_handle, Char16* cmd_line
 
 macro int @win_to_void_main(#m, void* handle, void* prev_handle, Char16* cmd_line, int show_cmd)
 {
+	win32_set_utf8_codepage();
 	String[] args = win_command_line_to_strings(cmd_line);
 	defer release_wargs(args);
 	#m(handle, prev_handle, args, show_cmd);
@@ -151,6 +178,7 @@ macro int @win_to_void_main(#m, void* handle, void* prev_handle, Char16* cmd_lin
 
 macro int @wmain_to_err_main_args(#m, int argc, Char16** argv)
 {
+	win32_set_utf8_codepage();
 	String[] args = wargs_strings(argc, argv);
 	defer release_wargs(args);
 	if (catch #m(args)) return 1;
@@ -159,6 +187,7 @@ macro int @wmain_to_err_main_args(#m, int argc, Char16** argv)
 
 macro int @wmain_to_int_main_args(#m, int argc, Char16** argv)
 {
+	win32_set_utf8_codepage();
 	String[] args = wargs_strings(argc, argv);
 	defer release_wargs(args);
 	return #m(args);
@@ -166,6 +195,7 @@ macro int @wmain_to_int_main_args(#m, int argc, Char16** argv)
 
 macro int @_wmain_runner(#m, int argc, Char16** argv)
 {
+	win32_set_utf8_codepage();
 	String[] args = wargs_strings(argc, argv);
 	defer release_wargs(args);
 	return #m(args) ? 0 : 1;
@@ -173,6 +203,7 @@ macro int @_wmain_runner(#m, int argc, Char16** argv)
 
 macro int @wmain_to_void_main_args(#m, int argc, Char16** argv)
 {
+	win32_set_utf8_codepage();
 	String[] args = wargs_strings(argc, argv);
 	defer release_wargs(args);
 	#m(args);

lib/std/core/refcount.c3

@@ -8,7 +8,7 @@
  @require !$defined(Type.dealloc) ||| $defined(Type.dealloc(&&(Type){})) : "'dealloc' must only take a pointer to the underlying type"
  @require !$defined(Type.dealloc) ||| $typeof((Type){}.dealloc()) ==  void : "'dealloc' must return 'void'"
 *>
-module std::core::mem::ref { Type };
+module std::core::mem::ref <Type>;
 import std::thread, std::atomic;
 
 const OVERALIGNED @private = Type.alignof > mem::DEFAULT_MEM_ALIGNMENT;

lib/std/core/runtime.c3

@@ -27,7 +27,7 @@ macro @enum_lookup($Type, #value, value)
 	$foreach $val : $Type.values:
 		if ($val.#value == value) return $val;
 	$endforeach
-	return NOT_FOUND?;
+	return NOT_FOUND~;
 }
 
 macro @enum_lookup_new($Type, $name, value)
@@ -35,7 +35,7 @@ macro @enum_lookup_new($Type, $name, value)
 	$foreach $val : $Type.values:
 		if ($val.$eval($name) == value) return $val;
 	$endforeach
-	return NOT_FOUND?;
+	return NOT_FOUND~;
 }
 
 

lib/std/core/runtime_test.c3

@@ -27,6 +27,8 @@ struct TestContext
 	bool is_in_panic;
 	bool is_quiet_mode;
 	bool is_no_capture;
+    bool sort;
+    bool check_leaks;
 	String current_test_name;
 	TestFn setup_fn;
 	TestFn teardown_fn;
@@ -185,8 +187,6 @@ fn void unmute_output(bool has_error) @local
 fn bool run_tests(String[] args, TestUnit[] tests) @private
 {
 	usz max_name;
-	bool sort_tests = true;
-	bool check_leaks = true;
 	if (!tests.len)
 	{
 		io::printn("There are no test units to run.");
@@ -204,6 +204,8 @@ $endif
 	{
 		.assert_print_backtrace = true,
 		.breakpoint_on_assert = false,
+        .sort = true,
+        .check_leaks = true,
 		.log_level = LogPriority.ERROR,
 		.test_filter = "",
 		.has_ansi_codes = terminal_has_ansi_codes(),
@@ -218,9 +220,9 @@ $endif
 			case "--test-breakpoint":
 				context.breakpoint_on_assert = true;
 			case "--test-nosort":
-				sort_tests = false;
+				context.sort = false;
 			case "--test-noleak":
-				check_leaks = false;
+				context.check_leaks = false;
 			case "--test-nocapture":
 			case "--test-show-output":
 				context.is_no_capture = true;
@@ -261,7 +263,7 @@ $endif
 	test_context = &context;
 	log::set_priority_all(test_context.log_level);
 
-	if (sort_tests)
+	if (context.sort)
 	{
 		quicksort(tests, &cmp_test_unit);
 	}
@@ -321,14 +323,17 @@ $endif
 		{
 			mute_output();
 			mem.clear();
-			if (check_leaks) allocator::thread_allocator = &mem;
-			unit.func();
+			if (context.check_leaks) allocator::thread_allocator = &mem;
+			@pool()
+            {
+                unit.func();
+            };
 			// 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;
+			if (context.check_leaks) allocator::thread_allocator = context.stored.allocator;
 
 			unmute_output(false); // all good, discard output
 			if (mem.has_leaks())

lib/std/core/slice2d.c3

@@ -1,10 +1,10 @@
-module std::core::array::slice {Type};
+module std::core::array;
 
 <*
  A slice2d allows slicing an array like int[10][10] into an arbitrary "int[][]"-like counterpart
  Typically you'd use array::slice2d(...) to create one.
 *>
-struct Slice2d
+struct Slice2d <Type>
 {
 	Type* ptr;
 	usz inner_len;

lib/std/core/string.c3

@@ -120,7 +120,7 @@ fn String bformat(char[] buffer, String fmt, args...) @format(1)
 	OutputFn format_fn = fn void?(void* buf, char c) {
 		char[]* buffer_ref = buf;
 		char[] buffer = *buffer_ref;
-		if (buffer.len == 0) return io::BUFFER_EXCEEDED?;
+		if (buffer.len == 0) return io::BUFFER_EXCEEDED~;
 		buffer[0] = c;
 		*buffer_ref = buffer[1..];
 	};
@@ -190,13 +190,18 @@ fn String join(Allocator allocator, String[] s, String joiner)
  @param [&inout] allocator : `The allocator to use for the String`
  @return "The new string with the elements replaced"
 *>
-fn String String.replace(self, Allocator allocator, String needle, String new_str) @nodiscard
+fn String String.replace(self, Allocator allocator, String needle, String new_str) @nodiscard => @pool()
 {
-	@pool()
+	Splitter s = self.tokenize_all(needle);
+	DString d;
+	d.init(tmem, new_str.len * 2 + self.len + 16);
+	(void)d.append(s.next());
+	while (try element = s.next())
 	{
-		String[] split = self.tsplit(needle);
-		return dstring::join(tmem, split, new_str).copy_str(allocator);
-	};
+		d.append(new_str);
+		d.append(element);
+	}
+	return d.copy_str(allocator);
 }
 
 <*
@@ -209,8 +214,16 @@ fn String String.replace(self, Allocator allocator, String needle, String new_st
 *>
 fn String String.treplace(self, String needle, String new_str)
 {
-	String[] split = self.tsplit(needle);
-	return dstring::join(tmem, split, new_str).str_view();
+	Splitter s = self.tokenize_all(needle);
+	DString d;
+	d.init(tmem, new_str.len * 2 + self.len + 16);
+	(void)d.append(s.next());
+	while (try element = s.next())
+	{
+		d.append(new_str);
+		d.append(element);
+	}
+	return d.str_view();
 }
 
 
@@ -355,7 +368,7 @@ fn String[] String.split(self, Allocator allocator, String delimiter, usz max =
 	bool no_more = false;
 	while (!no_more)
 	{
-		usz? index = i == max - 1 ? NOT_FOUND? : self.index_of(delimiter);
+		usz? index = i == max - 1 ? NOT_FOUND~ : self.index_of(delimiter);
 		String res @noinit;
 		if (try index)
 		{
@@ -414,7 +427,7 @@ fn String[]? String.split_to_buffer(s, String delimiter, String[] buffer, usz ma
 	bool no_more = false;
 	while (!no_more)
 	{
-		usz? index = i == max - 1 ? NOT_FOUND? : s.index_of(delimiter);
+		usz? index = i == max - 1 ? NOT_FOUND~ : s.index_of(delimiter);
 		String res @noinit;
 		if (try index)
 		{
@@ -432,7 +445,7 @@ fn String[]? String.split_to_buffer(s, String delimiter, String[] buffer, usz ma
 		}
 		if (i == max_capacity)
 		{
-			return BUFFER_EXCEEDED?;
+			return BUFFER_EXCEEDED~;
 		}
 		buffer[i++] = res;
 	}
@@ -513,7 +526,7 @@ fn usz? String.index_of_char(self, char character)
 	{
 		if (c == character) return i;
 	}
-	return NOT_FOUND?;
+	return NOT_FOUND~;
 }
 
 <*
@@ -536,7 +549,7 @@ fn usz? String.index_of_chars(String self, char[] characters)
         }
     }
 
-    return NOT_FOUND?;
+    return NOT_FOUND~;
 }
 
 <*
@@ -553,12 +566,12 @@ fn usz? String.index_of_chars(String self, char[] characters)
 fn usz? String.index_of_char_from(self, char character, usz start_index)
 {
 	usz len = self.len;
-	if (len <= start_index) return NOT_FOUND?;
+	if (len <= start_index) return NOT_FOUND~;
 	for (usz i = start_index; i < len; i++)
 	{
 		if (self[i] == character) return i;
 	}
-	return NOT_FOUND?;
+	return NOT_FOUND~;
 }
 
 <*
@@ -577,7 +590,7 @@ fn usz? String.rindex_of_char(self, char character)
 	{
 		if (c == character) return i;
 	}
-	return NOT_FOUND?;
+	return NOT_FOUND~;
 }
 
 <*
@@ -602,7 +615,7 @@ fn usz? String.index_of(self, String substr)
 			if (c == first && self[i : needed] == substr) return i;
 		}
 	}
-	return NOT_FOUND?;
+	return NOT_FOUND~;
 }
 
 <*
@@ -627,7 +640,7 @@ fn usz? String.rindex_of(self, String substr)
 			if (c == first && self[i : needed] == substr) return i;
 		}
 	}
-	return NOT_FOUND?;
+	return NOT_FOUND~;
 }
 
 fn bool ZString.eq(self, ZString other) @operator(==)
@@ -1013,12 +1026,12 @@ macro String.to_integer(self, $Type, int base = 10)
 	usz index = 0;
 	char* ptr = self.ptr;
 	while (index < len && ptr[index].is_blank()) index++;
-	if (len == index) return EMPTY_STRING?;
+	if (len == index) return EMPTY_STRING~;
 	bool is_negative;
 	switch (self[index])
 	{
 		case '-':
-			if ($Type.min == 0) return NEGATIVE_VALUE?;
+			if ($Type.min == 0) return NEGATIVE_VALUE~;
 			is_negative = true;
 			index++;
 		case '+':
@@ -1026,7 +1039,7 @@ macro String.to_integer(self, $Type, int base = 10)
 		default:
 			break;
 	}
-	if (len == index) return MALFORMED_INTEGER?;
+	if (len == index) return MALFORMED_INTEGER~;
 	$Type base_used = ($Type)base;
 	if (self[index] == '0' && base == 10)
 	{
@@ -1049,7 +1062,7 @@ macro String.to_integer(self, $Type, int base = 10)
 			default:
 				break;
 		}
-		if (len == index) return MALFORMED_INTEGER?;
+		if (len == index) return MALFORMED_INTEGER~;
 	}
 	$Type value = 0;
 	while (index != len)
@@ -1059,18 +1072,18 @@ macro String.to_integer(self, $Type, int base = 10)
 		{
 			case base_used < 10 || c < 'A':   c -= '0';
 			case c <= 'F':                    c -= 'A' - 10;
-			case c < 'a' || c > 'f':          return MALFORMED_INTEGER?;
+			case c < 'a' || c > 'f':          return MALFORMED_INTEGER~;
 			default:                          c -= 'a' - 10;
 		}
-		if (c >= base_used) return MALFORMED_INTEGER?;
+		if (c >= base_used) return MALFORMED_INTEGER~;
 		do
 		{
 			if (is_negative)
 			{
-				value = value.overflow_mul(base_used).overflow_sub(c) ?? INTEGER_OVERFLOW?!;
+				value = value.overflow_mul(base_used).overflow_sub(c) ?? INTEGER_OVERFLOW~!;
 				break;
 			}
-			value = value.overflow_mul(base_used).overflow_add(c) ?? INTEGER_OVERFLOW?!;
+			value = value.overflow_mul(base_used).overflow_add(c) ?? INTEGER_OVERFLOW~!;
 		};
 	}
 	return value;
@@ -1184,16 +1197,21 @@ fn void Splitter.reset(&self)
 	self.current = 0;
 }
 
+fn bool Splitter.at_end(&self)
+{
+	return self.current > self.string.len;
+}
+
 fn String? Splitter.next(&self)
 {
 	while (true)
 	{
 		usz len = self.string.len;
 		usz current = self.current;
-		if (current > len) return NO_MORE_ELEMENT?;
+		if (current > len) return NO_MORE_ELEMENT~;
 		if (current == len)
 		{
-			if (self.type != TOKENIZE_ALL) return NO_MORE_ELEMENT?;
+			if (self.type != TOKENIZE_ALL) return NO_MORE_ELEMENT~;
 			self.current++;
 			return self.string[current - 1:0];
 		}
@@ -1205,7 +1223,7 @@ fn String? Splitter.next(&self)
 			if (!next && self.type == TOKENIZE) continue;
 			return remaining[:next];
 		}
-		self.current = len;
+		self.current = len + 1;
 		return remaining;
 	}
 }

lib/std/core/string_escape.c3

@@ -1,4 +1,4 @@
-// Copyright (c) 2024 Christoffer Lerno. All rights reserved.
+// Copyright (c) 2024-2025 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.
 
@@ -22,43 +22,57 @@ faultdef INVALID_ESCAPE_SEQUENCE, UNTERMINATED_STRING, INVALID_HEX_ESCAPE, INVAL
 *>
 fn String String.escape(String s, Allocator allocator, bool strip_quotes = true)
 {
-    // Conservative allocation: most strings need minimal escaping
-    usz initial_capacity = s.len + s.len / 5 + 2; // ~1.2x + quotes
-    DString result = dstring::new_with_capacity(allocator, initial_capacity);
+	// Conservative allocation: most strings need minimal escaping
+	usz initial_capacity = s.len + s.len / 5 + 2; // ~1.2x + quotes
 
-    if (!strip_quotes) result.append_char('"');
-
-    foreach (char c : s)
-    {
-        switch (c)
-        {
-            case '"':  result.append(`\"`);
-            case '\\': result.append(`\\`);
-            case '\b': result.append(`\b`);
-            case '\f': result.append(`\f`);
-            case '\n': result.append(`\n`);
-            case '\r': result.append(`\r`);
-            case '\t': result.append(`\t`);
-            case '\v': result.append(`\v`);
-            case '\0': result.append(`\0`);
-            default:
-                if (c >= 32 && c <= 126)
-                {
-                    // Printable ASCII
-                    result.append_char(c);
-                }
-                else
-                {
-                    // Non-printable, use hex escape
-                    result.appendf("\\x%02x", (uint)c);
-                }
-        }
-    }
-
-    if (!strip_quotes) result.append_char('"');
-    return result.copy_str(allocator);
+	if (allocator == tmem)
+	{
+		DString result = dstring::new_with_capacity(tmem, initial_capacity);
+		escape_dstring(s, result, strip_quotes);
+		return result.str_view();
+	}
+	@pool()
+	{
+		DString result = dstring::temp_with_capacity(initial_capacity);
+		escape_dstring(s, result, strip_quotes);
+		return result.copy_str(allocator);
+	};
 }
 
+fn void escape_dstring(String s, DString result, bool strip_quotes) @private
+{
+	if (!strip_quotes) result.append_char('"');
+
+	foreach (char c : s)
+	{
+		switch (c)
+		{
+			case '"':  result.append(`\"`);
+			case '\\': result.append(`\\`);
+			case '\b': result.append(`\b`);
+			case '\f': result.append(`\f`);
+			case '\n': result.append(`\n`);
+			case '\r': result.append(`\r`);
+			case '\t': result.append(`\t`);
+			case '\v': result.append(`\v`);
+			case '\0': result.append(`\0`);
+			default:
+				if (c >= 32 && c <= 126)
+				{
+					// Printable ASCII
+					result.append_char(c);
+				}
+				else
+				{
+					// Non-printable, use hex escape
+					result.appendf("\\x%02x", (uint)c);
+				}
+		}
+	}
+
+	if (!strip_quotes) result.append_char('"');
+
+}
 <*
  Escape a string using the temp allocator.
 
@@ -76,33 +90,33 @@ fn String String.tescape(String s, bool strip_quotes = false) => s.escape(tmem,
 *>
 fn usz escape_len(String s)
 {
-    usz len = 2; // For quotes
-    foreach (char c : s)
-    {
-        switch (c)
-        {
-            case '"':
-            case '\\':
-            case '\b':
-            case '\f':
-            case '\n':
-            case '\r':
-            case '\t':
-            case '\v':
-            case '\0':
-                len += 2; // \X
-            default:
-                if (c >= 32 && c <= 126)
-                {
-                    len += 1;
-                }
-                else
-                {
-                    len += 4; // \xHH
-                }
-        }
-    }
-    return len;
+	usz len = 2; // For quotes
+	foreach (char c : s)
+	{
+		switch (c)
+		{
+			case '"':
+			case '\\':
+			case '\b':
+			case '\f':
+			case '\n':
+			case '\r':
+			case '\t':
+			case '\v':
+			case '\0':
+				len += 2; // \X
+			default:
+				if (c >= 32 && c <= 126)
+				{
+					len += 1;
+				}
+				else
+				{
+					len += 4; // \xHH
+				}
+		}
+	}
+	return len;
 }
 
 <*
@@ -111,90 +125,103 @@ fn usz escape_len(String s)
  @param allocator : "The allocator to use for the result"
  @param s : "The quoted string to unescape"
  @param allow_unquoted : "Set to true to unescape strings not surrounded by quotes, defaults to false"
+ @param lenient : "Be lenient with escapes, resolving unknown sequences to the escape character, defaults to false"
  @return "The unescaped string without quotes, safe to convert to ZString"
  @return? UNTERMINATED_STRING, INVALID_ESCAPE_SEQUENCE, INVALID_HEX_ESCAPE, INVALID_UNICODE_ESCAPE
 *>
-fn String? String.unescape(String s, Allocator allocator, bool allow_unquoted = false)
+fn String? String.unescape(String s, Allocator allocator, bool allow_unquoted = false, bool lenient = false)
 {
 	if (s.len >= 2 && s[0] == '"' && s[^1] == '"')
 	{
 		// Remove quotes.
 		s = s[1:^2];
 	}
-	else if (!allow_unquoted) return UNTERMINATED_STRING?;
+	else if (!allow_unquoted) return UNTERMINATED_STRING~;
 
-    // Handle empty string case
-    if (!s.len)
-    {
-        return "".copy(allocator);
-    }
-
-    DString result = dstring::new_with_capacity(allocator, s.len);
+	// Handle empty string case
+	if (!s.len)
+	{
+		return "".copy(allocator);
+	}
+	if (allocator == tmem)
+	{
+		DString result = dstring::new_with_capacity(tmem, s.len);
+		unescape_dstring(s, result, allow_unquoted, lenient)!;
+		return result.str_view();
+	}
+	@pool()
+	{
+		DString result = dstring::temp_with_capacity(s.len);
+		unescape_dstring(s, result, allow_unquoted, lenient)!;
+		return result.copy_str(allocator);
+	};
+}
 
+fn void? unescape_dstring(String s, DString result, bool allow_unquoted = false, bool lenient = false) @private
+{
 	usz len = s.len;
 	for (usz i = 0; i < len; i++)
 	{
 		char c = s[i];
-        if (c != '\\')
-        {
-            result.append_char(c);
-            continue;
-        }
+		if (c != '\\')
+		{
+			result.append_char(c);
+			continue;
+		}
 
-        // Handle escape sequence
-        if (i + 1 >= len) return INVALID_ESCAPE_SEQUENCE?;
+		// Handle escape sequence
+		if (i + 1 >= len) return INVALID_ESCAPE_SEQUENCE~;
 
-        char escape_char = s[++i];
-        switch (escape_char)
-        {
-            case '"':  result.append_char('"');
-            case '\\': result.append_char('\\');
-            case '/':  result.append_char('/');
-            case 'b':  result.append_char('\b');
-            case 'f':  result.append_char('\f');
-            case 'n':  result.append_char('\n');
-            case 'r':  result.append_char('\r');
-            case 't':  result.append_char('\t');
-            case 'v':  result.append_char('\v');
-            case '0':  result.append_char('\0');
-            case 'x':
-                // Hex escape \xHH
-                if (i + 2 >= len) return INVALID_HEX_ESCAPE?;
-                char h1 = s[++i];
-                char h2 = s[++i];
-                if (!h1.is_xdigit() || !h2.is_xdigit()) return INVALID_HEX_ESCAPE?;
-                uint val = h1 > '9' ? (h1 | 32) - 'a' + 10 : h1 - '0';
-                val = val << 4;
-                val += h2 > '9' ? (h2 | 32) - 'a' + 10 : h2 - '0';
-                result.append_char((char)val);
-            case 'u':
-                // Unicode escape \uHHHH
-                if (i + 4 >= len) return INVALID_UNICODE_ESCAPE?;
-                uint val;
-                for (int j = 0; j < 4; j++)
-                {
-                    char hex_char = s[++i];
-                    if (!hex_char.is_xdigit()) return INVALID_UNICODE_ESCAPE?;
-                    val = val << 4 + (hex_char > '9' ? (hex_char | 32) - 'a' + 10 : hex_char - '0');
-                }
-                result.append_char32(val);
-            case 'U':
-                // Unicode escape \UHHHHHHHH
-                if (i + 8 >= len) return INVALID_UNICODE_ESCAPE?;
-                uint val;
-                for (int j = 0; j < 8; j++)
-                {
-                    char hex_char = s[++i];
-                    if (!hex_char.is_xdigit()) return INVALID_UNICODE_ESCAPE?;
-                    val = val << 4 + (hex_char > '9' ? (hex_char | 32) - 'a' + 10 : hex_char - '0');
-                }
-                result.append_char32(val);
-            default:
-                return INVALID_ESCAPE_SEQUENCE?;
-        }
-    }
-
-    return result.copy_str(allocator);
+		char escape_char = s[++i];
+		switch (escape_char)
+		{
+			case '"':  result.append_char('"');
+			case '\\': result.append_char('\\');
+			case '/':  result.append_char('/');
+			case 'b':  result.append_char('\b');
+			case 'f':  result.append_char('\f');
+			case 'n':  result.append_char('\n');
+			case 'r':  result.append_char('\r');
+			case 't':  result.append_char('\t');
+			case 'v':  result.append_char('\v');
+			case '0':  result.append_char('\0');
+			case 'x':
+				// Hex escape \xHH
+				if (i + 2 >= len) return INVALID_HEX_ESCAPE~;
+				char h1 = s[++i];
+				char h2 = s[++i];
+				if (!h1.is_xdigit() || !h2.is_xdigit()) return INVALID_HEX_ESCAPE~;
+				uint val = h1 > '9' ? (h1 | 32) - 'a' + 10 : h1 - '0';
+				val = val << 4;
+				val += h2 > '9' ? (h2 | 32) - 'a' + 10 : h2 - '0';
+				result.append_char((char)val);
+			case 'u':
+				// Unicode escape \uHHHH
+				if (i + 4 >= len) return INVALID_UNICODE_ESCAPE~;
+				uint val;
+				for (int j = 0; j < 4; j++)
+				{
+					char hex_char = s[++i];
+					if (!hex_char.is_xdigit()) return INVALID_UNICODE_ESCAPE~;
+					val = val << 4 + (hex_char > '9' ? (hex_char | 32) - 'a' + 10 : hex_char - '0');
+				}
+				result.append_char32(val);
+			case 'U':
+				// Unicode escape \UHHHHHHHH
+				if (i + 8 >= len) return INVALID_UNICODE_ESCAPE~;
+				uint val;
+				for (int j = 0; j < 8; j++)
+				{
+					char hex_char = s[++i];
+					if (!hex_char.is_xdigit()) return INVALID_UNICODE_ESCAPE~;
+					val = val << 4 + (hex_char > '9' ? (hex_char | 32) - 'a' + 10 : hex_char - '0');
+				}
+				result.append_char32(val);
+			default:
+				if (!lenient) return INVALID_ESCAPE_SEQUENCE~;
+				result.append_char(escape_char);
+		}
+	}
 }
 
 <*
@@ -202,10 +229,11 @@ fn String? String.unescape(String s, Allocator allocator, bool allow_unquoted =
 
  @param s : "The quoted string to unescape"
  @param allow_unquoted : "Set to true to unescape strings not surrounded by quotes, defaults to false"
+ @param lenient : "Be lenient with escapes, resolving unknown sequences to the escape character, defaults to false"
  @return "The unescaped string without quotes"
  @return? UNTERMINATED_STRING, INVALID_ESCAPE_SEQUENCE, INVALID_HEX_ESCAPE, INVALID_UNICODE_ESCAPE
 *>
-fn String? String.tunescape(String s, bool allow_unquoted = false) => s.unescape(tmem, allow_unquoted);
+fn String? String.tunescape(String s, bool allow_unquoted = false, bool lenient = false) => s.unescape(tmem, allow_unquoted, lenient);
 
 <*
  Check if a character needs to be escaped in a string literal.
@@ -215,19 +243,19 @@ fn String? String.tunescape(String s, bool allow_unquoted = false) => s.unescape
 *>
 fn bool needs_escape(char c)
 {
-    switch (c)
-    {
-        case '"':
-        case '\\':
-        case '\b':
-        case '\f':
-        case '\n':
-        case '\r':
-        case '\t':
-        case '\v':
-        case '\0':
-            return true;
-        default:
-            return c < 32 || c > 126;
-    }
+	switch (c)
+	{
+		case '"':
+		case '\\':
+		case '\b':
+		case '\f':
+		case '\n':
+		case '\r':
+		case '\t':
+		case '\v':
+		case '\0':
+			return true;
+		default:
+			return c < 32 || c > 126;
+	}
 }

lib/std/core/string_iterator.c3

@@ -15,7 +15,7 @@ fn Char32? StringIterator.next(&self)
 {
     usz len = self.utf8.len;
     usz current = self.current;
-    if (current >= len) return NO_MORE_ELEMENT?;
+    if (current >= len) return NO_MORE_ELEMENT~;
     usz read = (len - current < 4 ? len - current : 4);
     Char32 res = conv::utf8_to_char32(&self.utf8[current], &read)!;
     self.current += read;
@@ -26,7 +26,7 @@ fn Char32? StringIterator.peek(&self)
 {
     usz len = self.utf8.len;
     usz current = self.current;
-    if (current >= len) return NO_MORE_ELEMENT?;
+    if (current >= len) return NO_MORE_ELEMENT~;
     usz read = (len - current < 4 ? len - current : 4);
     Char32 res = conv::utf8_to_char32(&self.utf8[current], &read)!;
     return res;
@@ -43,7 +43,7 @@ fn Char32? StringIterator.get(&self)
     usz current = self.current;
     usz read = (len - current < 4 ? len - current : 4);
     usz index = current > read ? current - read : 0;
-    if (index >= len) return NO_MORE_ELEMENT?;
+    if (index >= len) return NO_MORE_ELEMENT~;
     Char32 res = conv::utf8_to_char32(&self.utf8[index], &read)!;
     return res;
 }

lib/std/core/string_to_real.c3

@@ -64,7 +64,7 @@ macro double? decfloat(char[] chars, int $bits, int $emin, int sign)
 		got_rad = true;
 		if (index == last_char)
 		{
-			if (!got_digit) return MALFORMED_FLOAT?;
+			if (!got_digit) return MALFORMED_FLOAT~;
 			return sign * 0.0;
 		}
 		if (index != last_char && (c = chars[++index]) == '0')
@@ -83,7 +83,7 @@ macro double? decfloat(char[] chars, int $bits, int $emin, int sign)
 		switch
 		{
 			case c == '.':
-				if (got_rad) return MALFORMED_FLOAT?;
+				if (got_rad) return MALFORMED_FLOAT~;
 				got_rad = true;
 				lrp = dc;
 			case k < KMAX - 3:
@@ -113,24 +113,24 @@ macro double? decfloat(char[] chars, int $bits, int $emin, int sign)
 		c = chars[++index];
 	}
 	if (!got_rad) lrp = dc;
-	if (!got_digit) return MALFORMED_FLOAT?;
+	if (!got_digit) return MALFORMED_FLOAT~;
 	if ((c | 32) == 'e')
 	{
-		if (last_char == index) return MALFORMED_FLOAT?;
-		long e10 = String.to_long((String)chars[index + 1..]) ?? MALFORMED_FLOAT?!;
+		if (last_char == index) return MALFORMED_FLOAT~;
+		long e10 = String.to_long((String)chars[index + 1..]) ?? MALFORMED_FLOAT~!;
 		lrp += e10;
 	}
 	else if (index != last_char)
 	{
-		return MALFORMED_FLOAT?;
+		return 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 FLOAT_OUT_OF_RANGE?;
-	if (lrp < $emin - 2 * math::DOUBLE_MANT_DIG) return FLOAT_OUT_OF_RANGE?;
+	if (lrp > - $emin / 2) return FLOAT_OUT_OF_RANGE~;
+	if (lrp < $emin - 2 * math::DOUBLE_MANT_DIG) return FLOAT_OUT_OF_RANGE~;
 
 	// Align incomplete final B1B digit
 	if (j)
@@ -320,7 +320,7 @@ macro double? decfloat(char[] chars, int $bits, int $emin, int sign)
 			y *= 0.5;
 			e2++;
 		}
-		if (e2 + math::DOUBLE_MANT_DIG > emax || (denormal && frac)) return MALFORMED_FLOAT?;
+		if (e2 + math::DOUBLE_MANT_DIG > emax || (denormal && frac)) return MALFORMED_FLOAT~;
 	}
 	return math::scalbn(y, e2);
 }
@@ -351,7 +351,7 @@ macro double? hexfloat(char[] chars, int $bits, int $emin, int sign)
 		got_rad = true;
 		if (index == last_char)
 		{
-			if (!got_digit) return MALFORMED_FLOAT?;
+			if (!got_digit) return MALFORMED_FLOAT~;
 			return sign * 0.0;
 		}
 		if (index != last_char && (c = chars[++index]) == '0')
@@ -369,7 +369,7 @@ macro double? hexfloat(char[] chars, int $bits, int $emin, int sign)
 	{
 		if (c == '.')
 		{
-			if (got_rad) return MALFORMED_FLOAT?;
+			if (got_rad) return MALFORMED_FLOAT~;
 			got_rad = true;
 			rp = dc;
 		}
@@ -393,20 +393,20 @@ macro double? hexfloat(char[] chars, int $bits, int $emin, int sign)
 		if (index == last_char) break;
 		c = chars[++index];
 	}
-	if (!got_digit) return MALFORMED_FLOAT?;
+	if (!got_digit) return 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..]) ?? (MALFORMED_FLOAT?)!;
+		long e2val = String.to_long((String)chars[index + 1..]) ?? MALFORMED_FLOAT~!;
 		e2 = e2val;
 	}
 	e2 += 4 * rp - 32;
 	if (!x) return sign * 0.0;
-	if (e2 > -$emin) return FLOAT_OUT_OF_RANGE?;
-	if (e2 < $emin - 2 * math::DOUBLE_MANT_DIG) return FLOAT_OUT_OF_RANGE?;
+	if (e2 > -$emin) return FLOAT_OUT_OF_RANGE~;
+	if (e2 < $emin - 2 * math::DOUBLE_MANT_DIG) return FLOAT_OUT_OF_RANGE~;
 
 	while (x < 0x80000000)
 	{
@@ -441,7 +441,7 @@ macro double? hexfloat(char[] chars, int $bits, int $emin, int sign)
 	}
 	y = bias + sign * (double)x + sign * y;
 	y -= bias;
-	if (!y) return FLOAT_OUT_OF_RANGE?;
+	if (!y) return FLOAT_OUT_OF_RANGE~;
 
 	return math::scalbn(y, (int)e2);
 }
@@ -463,7 +463,7 @@ macro String.to_real(chars, $Type) @private
 	$endswitch
 
 	chars = chars.trim();
-	if (!chars.len) return MALFORMED_FLOAT?;
+	if (!chars.len) return MALFORMED_FLOAT~;
 
 	if (chars.len != 1)
 	{
@@ -477,7 +477,7 @@ macro String.to_real(chars, $Type) @private
 		}
 	}
 	chars = chars.trim();
-	if (!chars.len) return MALFORMED_FLOAT?;
+	if (!chars.len) return MALFORMED_FLOAT~;
 
 	if (chars == "infinity" || chars == "INFINITY") return sign * $Type.inf;
 	if (chars == "NAN" || chars == "nan") return $Type.nan;

lib/std/core/test.c3

@@ -12,7 +12,7 @@ faultdef DIVISION_BY_ZERO;
 
 fn double? divide(int a, int b)
 {
-	if (b == 0) return MathError.DIVISION_BY_ZERO?;
+	if (b == 0) return MathError.DIVISION_BY_ZERO~;
 	return (double)(a) / (double)(b);
 }
 

lib/std/core/types.c3

@@ -29,47 +29,47 @@ macro any_to_int(any v, $Type)
 	{
 		case ichar:
 			ichar c = *(char*)v.ptr;
-			if (is_mixed_signed && c < 0) return VALUE_OUT_OF_UNSIGNED_RANGE?;
+			if (is_mixed_signed && c < 0) return VALUE_OUT_OF_UNSIGNED_RANGE~;
 			return ($Type)c;
 		case short:
 			short s = *(short*)v.ptr;
-			if (is_mixed_signed && s < 0) return VALUE_OUT_OF_UNSIGNED_RANGE?;
-			if (s > max || s < min) return VALUE_OUT_OF_RANGE?;
+			if (is_mixed_signed && s < 0) return VALUE_OUT_OF_UNSIGNED_RANGE~;
+			if (s > max || s < min) return VALUE_OUT_OF_RANGE~;
 			return ($Type)s;
 		case int:
 			int i = *(int*)v.ptr;
-			if (is_mixed_signed && i < 0) return VALUE_OUT_OF_UNSIGNED_RANGE?;
-			if (i > max || i < min) return VALUE_OUT_OF_RANGE?;
+			if (is_mixed_signed && i < 0) return VALUE_OUT_OF_UNSIGNED_RANGE~;
+			if (i > max || i < min) return VALUE_OUT_OF_RANGE~;
 			return ($Type)i;
 		case long:
 			long l = *(long*)v.ptr;
-			if (is_mixed_signed && l < 0) return VALUE_OUT_OF_UNSIGNED_RANGE?;
-			if (l > max || l < min) return VALUE_OUT_OF_RANGE?;
+			if (is_mixed_signed && l < 0) return VALUE_OUT_OF_UNSIGNED_RANGE~;
+			if (l > max || l < min) return VALUE_OUT_OF_RANGE~;
 			return ($Type)l;
 		case int128:
 			int128 i = *(int128*)v.ptr;
-			if (is_mixed_signed && i < 0) return VALUE_OUT_OF_UNSIGNED_RANGE?;
-			if (i > max || i < min) return VALUE_OUT_OF_RANGE?;
+			if (is_mixed_signed && i < 0) return VALUE_OUT_OF_UNSIGNED_RANGE~;
+			if (i > max || i < min) return VALUE_OUT_OF_RANGE~;
 			return ($Type)i;
 		case char:
 			char c = *(char*)v.ptr;
-			if (c > max) return VALUE_OUT_OF_RANGE?;
+			if (c > max) return VALUE_OUT_OF_RANGE~;
 			return ($Type)c;
 		case ushort:
 			ushort s = *(ushort*)v.ptr;
-			if (s > max || s < min) return VALUE_OUT_OF_RANGE?;
+			if (s > max || s < min) return VALUE_OUT_OF_RANGE~;
 			return ($Type)s;
 		case uint:
 			uint i = *(uint*)v.ptr;
-			if (i > max || i < min) return VALUE_OUT_OF_RANGE?;
+			if (i > max || i < min) return VALUE_OUT_OF_RANGE~;
 			return ($Type)i;
 		case ulong:
 			ulong l = *(ulong*)v.ptr;
-			if (l > max || l < min) return VALUE_OUT_OF_RANGE?;
+			if (l > max || l < min) return VALUE_OUT_OF_RANGE~;
 			return ($Type)l;
 		case uint128:
 			uint128 i = *(uint128*)v.ptr;
-			if (i > max || i < min) return VALUE_OUT_OF_RANGE?;
+			if (i > max || i < min) return VALUE_OUT_OF_RANGE~;
 			return ($Type)i;
 		default:
 			unreachable();
@@ -329,8 +329,8 @@ macro lower_to_atomic_compatible_type($Type) @const
   $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_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, $Type2) @const
 {
@@ -345,7 +345,7 @@ macro bool has_equals($Type) @const => $defined(($Type){} == ($Type){});
 
 macro bool is_equatable_type($Type) @const
 {
-	$if $defined($Type.less) || $defined($Type.compare_to) || $defined($Type.equals):
+	$if $defined($Type.less) ||| $defined($Type.compare_to) ||| $defined($Type.equals):
 		return true;
 	$else
 		return $Type.is_eq;
@@ -357,7 +357,7 @@ macro bool is_equatable_type($Type) @const
 *>
 macro bool implements_copy($Type) @const
 {
-	return $defined($Type.copy) && $defined($Type.free);
+	return $defined($Type.copy) &&& $defined($Type.free);
 }
 
 macro bool @equatable_value(#value) @const
@@ -367,7 +367,7 @@ macro bool @equatable_value(#value) @const
 
 macro bool @comparable_value(#value) @const
 {
-	$if $defined(#value.less) || $defined(#value.compare_to):
+	$if $defined(#value.less) ||| $defined(#value.compare_to):
 		return true;
 	$else
 		return $typeof(#value).is_ordered;

lib/std/crypto/chacha20.c3

@@ -0,0 +1,233 @@
+// Copyright (c) 2025 Zack Puhl <github@xmit.xyz>. 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.
+//
+// ChaCha20 code dedicated from repo: https://github.com/NotsoanoNimus/chacha20_aead.c3l (but massively cleaned)
+module std::crypto::chacha20;
+
+
+<* The typical cipher block size in bytes. *>
+const BLOCK_SIZE = 64;
+
+<* Required key size in bytes. *>
+const KEY_SIZE = 32;
+
+<* ChaCha20 "nonce" (initialization vector) size. *>
+const NONCE_SIZE = 12;
+
+<* A required ChaCha20 "magic" value used for state initialization. *>
+const char[] MAGIC = "expand 32-byte k";
+
+<*
+ Once a single ChaCha20 context has processed this many bytes, a new nonce MUST be used,
+ unless the static `permit_overflow` runtime module variable is set to true.
+*>
+const CHACHA20_NONCE_REUSE_LIMIT = 64 * (1ull << 32);
+
+<*
+ SECURITY WARNING:
+ This boolean should always remain 'false'. If set to 'true', you accept the security
+ implications of nonce re-use caused by an overflow in the cipher's  'counter' field.
+
+ This security warning is only applicable when a single ChaCha20 context is being used
+ to process more than about 256 GiB of data.
+*>
+bool permit_overflow = false;
+
+
+<* A context structure used to track an ongoing ChaCha20 transformation. *>
+struct ChaCha20
+{
+	<* The position within a block before permuting the rounds. *>
+	usz position;
+	<* Count of bytes processed. Useful to track an approach to the 256GiB limit of a single context. *>
+	ulong bytes_processed;
+	<* The key stream or state used during cipher block operations. *>
+	uint[16] key_stream @align(ulong.sizeof);
+	<* The secret key for the context. *>
+	char[32] key;
+	<* The one-time nonce (or IV - initialization vector) used for the context. *>
+	char[12] nonce;
+	<* Internal state of the cipher. *>
+	uint[16] state;
+}
+
+
+<* The meat and potatoes of the ChaCha20 stream cipher. *>
+macro quarter_round(uint* x, int a, int b, int c, int d) @local
+{
+	x[a] += x[b]; x[d] = (x[d] ^ x[a]).rotl(16);
+	x[c] += x[d]; x[b] = (x[b] ^ x[c]).rotl(12);
+	x[a] += x[b]; x[d] = (x[d] ^ x[a]).rotl(8);
+	x[c] += x[d]; x[b] = (x[b] ^ x[c]).rotl(7);
+}
+
+<* Process the next (or final) chunk of ingested data. *>
+fn void ChaCha20.mutate_keystream(&self) @local @inline
+{
+	self.key_stream[..] = self.state[..];
+
+	for (usz i = 0; i < 10; i++) // unrolling this does not improve performance measurably
+	{
+		quarter_round(&self.key_stream[0],   0,  4,  8, 12);
+		quarter_round(&self.key_stream[0],   1,  5,  9, 13);
+		quarter_round(&self.key_stream[0],   2,  6, 10, 14);
+		quarter_round(&self.key_stream[0],   3,  7, 11, 15);
+		quarter_round(&self.key_stream[0],   0,  5, 10, 15);
+		quarter_round(&self.key_stream[0],   1,  6, 11, 12);
+		quarter_round(&self.key_stream[0],   2,  7,  8, 13);
+		quarter_round(&self.key_stream[0],   3,  4,  9, 14);
+	}
+
+	// NOTE: This would 'feel' like a performance hit, but testing the benchmark doesn't show any noticeable
+	//   difference on -O5 between this and a for-loop, or even an unrolled loop with compile-time '$for'.
+	array::@zip_into(self.key_stream[..], self.state[..], fn (a, b) => a + b);
+
+	self.state[12]++; // increment the block counter (rollovers are ok)
+}
+
+<*
+ Initialize a ChaCha20 transformation context.
+
+ @param key : `The secret key used for the transformation operation.`
+ @param nonce : `The one-time nonce to use for the transformation operation.`
+ @param counter : `An optional counter value to adjust the stream's position.`
+
+ @require key.len == KEY_SIZE : `Input key slice is not the correct length (32 bytes).`
+ @require nonce.len ==  NONCE_SIZE : `Input nonce slice is not the correct length (12 bytes).`
+*>
+fn void ChaCha20.init(&self, char[KEY_SIZE] key, char[NONCE_SIZE] nonce, uint counter = 0)
+{
+	// Init block.
+	self.position = BLOCK_SIZE; // start at the "end" of a block on init
+	self.bytes_processed = 0;
+	self.key[..] = key[..];
+	self.nonce[..] = nonce[..];
+	((char*)&self.state[0])[:MAGIC.len] = MAGIC[..];
+	((char*)&self.state[4])[:KEY_SIZE] = key[..];
+	self.state[12] = counter;
+	((char*)&self.state[13])[:NONCE_SIZE] = nonce[..];
+}
+
+<*
+ Transform some input data using the current context structure.
+
+ @param[inout] data : `The data to transform (encrypt or decrypt).`
+*>
+fn void ChaCha20.transform(&self, char[] data)
+{
+	if (!data.len) return;
+
+	usz original_length = data.len;
+	char[] key_stream = @as_char_view(self.key_stream);
+
+	// 1. Process remaining bytes in the current keystream block.
+	if (self.position < BLOCK_SIZE)
+	{
+		usz len = data.len < (BLOCK_SIZE - self.position) ? data.len : (BLOCK_SIZE - self.position);
+		for (usz i = 0; i < len; i++) data[i] ^= key_stream[self.position + i];
+		self.position += len;
+		data = data[len..];
+	}
+
+	// 2. Get the amount of bytes offset from the nearest alignment boundary.
+	//   Process full blocks at a time, word by word according to the system's architecture.
+	//   Any extra bytes on each side are dynamically processed byte-by-byte.
+    usz offset = usz.sizeof - (((usz)data.ptr % usz.sizeof) ?: usz.sizeof);
+
+	for (usz x = offset; data.len >= BLOCK_SIZE; data = data[BLOCK_SIZE..], x = offset)
+	{
+		self.mutate_keystream();
+        if (offset) foreach (i, &b : data[:offset]) *b ^= key_stream[i];
+        char[] aligned_data = data[offset..];
+		for (; x <= (BLOCK_SIZE - usz.sizeof); x += usz.sizeof)
+		{
+			((usz*)aligned_data.ptr)[x / usz.sizeof] ^= mem::load((usz*)(&key_stream[x]), $align: 1);
+		}
+        for (; x < BLOCK_SIZE; x++) data[x] ^= key_stream[x];
+	}
+
+	// 3. Process any remaining bytes.
+	if (data.len > 0)
+	{
+		self.mutate_keystream();
+		for (usz i = 0; i < data.len; i++) data[i] ^= key_stream[i];
+		self.position = data.len;
+	}
+
+	// All done. Capture the transformed length of data and check limits.
+	self.bytes_processed += original_length;
+	if (@unlikely(self.bytes_processed >= CHACHA20_NONCE_REUSE_LIMIT && !permit_overflow))
+	{
+		abort(
+			"ChaCha20 transform limit (~256 GiB) exceeded. You can set 'chacha20::permit_overflow = true;' at"
+			" runtime to disable this panic, but you accept the terrible SECURITY IMPLICATIONS of doing so."
+		);
+	}
+}
+
+<* Destroy the current context structure by zeroing all fields. *>
+fn void ChaCha20.destroy(&self) => mem::zero_volatile(@as_char_view(*self));
+
+
+<*
+ Perform an in-place transformation of some data in a buffer, without cloning the data to a new buffer.
+
+ @param[inout] data : `The data to transform (encrypt or decrypt).`
+ @param key : `The secret key used for the transformation operation.`
+ @param nonce : `The one-time nonce to use for the transformation operation.`
+ @param counter : `An optional counter value to adjust the stream's position.`
+
+ @require key.len == KEY_SIZE : `Input key slice is not the correct length (32 bytes).`
+ @require nonce.len ==  NONCE_SIZE : `Input nonce slice is not the correct length (12 bytes).`
+*>
+fn void crypt(char[] data, char[KEY_SIZE] key, char[NONCE_SIZE] nonce, uint counter = 0) @private
+{
+	if (@unlikely(!data.len)) return;
+	ChaCha20 c @noinit;
+	defer c.destroy();
+	c.init(key, nonce, counter);
+	c.transform(data);
+}
+alias encrypt_mut = crypt;
+alias decrypt_mut = crypt;
+
+<*
+ Perform a transformation of some data cloned from a source buffer.
+
+ @param[&inout] allocator : `The memory allocator which controls allocation of the cloned input data.`
+ @param[inout] data : `The data to transform (encrypt or decrypt).`
+ @param key : `The secret key used for the transformation operation.`
+ @param nonce : `The one-time nonce to use for the transformation operation.`
+ @param counter : `An optional counter value to adjust the stream's position.`
+
+ @require key.len == KEY_SIZE : `Input key slice is not the correct length (32 bytes).`
+ @require nonce.len ==  NONCE_SIZE : `Input nonce slice is not the correct length (12 bytes).`
+*>
+fn char[] crypt_clone(Allocator allocator, char[] data, char[KEY_SIZE] key, char[NONCE_SIZE] nonce, uint counter = 0) @private
+{
+	if (@unlikely(!data.len)) return {};
+	char[] buff = allocator::clone_slice(allocator, data);
+	crypt(buff, key, nonce, counter);
+	return buff;
+}
+alias encrypt = crypt_clone;
+alias decrypt = crypt_clone;
+
+<*
+ Perform a transformation of some data cloned from a source buffer by the temp allocator.
+
+ @param[inout] data : `The data to transform (encrypt or decrypt).`
+ @param key : `The secret key used for the transformation operation.`
+ @param nonce : `The one-time nonce to use for the transformation operation.`
+ @param counter : `An optional counter value to adjust the stream's position.`
+
+ @require key.len == KEY_SIZE : `Input key slice is not the correct length (32 bytes).`
+ @require nonce.len ==  NONCE_SIZE : `Input nonce slice is not the correct length (12 bytes).`
+*>
+fn char[] tcrypt_clone(char[] data, char[KEY_SIZE] key, char[NONCE_SIZE] nonce, uint counter = 0) @private
+{
+	return crypt_clone(tmem, data, key, nonce, counter);
+}
+alias tencrypt = tcrypt_clone;
+alias tdecrypt = tcrypt_clone;

lib/std/encoding/base32.c3

@@ -101,12 +101,12 @@ fn char[]? decode_buffer(char[] src, char[] dst, char padding = DEFAULT_PAD, Bas
 		{
 			if (src.len == 0)
 			{
-				if (padding > 0) return encoding::INVALID_PADDING?;
+				if (padding > 0) return encoding::INVALID_PADDING~;
 				break;
 			}
 			if (src[0] == padding) break;
 			buf[i] = alphabet.reverse[src[0]];
-			if (buf[i] == INVALID) return encoding::INVALID_CHARACTER?;
+			if (buf[i] == INVALID) return encoding::INVALID_CHARACTER~;
 			src = src[1..];
 		}
 
@@ -150,7 +150,7 @@ fn char[]? decode_buffer(char[] src, char[] dst, char padding = DEFAULT_PAD, Bas
 				dst[0] = buf[1] >> 2 | buf[0] << 3;
 				n++;
 			default:
-				return encoding::INVALID_CHARACTER?;
+				return encoding::INVALID_CHARACTER~;
 		}
 		if (dst.len < 5) break;
 		dst = dst[5..];

lib/std/encoding/base64.c3

@@ -87,11 +87,11 @@ fn usz? decode_len(usz n, char padding)
 	usz trailing = n % 4;
 	if (padding)
 	{
-		if (trailing != 0) return encoding::INVALID_PADDING?;
+		if (trailing != 0) return encoding::INVALID_PADDING~;
 		// source size is multiple of 4
 		return dn;
 	}
-	if (trailing == 1) return encoding::INVALID_PADDING?;
+	if (trailing == 1) return encoding::INVALID_PADDING~;
 	return dn + trailing * 3 / 4;
 }
 
@@ -196,7 +196,7 @@ fn char[]? decode_buffer(char[] src, char[] dst, char padding = DEFAULT_PAD, Bas
 			case c1:
 			case c2:
 			case c3:
-				return encoding::INVALID_CHARACTER?;
+				return encoding::INVALID_CHARACTER~;
 		}
 		uint group = (uint)c0 << 18 | (uint)c1 << 12 | (uint)c2 << 6 | (uint)c3;
 		dst[0] = (char)(group >> 16);
@@ -211,7 +211,7 @@ fn char[]? decode_buffer(char[] src, char[] dst, char padding = DEFAULT_PAD, Bas
 	src = src[^trailing..];
 	char c0 = alphabet.reverse[src[0]];
 	char c1 = alphabet.reverse[src[1]];
-	if (c0 == 0xFF || c1 == 0xFF) return encoding::INVALID_PADDING?;
+	if (c0 == 0xFF || c1 == 0xFF) return encoding::INVALID_PADDING~;
 	if (!padding)
 	{
 		switch (src.len)
@@ -221,7 +221,7 @@ fn char[]? decode_buffer(char[] src, char[] dst, char padding = DEFAULT_PAD, Bas
 				dst[0] = (char)(group >> 16);
 			case 3:
 				char c2 = alphabet.reverse[src[2]];
-				if (c2 == 0xFF) return encoding::INVALID_CHARACTER?;
+				if (c2 == 0xFF) return encoding::INVALID_CHARACTER~;
 				uint group = (uint)c0 << 18 | (uint)c1 << 12 | (uint)c2 << 6;
 				dst[0] = (char)(group >> 16);
 				dst[1] = (char)(group >> 8);
@@ -235,13 +235,13 @@ fn char[]? decode_buffer(char[] src, char[] dst, char padding = DEFAULT_PAD, Bas
 		switch (padding)
 		{
 			case src[2]:
-				if (src[3] != padding) return encoding::INVALID_PADDING?;
+				if (src[3] != padding) return encoding::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 encoding::INVALID_CHARACTER?;
+				if (c2 == 0xFF) return encoding::INVALID_CHARACTER~;
 				uint group = (uint)c0 << 18 | (uint)c1 << 12 | (uint)c2 << 6;
 				dst[0] = (char)(group >> 16);
 				dst[1] = (char)(group >> 8);

lib/std/encoding/csv.c3

@@ -79,7 +79,7 @@ macro void? @each_row(InStream stream, String separator = ",", int max_rows = in
 			if (catch err = s)
 			{
 				if (err == io::EOF) return;
-				return err?;
+				return err~;
 			}
 			@body(s.split(mem, separator));
 		};

lib/std/encoding/hex.c3

@@ -81,7 +81,7 @@ fn usz? decode_bytes(char[] src, char[] dst)
 	{
 		char a = HEXREVERSE[src[j - 1]];
 		char b = HEXREVERSE[src[j]];
-		if (a > 0x0f || b > 0x0f) return encoding::INVALID_CHARACTER?;
+		if (a > 0x0f || b > 0x0f) return encoding::INVALID_CHARACTER~;
 		dst[i] = (a << 4) | b;
 		i++;
 	}

lib/std/encoding/json.c3

@@ -85,13 +85,13 @@ fn Object*? parse_from_token(JsonContext* context, JsonTokenType token) @local
 		case COMMA:
 		case RBRACE:
 		case RBRACKET:
-		case COLON: return UNEXPECTED_CHARACTER?;
+		case COLON: return 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 io::EOF?;
+		case EOF: return io::EOF~;
 	}
 }
 fn Object*? parse_any(JsonContext* context) @local
@@ -117,7 +117,7 @@ fn JsonTokenType? lex_number(JsonContext *context, char c) @local
 			c = read_next(context)!;
 			if (leading_zero)
 			{
-				if (c.is_digit()) return INVALID_NUMBER?;
+				if (c.is_digit()) return INVALID_NUMBER~;
 				leading_zero = false;
 			}
 		}
@@ -140,7 +140,7 @@ fn JsonTokenType? lex_number(JsonContext *context, char c) @local
 					t.append(c);
 					c = read_next(context)!;
 			}
-			if (!c.is_digit()) return INVALID_NUMBER?;
+			if (!c.is_digit()) return INVALID_NUMBER~;
 			while (c.is_digit())
 			{
 				t.append(c);
@@ -148,7 +148,7 @@ fn JsonTokenType? lex_number(JsonContext *context, char c) @local
 			}
 		}
 		pushback(context, c);
-		double? d = t.str_view().to_double() ?? INVALID_NUMBER?;
+		double? d = t.str_view().to_double() ?? INVALID_NUMBER~;
 		context.last_number = d!;
 		return NUMBER;
 	};
@@ -160,14 +160,14 @@ fn Object*? parse_map(JsonContext* context) @local
 	defer catch map.free();
 	JsonTokenType token = advance(context)!;
 	defer context.depth--;
-	if (++context.depth >= max_depth) return json::MAX_DEPTH_REACHED?;
+	if (++context.depth >= max_depth) return json::MAX_DEPTH_REACHED~;
 
 	@stack_mem(256; Allocator mem)
 	{
 		DString temp_key = dstring::new_with_capacity(mem, 32);
 		while (token != JsonTokenType.RBRACE)
 		{
-			if (token != JsonTokenType.STRING) return UNEXPECTED_CHARACTER?;
+			if (token != JsonTokenType.STRING) return UNEXPECTED_CHARACTER~;
 			DString string = context.last_string;
 			// Copy the key to our temp holder, since our
 			// last_string may be used in parse_any
@@ -182,7 +182,7 @@ fn Object*? parse_map(JsonContext* context) @local
 				token = advance(context)!;
 				continue;
 			}
-			if (token != JsonTokenType.RBRACE) return UNEXPECTED_CHARACTER?;
+			if (token != JsonTokenType.RBRACE) return UNEXPECTED_CHARACTER~;
 		}
 		return map;
 	};
@@ -193,7 +193,7 @@ fn Object*? parse_array(JsonContext* context) @local
 	Object* list = object::new_obj(context.allocator);
 	defer catch list.free();
 	defer context.depth--;
-	if (++context.depth >= max_depth) return json::MAX_DEPTH_REACHED?;
+	if (++context.depth >= max_depth) return json::MAX_DEPTH_REACHED~;
 	JsonTokenType token = advance(context)!;
 	while (token != JsonTokenType.RBRACKET)
 	{
@@ -205,7 +205,7 @@ fn Object*? parse_array(JsonContext* context) @local
 			token = advance(context)!;
 			continue;
 		}
-		if (token != JsonTokenType.RBRACKET) return UNEXPECTED_CHARACTER?;
+		if (token != JsonTokenType.RBRACKET) return UNEXPECTED_CHARACTER~;
 	}
 	return list;
 }
@@ -236,7 +236,7 @@ fn char? read_next(JsonContext* context) @local
 			context.reached_end = true;
 			return '\0';
 		}
-		return err?;
+		return err~;
 	}
 	if (c == 0)
 	{
@@ -249,7 +249,7 @@ fn JsonTokenType? advance(JsonContext* context) @local
 {
 	char c;
 	// Skip whitespace
-	while WS: (c = read_next(context)!)
+	while WS: ((c = read_next(context)!))
 	{
 		switch (c)
 		{
@@ -272,12 +272,12 @@ fn JsonTokenType? advance(JsonContext* context) @local
 				while COMMENT: (true)
 				{
 					// Skip to */
-					while (c = read_next(context)!)
+					while ((c = read_next(context)!))
 					{
 						if (c == '\n') context.line++;
 						if (c != '*') continue;
 						// Skip through all the '*'
-						while (c = read_next(context)!)
+						while ((c = read_next(context)!))
 						{
 							if (c == '\n') context.line++;
 							if (c != '*') break;
@@ -293,7 +293,7 @@ fn JsonTokenType? advance(JsonContext* context) @local
 	switch (c)
 	{
 		case '\0':
-			return io::EOF?;
+			return io::EOF~;
 		case '{':
 			return LBRACE;
 		case '}':
@@ -321,7 +321,7 @@ fn JsonTokenType? advance(JsonContext* context) @local
 			match(context, "ull")!;
 			return NULL;
 		default:
-			return UNEXPECTED_CHARACTER?;
+			return UNEXPECTED_CHARACTER~;
 	}
 }
 
@@ -330,13 +330,13 @@ fn void? match(JsonContext* context, String str) @local
 	foreach (c : str)
 	{
 		char l = read_next(context)!;
-		if (l != c) return UNEXPECTED_CHARACTER?;
+		if (l != c) return UNEXPECTED_CHARACTER~;
 	}
 }
 
 fn void? parse_expected(JsonContext* context, JsonTokenType token) @local
 {
-	if (advance(context)! != token) return UNEXPECTED_CHARACTER?;
+	if (advance(context)! != token) return UNEXPECTED_CHARACTER~;
 }
 
 fn JsonTokenType? lex_string(JsonContext* context)
@@ -348,9 +348,9 @@ fn JsonTokenType? lex_string(JsonContext* context)
 		switch (c)
 		{
 			case '\0':
-				return io::EOF?;
+				return io::EOF~;
 			case 1..31:
-				return UNEXPECTED_CHARACTER?;
+				return UNEXPECTED_CHARACTER~;
 			case '"':
 				break LOOP;
 			case '\\':
@@ -363,9 +363,9 @@ fn JsonTokenType? lex_string(JsonContext* context)
 		switch (c)
 		{
 			case '\0':
-				return io::EOF?;
+				return io::EOF~;
 			case 1..31:
-				return UNEXPECTED_CHARACTER?;
+				return UNEXPECTED_CHARACTER~;
 			case '"':
 			case '\\':
 			case '/':
@@ -385,13 +385,13 @@ fn JsonTokenType? lex_string(JsonContext* context)
 				for (int i = 0; i < 4; i++)
 				{
 					c = read_next(context)!;
-					if (!c.is_xdigit()) return INVALID_ESCAPE_SEQUENCE?;
+					if (!c.is_xdigit()) return INVALID_ESCAPE_SEQUENCE~;
 					val = val << 4 + (c > '9' ? (c | 32) - 'a' + 10 : c - '0');
 				}
 				context.last_string.append_char32(val);
 				continue;
 			default:
-				 return INVALID_ESCAPE_SEQUENCE?;
+				 return INVALID_ESCAPE_SEQUENCE~;
 		}
 		context.last_string.append(c);
 	}

lib/std/experimental/FrameScheduler.c3

@@ -1,4 +1,4 @@
-module std::experimental::scheduler{Event};
+module std::experimental::scheduler <Event>;
 import std::collections, std::thread, std::time;
 
 struct DelayedSchedulerEvent @local
@@ -43,7 +43,7 @@ macro void FrameScheduler.@destroy(&self; @destruct(Event e))
 	self.events.free();
 	self.pending_events.free();
 	self.delayed_events.free();
-	(void)self.mtx.destroy();
+	self.mtx.destroy();
 }
 
 fn void FrameScheduler.queue_delayed_event(&self, Event event, Duration delay)
@@ -76,7 +76,7 @@ fn Event? FrameScheduler.pop_event(&self)
 	while (true)
 	{
 		if (try event = self.events.pop()) return event;
-    	if (!@atomic_load(self.pending)) return NO_MORE_ELEMENT?;
+    	if (!@atomic_load(self.pending)) return NO_MORE_ELEMENT~;
         self.mtx.@in_lock()
         {
             self.events.add_all(&self.pending_events);
@@ -88,7 +88,7 @@ fn Event? FrameScheduler.pop_event(&self)
                 self.events.push(self.delayed_events.pop()!!);
             }
             @atomic_store(self.pending, self.delayed_events.len() > 0);
-            if (!self.events.len()) return NO_MORE_ELEMENT?;
+            if (!self.events.len()) return NO_MORE_ELEMENT~;
         };
 	}
 }

lib/std/hash/a5hash.c3

@@ -62,8 +62,8 @@ fn ulong hash(char[] data, ulong seed = 0)
 			for (; data.len > 16; data = data[16..])
 			{
 				@a5mul(
-					@unaligned_load(((ulong*)data.ptr)[0], 1) ^ seed1,
-					@unaligned_load(((ulong*)data.ptr)[1], 1) ^ seed2,
+					mem::load((ulong*)data.ptr, 1) ^ seed1,
+					mem::load((ulong*)data.ptr + 1, 1) ^ seed2,
 					seed1, seed2
 				);
 
@@ -71,16 +71,16 @@ fn ulong hash(char[] data, ulong seed = 0)
 				seed2 += val10;
 			}
 
-			a = @unaligned_load(*(ulong*)(data.ptr + (uptr)data.len - 16), 1);
-			b = @unaligned_load(*(ulong*)(data.ptr + (uptr)data.len - 8), 1);
+			a = mem::load((ulong*)(data.ptr + (uptr)data.len - 16), 1);
+			b = mem::load((ulong*)(data.ptr + (uptr)data.len - 8), 1);
 		}
 		else
 		{
-			a = ((ulong)@unaligned_load(*(uint*)&data[0], 1) << 32)
-				| @unaligned_load(*(uint*)&data[^4], 1);
+			a = ((ulong)mem::load((uint*)&data[0], 1) << 32)
+				| mem::load((uint*)&data[^4], 1);
 
-			b = ((ulong)@unaligned_load(*(uint*)&data[(data.len >> 3) * 4], 1) << 32)
-				| @unaligned_load(*(uint*)(data.ptr + data.len - 4 - (data.len >> 3) * 4), 1);
+			b = ((ulong)mem::load((uint*)&data[(data.len >> 3) * 4], 1) << 32)
+				| mem::load((uint*)(data.ptr + data.len - 4 - (data.len >> 3) * 4), 1);
 		}
 	}
 	else

lib/std/hash/blake2.c3

@@ -0,0 +1,383 @@
+// Copyright (c) 2025 Zack Puhl <github@xmit.xyz>. 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.
+//
+// This is based on the original BLAKE2 reference implementation, but it's been
+//   significantly changed. You can really see how C3's features shine over the
+//   original C source.
+//
+// You'll note that this may have been better done with generic modules. I'm not
+//   an expert, but I'd argue that it's unnecessary for this particular algorithm.
+//   There are only two important versions of the algorithm to implement in a "common"
+//   (generic) way, and the output size of each type is controlled rather naturally.
+//
+module std::hash::blake2;
+
+
+// Common hash output sizes. The library is NOT limited to these six output sizes, but
+//   these instead stand in for what would have been "magic numbers" throughout the code.
+const SIZE_128 = 16;
+const SIZE_160 = 20;
+const SIZE_224 = 28;
+const SIZE_256 = 32;
+const SIZE_384 = 48;
+const SIZE_512 = 64;
+
+
+macro @g(r, m, $s, i, #a, #b, #c, #d) @local
+{
+	// I really dislike using these conditional one-liners, but... it's the only difference between 2s and 2b besides buffer sizes (mainly).
+	#a += #b + m[$s[r][2*i+0]];
+	#d = (#d ^ #a).rotr($sizeof(#a) == ulong.sizeof ??? 32 : 16);
+	#c += #d;
+	#b = (#b ^ #c).rotr($sizeof(#a) == ulong.sizeof ??? 24 : 12);
+	#a += #b + m[$s[r][2*i+1]];
+	#d = (#d ^ #a).rotr($sizeof(#a) == ulong.sizeof ??? 16 : 8);
+	#c += #d;
+	#b = (#b ^ #c).rotr($sizeof(#a) == ulong.sizeof ??? 63 : 7);
+}
+
+macro @round(r, m, $s, #v) @local
+{
+	@g(r, m, $s, 0, #v[ 0], #v[ 4], #v[ 8], #v[12]);
+	@g(r, m, $s, 1, #v[ 1], #v[ 5], #v[ 9], #v[13]);
+	@g(r, m, $s, 2, #v[ 2], #v[ 6], #v[10], #v[14]);
+	@g(r, m, $s, 3, #v[ 3], #v[ 7], #v[11], #v[15]);
+	@g(r, m, $s, 4, #v[ 0], #v[ 5], #v[10], #v[15]);
+	@g(r, m, $s, 5, #v[ 1], #v[ 6], #v[11], #v[12]);
+	@g(r, m, $s, 6, #v[ 2], #v[ 7], #v[ 8], #v[13]);
+	@g(r, m, $s, 7, #v[ 3], #v[ 4], #v[ 9], #v[14]);
+}
+
+macro @common_compress(#instance, $rounds, $iv, $sigma, block) @local
+{
+	$typeof(#instance.h[0])[16] m, v;
+
+	((char*)&m)[:$sizeof(block)] = block[..];
+
+	v[:8] = #instance.h[..];
+	v[ 8] = $iv[0];
+	v[ 9] = $iv[1];
+	v[10] = $iv[2];
+	v[11] = $iv[3];
+	v[12] = $iv[4] ^ #instance.t[0];
+	v[13] = $iv[5] ^ #instance.t[1];
+	v[14] = $iv[6] ^ #instance.f[0];
+	v[15] = $iv[7] ^ #instance.f[1];
+
+	$for usz $i = 0; $i < $rounds; $i++:
+		@round($i, m, $sigma, v);
+	$endfor
+
+	$for usz $i = 0; $i < 8; $i++:
+		#instance.h[$i] ^= v[$i] ^ v[$i + 8];
+	$endfor
+}
+
+macro @add_ctr(#instance, usz amount) @local
+{
+	#instance.t[0] += ($typeof(#instance.t[0]))amount;
+	#instance.t[1] += ($typeof(#instance.t[0]))(#instance.t[0] < amount); // adds 1 on overflow of [0]
+}
+
+macro @common_init(#instance, $ParamType, $iv, usz out_len, char[] key = {}, char[] salt = {}, char[] personal = {}) @local
+{
+	mem::zero_volatile(@as_char_view(*#instance)); // explicitly because habits around hash init usually involve @noinit
+
+	#instance.h[..] = $iv[..];
+	#instance.outlen = out_len;
+
+	$ParamType p = {
+		.digest_length = (char)out_len,
+		.key_length = (char)key.len,
+		.fanout = 1,
+		.depth = 1,
+	};
+	if (salt.len) p.salt[:salt.len] = salt[..];
+	if (personal.len) p.personal[:personal.len] = personal[..];
+
+	array::@zip_into(((char*)&#instance.h)[:$sizeof(p)], ((char*)&p)[:$sizeof(p)], fn (a, b) => a ^ b); // bytes(self.h) ^= bytes(p)
+
+	if (key.len) 
+	{
+		char[$sizeof($iv[0])*16] dummy = {};
+		dummy[:key.len] = key[..];
+		#instance.update(dummy[..]); // consume a FULL block
+		mem::zero_volatile(dummy[..]); // do not optimize clearing this from the stack
+	}
+}
+
+macro @common_update(#instance, $block_size, char[] data) @local
+{
+	if (@unlikely(!data.len)) return;
+
+	usz fill = $block_size - #instance.buflen;
+	if (data.len > fill)
+	{
+		#instance.buf[#instance.buflen:fill] = data[:fill];
+		#instance.buflen = 0;
+
+		@add_ctr(#instance, $block_size);
+		#instance.compress(#instance.buf);
+
+		data = data[fill..];
+
+		for (; data.len > $block_size; data = data[$block_size..])
+		{
+			@add_ctr(#instance, $block_size);
+			#instance.compress(data[:$block_size]);
+		}
+	}
+	#instance.buf[#instance.buflen:data.len] = data[..];
+	#instance.buflen += data.len;
+}
+
+macro @common_final(#instance, $output_length) @local
+{
+	char[$output_length] result = {};
+	if ($output_length != #instance.outlen) return result;
+
+	@add_ctr(#instance, #instance.buflen);
+	if (#instance.f[0]) return result; // technically an error return
+
+	if (#instance.last_node) #instance.f[1] = $typeof(#instance.h[0]).max;
+	#instance.f[0] = $typeof(#instance.h[0]).max;
+
+	mem::zero_volatile(#instance.buf[#instance.buflen..]); // pad buffer with zeroes
+	#instance.compress(#instance.buf);
+
+    defer mem::zero_volatile(@as_char_view(*#instance)); // destroy the current context implicitly
+
+	result[:#instance.outlen] = @as_char_view(#instance.h)[:#instance.outlen];
+	return result;
+}
+
+
+// ======================================================================================
+// BEGIN Blake2b contents. Do not separate this from Blake2s: there's not really a point.
+//
+const BLAKE2B_BLOCKBYTES	@local = 128;
+const BLAKE2B_OUTBYTES		@local = 64;
+const BLAKE2B_KEYBYTES		@local = 64;
+const BLAKE2B_SALTBYTES		@local = 16;
+const BLAKE2B_PERSONALBYTES	@local = 16;
+const ulong[8] BLAKE2B_IV @local = {
+    0x6a09e667f3bcc908, 0xbb67ae8584caa73b, 0x3c6ef372fe94f82b, 0xa54ff53a5f1d36f1,
+    0x510e527fade682d1, 0x9b05688c2b3e6c1f, 0x1f83d9abfb41bd6b, 0x5be0cd19137e2179
+};
+const char[16][12] BLAKE2B_SIGMA @local = {
+    x'000102030405060708090a0b0c0d0e0f',
+    x'0e0a0408090f0d06010c00020b070503',
+    x'0b080c0005020f0d0a0e030607010904',
+    x'070903010d0c0b0e0206050a04000f08',
+    x'0900050702040a0f0e010b0c0608030d',
+    x'020c060a000b0803040d07050f0e0109',
+    x'0c05010f0e0d040a000706030902080b',
+    x'0d0b070e0c01030905000f040806020a',
+    x'060f0e090b0300080c020d0701040a05',
+    x'0a020804070601050f0b090e030c0d00',
+    x'000102030405060708090a0b0c0d0e0f',
+    x'0e0a0408090f0d06010c00020b070503',
+};
+
+struct Blake2b
+{
+	ulong[8]					h;
+	ulong[2]					t;
+	ulong[2]					f;
+	char[BLAKE2B_BLOCKBYTES]	buf;
+	usz							buflen;
+	usz							outlen;
+	char						last_node;
+}
+
+struct Blake2bParam @packed @local
+{
+	char						digest_length;	/*  1 */
+	char						key_length;		/*  2 */
+	char						fanout;			/*  3 */
+	char						depth;			/*  4 */
+	uint						leaf_length;	/*  8 */
+	uint						node_offset;	/* 12 */
+	uint						xof_length;		/* 16 */
+	char						node_depth;		/* 17 */
+	char						inner_length;	/* 18 */
+	char[14]					reserved;		/* 32 */
+	char[BLAKE2B_SALTBYTES]		salt;			/* 48 */
+	char[BLAKE2B_PERSONALBYTES]	personal;		/* 64 */
+}
+
+<*
+ @require $defined(data[0]) &&& $typeof(data[0]) == char : "Input data must be a char slice, char array, or string value."
+*>
+macro blake2b_hash($out_len, data, char[] key = {}, char[] salt = {})
+{
+	Blake2b b @noinit;
+    b.init($out_len, key, salt);
+	b.update(data[..]);
+	return b.final($out_len);
+}
+alias b = blake2b_hash;
+
+// See RFC 7693, Section 4 for common parameter sets.
+macro blake2b_224(data, char[] key = {}, char[] salt = {}) => blake2b_hash(SIZE_224, data, key, salt);
+macro blake2b_256(data, char[] key = {}, char[] salt = {}) => blake2b_hash(SIZE_256, data, key, salt);
+macro blake2b_384(data, char[] key = {}, char[] salt = {}) => blake2b_hash(SIZE_384, data, key, salt);
+macro blake2b_512(data, char[] key = {}, char[] salt = {}) => blake2b_hash(SIZE_512, data, key, salt);
+
+alias b_224 = blake2b_224;
+alias b_256 = blake2b_256;
+alias b_384 = blake2b_384;
+alias b_512 = blake2b_512;
+
+<*
+ Blake2b initialization method. Presents various options 
+
+ @param out_len : "The desired output length from the hash function."
+ @param[in] key : "An optional key value to use (keys the entire hash value to give HMAC-like functionality)."
+ @param[in] salt : "An optional salt value to use when generating the hash."
+ @param[in] personal : "An optional personalization value to use when generating the hash."
+
+ @require out_len > 0 && out_len <= BLAKE2B_OUTBYTES : "Blake2 output length must be within the proper range."
+ @require !key.ptr || (key.len > 0 && key.len <= BLAKE2B_KEYBYTES) : "A specified key's length must be within the proper range."
+ @require !salt.ptr || (salt.len > 0 && salt.len <= BLAKE2B_SALTBYTES) : "A specified salt's length must be within the proper range."
+ @require !personal.ptr || (personal.len > 0 && personal.len <= BLAKE2B_PERSONALBYTES) : "A specified personalization's length must be within the proper range."
+*>
+fn void Blake2b.init(&self, usz out_len, char[] key = {}, char[] salt = {}, char[] personal = {})
+	=> @common_init(self, Blake2bParam, BLAKE2B_IV, out_len, key, salt, personal);
+
+<*
+ Core compression inline function for Blake2b.
+*>
+fn void Blake2b.compress(&self, char[BLAKE2B_BLOCKBYTES] block) @local @inline
+	=> @common_compress(self, 12, BLAKE2B_IV, BLAKE2B_SIGMA, block);
+
+<*
+ Add more data to the hash context or stream.
+
+ @param[in] data : "The data to ingest into the hash context."
+*>
+fn void Blake2b.update(&self, char[] data)
+	=> @common_update(self, BLAKE2B_BLOCKBYTES, data);
+
+<*
+ Finalize the hash context and return the hash result at the given size.
+
+ @param $output_length : "The length of the output array which is returned by value, instead of as a slice."
+
+ @require $output_length == self.outlen : "The specified compile-time output size MUST be equal to the initialized output size."
+*>
+macro char[*] Blake2b.final(&self, $output_length)
+	=> @common_final(self, $output_length);
+
+
+// ======================================================================================
+// BEGIN Blake2s contents. Do not separate this from Blake2b: there's not really a point.
+//
+const BLAKE2S_BLOCKBYTES	@local = BLAKE2B_BLOCKBYTES 	/ 2;
+const BLAKE2S_OUTBYTES		@local = BLAKE2B_OUTBYTES 		/ 2;
+const BLAKE2S_KEYBYTES		@local = BLAKE2B_KEYBYTES 		/ 2;
+const BLAKE2S_SALTBYTES		@local = BLAKE2B_SALTBYTES 		/ 2;
+const BLAKE2S_PERSONALBYTES	@local = BLAKE2B_PERSONALBYTES 	/ 2;
+const uint[8] BLAKE2S_IV @local = { 0x6A09E667, 0xBB67AE85, 0x3C6EF372, 0xA54FF53A, 0x510E527F, 0x9B05688C, 0x1F83D9AB, 0x5BE0CD19 };
+const char[16][10] BLAKE2S_SIGMA @local = {
+    x'000102030405060708090a0b0c0d0e0f',
+    x'0e0a0408090f0d06010c00020b070503',
+    x'0b080c0005020f0d0a0e030607010904',
+    x'070903010d0c0b0e0206050a04000f08',
+    x'0900050702040a0f0e010b0c0608030d',
+    x'020c060a000b0803040d07050f0e0109',
+    x'0c05010f0e0d040a000706030902080b',
+    x'0d0b070e0c01030905000f040806020a',
+    x'060f0e090b0300080c020d0701040a05',
+    x'0a020804070601050f0b090e030c0d00',
+};
+
+struct Blake2s
+{
+	uint[8]						h;
+	uint[2]						t;
+	uint[2]						f;
+	char[BLAKE2S_BLOCKBYTES]	buf;
+	usz							buflen;
+	usz							outlen;
+	char						last_node;
+}
+
+struct Blake2sParam @packed @local
+{
+	char						digest_length;	/*  1 */
+	char						key_length;		/*  2 */
+	char						fanout;			/*  3 */
+	char						depth;			/*  4 */
+	uint						leaf_length;	/*  8 */
+	uint						node_offset;	/* 12 */
+	ushort						xof_length;		/* 14 */
+	char						node_depth;		/* 15 */
+	char						inner_length;	/* 16 */
+	char[BLAKE2S_SALTBYTES]		salt;			/* 24 */
+	char[BLAKE2S_PERSONALBYTES]	personal;		/* 32 */
+}
+
+<*
+ @require $defined(data[0]) &&& $typeof(data[0]) == char : "Input data must be a char slice, char array, or string value."
+*>
+macro blake2s_hash($out_len, data, char[] key = {}, char[] salt = {})
+{
+	Blake2s b @noinit;
+    b.init($out_len, key, salt);
+	b.update(data[..]);
+	return b.final($out_len);
+}
+alias s = blake2s_hash;
+
+// See RFC 7693, Section 4 for common parameter sets.
+macro blake2s_128(data, char[] key = {}, char[] salt = {}) => blake2s_hash(SIZE_128, data, key, salt);
+macro blake2s_160(data, char[] key = {}, char[] salt = {}) => blake2s_hash(SIZE_160, data, key, salt);
+macro blake2s_224(data, char[] key = {}, char[] salt = {}) => blake2s_hash(SIZE_224, data, key, salt);
+macro blake2s_256(data, char[] key = {}, char[] salt = {}) => blake2s_hash(SIZE_256, data, key, salt);
+
+alias s_128 = blake2s_128;
+alias s_160 = blake2s_160;
+alias s_224 = blake2s_224;
+alias s_256 = blake2s_256;
+
+<*
+ Blake2s initialization method. Presents various options 
+
+ @param out_len : "The desired output length from the hash function."
+ @param[in] key : "An optional key value to use (keys the entire hash value to give HMAC-like functionality)."
+ @param[in] salt : "An optional salt value to use when generating the hash."
+ @param[in] personal : "An optional personalization value to use when generating the hash."
+
+ @require out_len > 0 && out_len <= BLAKE2S_OUTBYTES : "Blake2 output length must be within the proper range."
+ @require !key.ptr || (key.len > 0 && key.len <= BLAKE2S_KEYBYTES) : "A specified key's length must be within the proper range."
+ @require !salt.ptr || (salt.len > 0 && salt.len <= BLAKE2S_SALTBYTES) : "A specified salt's length must be within the proper range."
+ @require !personal.ptr || (personal.len > 0 && personal.len <= BLAKE2B_PERSONALBYTES) : "A specified personalization's length must be within the proper range."
+*>
+fn void Blake2s.init(&self, usz out_len, char[] key = {}, char[] salt = {}, char[] personal = {})
+	=> @common_init(self, Blake2sParam, BLAKE2S_IV, out_len, key, salt, personal);
+
+<*
+ Core compression inline function for Blake2s.
+*>
+fn void Blake2s.compress(&self, char[BLAKE2S_BLOCKBYTES] block) @local @inline
+	=> @common_compress(self, 10, BLAKE2S_IV, BLAKE2S_SIGMA, block);
+
+<*
+ Add more data to the hash context or stream.
+
+ @param[in] data : "The data to ingest into the hash context."
+*>
+fn void Blake2s.update(&self, char[] data)
+	=> @common_update(self, BLAKE2S_BLOCKBYTES, data);
+
+<*
+ Finalize the hash context and return the hash result at the given size.
+
+ @param $output_length : "The length of the output array which is returned by value, instead of as a slice."
+
+ @require $output_length == self.outlen : "The specified compile-time output size MUST be equal to the initialized output size."
+*>
+macro char[*] Blake2s.final(&self, $output_length)
+	=> @common_final(self, $output_length);

lib/std/hash/blake3.c3

@@ -0,0 +1,747 @@
+// Copyright (c) 2025-2026 Zack Puhl <github@xmit.xyz>. 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.
+//
+// This is based on the original BLAKE3 reference implementation:
+//   https://github.com/BLAKE3-team/BLAKE3/blob/master
+//
+module std::hash::blake3;
+
+
+const BLOCK_SIZE = 64;
+const CHUNK_SIZE = 1024;
+const KEY_SIZE = 32;
+const KEY_SIZE_WORDS = KEY_SIZE / uint.sizeof;
+const OUT_SIZE = 32;
+const MAX_DEPTH = 54;
+
+const uint[8] IV = {
+	0x6a09e667, 0xbb67ae85, 0x3c6ef372, 0xa54ff53a,
+	0x510e527f, 0x9b05688c, 0x1f83d9ab, 0x5be0cd19
+};
+
+const char[16][7] MESSAGE_SCHEDULE = {
+	x'000102030405060708090a0b0c0d0e0f',
+	x'0206030a0700040d010b0c05090e0f08',
+	x'03040a0c0d02070e060509000b0f0801',
+	x'0a070c090e030d0f04000b0205080106',
+	x'0c0d090b0f0a0e080702050300010604',
+	x'090e0b05080c0f010d03000a02060407',
+	x'0b0f0500010908060e0a020c0304070d',
+};
+
+
+// Get feature-based optimization options.
+//   For now, none of these are used until there's a chance to explore BLAKE3's (necessary) vectorization optimizations.
+//
+<* When true, force the use of slow-but-portable BLAKE3 functions. Do not vectorize the hash function. *>
+const FORCE_PORTABLE = true; //$feature(BLAKE3_FORCE_PORTABLE); // this is statically set to TRUE for now
+<* AARCH64: When not big-endian, use Neon. *>
+const USE_NEON = !FORCE_PORTABLE &&& (env::AARCH64 &&& !env::BIG_ENDIAN);
+<* Bundling some architecture booleans into one. *>
+const IS_X86 = !FORCE_PORTABLE &&& (env::X86_64 ||| env::X86);
+<*
+ The maximum possible degree of parallelization based on the current architecture.
+ This doesn't represent the ACTUAL degree available.
+*>
+const MAX_SIMD_DEGREE = IS_X86 ??? 16 : (USE_NEON ??? 4 : 1);
+<* There are cases in BLAKE3 where, at compile-time, it's necessary to easily get the max degree, or a minimum of 2. *>
+const MAX_SIMD_DEGREE_OR_2 = @max(MAX_SIMD_DEGREE, 2);
+
+
+<* Always set to true once BLAKE3 caches some initial CPU details. *>
+bool cpuinfo_initd @local = false;
+
+<*
+ Cache some information at runtime about the current processor and platform, as needed for optimizations.
+*>
+fn void init_blake3() @local @init
+{
+	$if IS_X86:
+		cpudetect::x86_initialize_cpu_features(); // query all x86 feature flags, one time
+	$endif
+	cpuinfo_initd = true;
+}
+
+<* Check whether a given CPU flag is set (x86/x86_64 only). *>
+macro bool @check_cpu_flag(X86Feature f) @local @if(IS_X86)
+	=> !!(cpudetect::x86_features & f.ordinal);
+
+<*
+ Return the actual SIMD degree of the processor at runtime.
+*>
+macro @simd_degree() @local
+{
+	if (!cpuinfo_initd) init_blake3();
+	assert(cpuinfo_initd == true, "Failed to run required BLAKE3 initializations.");
+
+	$switch:
+		$case IS_X86:
+			if (@check_cpu_flag(AVX512F) && @check_cpu_flag(AVX512VL)) return 16;
+			if (@check_cpu_flag(AVX2)) return 8;
+			if (@check_cpu_flag(SSE4_1) || @check_cpu_flag(SSE2)) return 4;
+		$case USE_NEON:
+			return 4;
+	$endswitch
+
+	return 1;
+}
+
+<* Flags used during hash computation based on its state. *>
+enum Blake3Flags : const inline char
+{
+	CHUNK_START			= 1 << 0,
+	CHUNK_END			= 1 << 1,
+	PARENT				= 1 << 2,
+	ROOT				= 1 << 3,
+	KEYED_HASH			= 1 << 4,
+	DERIVE_KEY_CONTEXT	= 1 << 5,
+	DERIVE_KEY_MATERIAL	= 1 << 6,
+}
+
+struct Blake3ChunkState @local
+{
+	uint[8]				cv;
+	ulong				chunk_counter;
+	char[BLOCK_SIZE]	buf;
+	char				buf_len;
+	char				blocks_compressed;
+	char				flags;
+}
+
+struct Blake3Output @local
+{
+	uint[KEY_SIZE_WORDS]	input_cv;
+	ulong					counter;
+	char[BLOCK_SIZE]		block;
+	char					block_len;
+	char					flags;
+}
+
+struct Blake3
+{
+	uint[KEY_SIZE_WORDS]				key;
+	Blake3ChunkState					chunk;
+	char								cv_stack_len;
+	char[(MAX_DEPTH + 1) * OUT_SIZE]	cv_stack;
+}
+
+
+<*
+ Generate an XOF hash based on the given inputs.
+
+ Consider the output hash w/ `seek = 0` and `$out_size = 41`:
+ ```
+ 2cc39783c223154fea8dfb7c1b1660f2ac2dcbd1c1de8277b0b0dd39b7e50d7d905630c8be290dfcf3
+ ```
+
+ Computing with the same input `key` and input `data`, but with a `seek = 3` and `$out_size = 8` yields:
+ ```
+ 83c223154fea8dfb
+
+ which is a slice cut out from the above hash:
+   2cc397  [83c223154fea8dfb]  7c1b1660f2ac2dcbd1c1de8277b0b0dd39b7e50d7d905630c8be290dfcf3
+ ```
+
+ In this way, the XOF primitive that BLAKE3 is built from allows the hash output to be a potentially
+ limitless result that one may slice to their liking using the right parameters.
+
+ @param [in] data : "The data to hash."
+ @param [in] key : "An optional 32-byte key to turn the result into a keyed hash."
+ @param seek : "An optional value specifying the offset into the XOF's yield where the resultant hash should begin."
+ @param $out_size : "An optional value specifying the desired length to slice from the XOF's yield."
+
+ @return "The hash as a character array of `$out_size` bytes."
+
+ @require !key.len || key.len == KEY_SIZE : "Key value must be empty or exactly 32 bytes."
+ @require $out_size > 0 : "You cannot use a zero $out_size."
+*>
+macro char[*] hash(char[] data, char[] key = {}, usz seek = 0, usz $out_size = 32)
+{
+	char[$out_size] result;
+	Blake3 b @noinit;
+	defer b.destroy();
+	b.init(key);
+	b.update(data);
+	b.final(result[..], $out_size, seek);
+	return result;
+}
+
+<*
+ Generate a hash from a context string. This call allows one to use the "context" to
+ auto-generate keying material for the resultant hash value. Effectively, this allows for
+ hashes made from data with completely variable-length keys, rather than having a key fixed
+ to 32 bytes. The 'context' nomenclature is from BLAKE3 itself, not my naming.
+
+ @param [in] data : "The data to hash."
+ @param [in] context : "An optional key to turn the result into a keyed hash."
+ @param seek : "An optional value specifying the offset into the XOF's yield where the resultant hash should begin."
+ @param $out_size : "An optional value specifying the desired length to slice from the XOF's yield."
+
+ @return "The context-based hash as a character array of `$out_size` bytes."
+
+ @require $out_size > 0 : "You cannot use a zero $out_size."
+*>
+macro char[*] ctx_hash(char[] data, char[] context, usz seek = 0, usz $out_size = 32)
+{
+	char[$out_size] result;
+	Blake3 b = new_from_context(context);
+	defer b.destroy();
+	b.update(data);
+	b.final(result[..], $out_size, seek);
+	return result;
+}
+
+<*
+ Generate a new Blake3 hashing structure from the given context string. The context string
+ acts as a variable-length key to seed the new hash structure, and makes it ready to ingest
+ incoming data with `update`.
+
+ @param [in] context : "The context byte array used to seed the returned Blake3 context."
+*>
+macro Blake3 new_from_context(char[] context)
+{
+	char[KEY_SIZE] context_based_key;
+	defer mem::zero_volatile(context_based_key[..]);
+
+	Blake3 key_from_ctx @noinit;
+	defer key_from_ctx.destroy();
+	key_from_ctx.init(explicit_flags: Blake3Flags.DERIVE_KEY_CONTEXT);
+	key_from_ctx.update(context);
+	key_from_ctx.final(context_based_key[..], KEY_SIZE);
+
+	Blake3 b @noinit;
+	b.init(key: context_based_key[..], explicit_flags: Blake3Flags.DERIVE_KEY_MATERIAL);
+	return b;
+}
+
+
+<*
+ Initialize a BLAKE3 context.
+
+ @param [in] key : "An optional key initializer to use."
+
+ @require !key.len || key.len == KEY_SIZE : "An explicit initialization key must be of KEY_SIZE (32 bytes)."
+*>
+fn void Blake3.init(&self, char[] key = {}, char explicit_flags = 0)
+{
+	mem::zero_volatile(@as_char_view(*self));
+
+	if (key.len)
+	{
+		foreach (i, &w : self.key) *w = mem::load((uint*)&key[i * $sizeof(self.key[0])], 1);
+		if (!explicit_flags) explicit_flags = Blake3Flags.KEYED_HASH;
+	}
+	else
+	{
+		self.key[..] = IV[..];
+	}
+
+	self.chunk.init(self.key[..], explicit_flags);
+}
+
+<*
+ Reset the state of the hashing context, in case it should be reused without reloading the key value.
+*>
+fn void Blake3.reset(&self) @local @inline
+{
+	self.chunk.reset(self.key[..], 0);
+	self.cv_stack_len = 0;
+}
+
+<*
+ Private function to merge tree results.
+*>
+fn void Blake3.merge_cv_stack(&self, ulong total_len) @local @inline
+{
+	usz post_merge_stack_len = (usz)@popcnt(total_len);
+	for (; self.cv_stack_len > post_merge_stack_len; self.cv_stack_len--)
+	{
+		char* parent_node = &self.cv_stack[(self.cv_stack_len - 2) * OUT_SIZE];
+		Blake3Output o = parent_output(parent_node, self.key[..], self.chunk.flags);
+		o.chaining_value(parent_node);
+	}
+}
+
+<*
+ Private function to add a new tree onto the stack.
+*>
+fn void Blake3.push_cv(&self, char* new_cv, ulong chunk_counter) @local @inline
+{
+	self.merge_cv_stack(chunk_counter);
+	self.cv_stack[self.cv_stack_len * OUT_SIZE : OUT_SIZE] = new_cv[:OUT_SIZE];
+	self.cv_stack_len++;
+}
+
+<*
+ Update the hash context by consuming incoming data.
+
+ @param [in] input : "The slice of new data to digest."
+ @param use_tbb : "Should remain `false` until other BLAKE3 optimizations are set up."
+*>
+fn void Blake3.update(&self, char[] input, bool use_tbb = false)
+{
+	if (!input.len) return;
+
+	if (self.chunk.len() > 0)
+	{
+		usz take = min(CHUNK_SIZE - self.chunk.len(), input.len);
+		self.chunk.update(input[:take]);
+		input = input[take..];
+
+		if (!input.len) return;
+
+		char[KEY_SIZE] chunk_cv;
+		Blake3Output o = self.chunk.output();
+		o.chaining_value(&chunk_cv);
+		self.push_cv(&chunk_cv, self.chunk.chunk_counter);
+		self.chunk.reset(self.key[..], self.chunk.chunk_counter + 1);
+	}
+
+	while (input.len > CHUNK_SIZE)
+	{
+		usz subtree_len = @round_down_to_power_of_2(input.len);
+		ulong count_so_far = self.chunk.chunk_counter * CHUNK_SIZE;
+
+		while ((((ulong)(subtree_len - 1)) & count_so_far) != 0) subtree_len /= 2;
+
+		ulong subtree_chunks = subtree_len / CHUNK_SIZE;
+		if (subtree_len <= CHUNK_SIZE)
+		{
+			Blake3ChunkState chunk_state;
+			chunk_state.init(self.key[..], self.chunk.flags);
+			chunk_state.chunk_counter = self.chunk.chunk_counter;
+			chunk_state.update(input[:subtree_len]);
+			char[OUT_SIZE] cv;
+			Blake3Output o = chunk_state.output();
+			o.chaining_value(&cv);
+			self.push_cv(&cv, chunk_state.chunk_counter);
+		}
+		else
+		{
+			char[2 * OUT_SIZE] cv_pair;
+			compress_subtree_to_parent_node(input[:subtree_len], self.key[..], self.chunk.chunk_counter, self.chunk.flags, cv_pair[..], use_tbb);
+			self.push_cv(&cv_pair[0], self.chunk.chunk_counter);
+			self.push_cv(&cv_pair[OUT_SIZE], self.chunk.chunk_counter + (subtree_chunks / 2));
+		}
+		self.chunk.chunk_counter += subtree_chunks;
+		input = input[subtree_len..];
+	}
+
+	if (input.len > 0)
+	{
+		self.chunk.update(input);
+		self.merge_cv_stack(self.chunk.chunk_counter);
+	}
+}
+
+<*
+ Yield the results of the hash into a specified output buffer, at the specified length.
+ Note that the `into` slice does not need to be properly cut to receive hash results; it
+ just needs to be wide enough to accommodate `into_len` yielded bytes from the XOF.
+
+ @param [in] into : "The storage buffer for the output hash value. Must be >= `into_len` bytes."
+ @param into_len : "How many bytes to receive from the XOF/hash output."
+ @param seek : "How far into the XOF's yield to begin the stored byte sequence."
+
+ @require into.len >= into_len : "The requested output size must be equal to or less than the size of the output slice."
+*>
+fn void Blake3.final(&self, char[] into, usz into_len, usz seek = 0)
+{
+	if (!into_len) return;
+
+	if (!self.cv_stack_len)
+	{
+		Blake3Output o = self.chunk.output();
+		o.root_bytes(seek, into[:into_len]);
+		return;
+	}
+
+	Blake3Output o @noinit;
+	usz cvs_remaining;
+	if (self.chunk.len() > 0)
+	{
+		cvs_remaining = self.cv_stack_len;
+		o = self.chunk.output();
+	}
+	else
+	{
+		cvs_remaining = (usz)self.cv_stack_len - 2;
+		o = parent_output(&self.cv_stack[cvs_remaining * KEY_SIZE], self.key[..], self.chunk.flags);
+	}
+
+	while (cvs_remaining > 0)
+	{
+		char[BLOCK_SIZE] parent_block;
+		cvs_remaining--;
+		parent_block[:32] = self.cv_stack[cvs_remaining * 32 : 32];
+		o.chaining_value(&parent_block[32]);
+		o = parent_output(&parent_block, self.key[..], self.chunk.flags);
+	}
+
+	o.root_bytes(seek, into[:into_len]);
+}
+
+<*
+ Destroy a BLAKE3 hashing context.
+*>
+fn void Blake3.destroy(&self) @inline
+{
+	mem::zero_volatile(@as_char_view(*self));
+}
+
+
+<*
+ Initialize a BLAKE3 chunk state.
+
+ @param [in] key
+ @param flags
+*>
+fn void Blake3ChunkState.init(&self, uint[] key, char flags) @local @inline
+{
+	mem::zero_volatile(@as_char_view(*self));
+	self.cv[..] = key[..];
+	self.flags = flags;
+}
+
+<*
+ Reset a BLAKE3 chunk state.
+
+ @param [in] key
+ @param chunk_counter
+*>
+fn void Blake3ChunkState.reset(&self, uint[] key, ulong chunk_counter) @local @inline
+{
+	self.init(key, self.flags); // maintain its own flags
+	self.chunk_counter = chunk_counter; // update chunk counter
+}
+
+<*
+ Get bytes length of consumed data.
+*>
+fn usz Blake3ChunkState.len(&self) @operator(len) @local @inline
+	=> (BLOCK_SIZE * (usz)self.blocks_compressed) + (usz)self.buf_len;
+
+<*
+ Ingest an amount of bytes into the chunk's buffer. NOTE: Doesn't check for underflow.
+
+ @param [in] data : "Data to ingest."
+*>
+fn usz Blake3ChunkState.fill_buf(&self, char[] data) @local @inline
+{
+	usz take = min(BLOCK_SIZE - (usz)self.buf_len, data.len);
+	self.buf[self.buf_len:take] = data[:take];
+	self.buf_len += (char)take;
+	return take;
+}
+
+<*
+ Determine whether to set the CHUNK_START flag.
+*>
+fn char Blake3ChunkState.maybe_start_flag(&self) @local @inline
+	=> !self.blocks_compressed ? Blake3Flags.CHUNK_START : 0;
+
+<*
+ Update the chunk with the provided input bytes.
+
+ @param [in] input : "Incoming bytes to update with."
+*>
+fn void Blake3ChunkState.update(&self, char[] input) @local
+{
+	if (self.buf_len)
+	{
+		usz take = self.fill_buf(input);
+		input = input[take..];
+		if (input.len)
+		{
+			compress_in_place(self.cv[..], self.buf[..], BLOCK_SIZE, self.chunk_counter, self.flags | self.maybe_start_flag());
+			self.blocks_compressed++;
+			self.buf_len = 0;
+			self.buf[..] = {};
+		}
+	}
+	for (; input.len > BLOCK_SIZE; self.blocks_compressed++, input = input[BLOCK_SIZE..])
+	{
+		compress_in_place(self.cv[..], input[:BLOCK_SIZE], BLOCK_SIZE, self.chunk_counter, self.flags | self.maybe_start_flag());
+	}
+	self.fill_buf(input);
+}
+
+<*
+ Convert the chunk state to an "output" type with the right flags.
+*>
+fn Blake3Output Blake3ChunkState.output(&self) @local @inline
+	=> make_output(self.cv[..], &self.buf, self.buf_len, self.chunk_counter, self.flags | self.maybe_start_flag() | Blake3Flags.CHUNK_END);
+
+<*
+ Generate and initialize an output structure with the provided parameters.
+
+ @param [in] key
+ @param [&in] in_block
+ @param block_len
+ @param counter
+ @param flags
+*>
+fn Blake3Output make_output(uint[] key, char* in_block, usz block_len, ulong counter, char flags) @local @noinline
+{
+	Blake3Output o;
+	o.input_cv[..] = key[..];
+	o.block[..] = in_block[:BLOCK_SIZE];
+	o.block_len = (char)block_len;
+	o.counter = counter;
+	o.flags = flags;
+	return o;
+}
+
+<*
+ Auto-generate a parent output structure, pre-initialized with some constant identifiers.
+
+ @param [&in] block
+ @param [in] key
+ @param flags
+*>
+macro Blake3Output parent_output(char* block, uint[] key, char flags) @local
+	=> make_output(key, block, BLOCK_SIZE, 0, flags | Blake3Flags.PARENT);
+
+<*
+ Compress then store the chaining value of the output structure.
+
+ @param [&inout] cv
+*>
+macro void Blake3Output.chaining_value(&self, char* cv) @local
+{
+	uint[KEY_SIZE_WORDS] cv_words;
+	cv_words[..] = self.input_cv[..];
+	compress_in_place(cv_words[..], self.block, self.block_len, self.counter, self.flags);
+	cv[:KEY_SIZE] = @as_char_view(cv_words)[:KEY_SIZE];
+}
+
+<*
+ Store the result of the output into the designated slice.
+
+ @param seek
+ @param [inout] into
+*>
+fn void Blake3Output.root_bytes(&self, usz seek, char[] into) @local
+{
+	if (!into.len) return;
+
+	ulong output_block_counter = seek / BLOCK_SIZE;
+	usz offset_within_block = seek % BLOCK_SIZE;
+	char[BLOCK_SIZE] wide_buf;
+
+	if (offset_within_block)
+	{
+		compress_xof(self.input_cv[..], self.block, self.block_len, output_block_counter, self.flags | Blake3Flags.ROOT, wide_buf[..]);
+		usz avail = BLOCK_SIZE - offset_within_block;
+		usz bytes = min(into.len, avail);
+		into[:bytes] = wide_buf[offset_within_block:bytes];
+		into = into[bytes..];
+		output_block_counter++;
+	}
+	if (into.len / BLOCK_SIZE)
+	{
+		@xof_many(self.input_cv[..], self.block, self.block_len, output_block_counter, self.flags | Blake3Flags.ROOT, into, into.len / BLOCK_SIZE);
+	}
+	output_block_counter += into.len / 64;
+	into = into[(usz)(into.len & -64ll) ..];
+	if (into.len)
+	{
+		compress_xof(self.input_cv[..], self.block, self.block_len, output_block_counter, self.flags | Blake3Flags.ROOT, wide_buf[..]);
+		into[..] = wide_buf[:into.len];
+	}
+}
+
+
+// =================================================================================================
+// =================================================================================================
+// =================================================================================================
+// WELCOME TO THE COMPUTATION GARDEN...
+//
+//     You wanna understand BLAKE3? You gotta get through us.
+//          ______________________________
+//        ༼ ºل͟º ༼ ºل͟º ༼ ºل͟º ༽ ºل͟º ༽ ºل͟º ༽
+//
+//
+macro uint @popcnt(#x) @local => (uint)#x.popcount();
+macro uint @highest_one(#x) @local => 63 ^ (uint)#x.clz();
+macro usz @round_down_to_power_of_2(#x) @local => (usz)1 << @highest_one(#x | 1);
+
+macro left_subtree_len(usz input_len) @local
+	=> @round_down_to_power_of_2((input_len - 1) / CHUNK_SIZE) * CHUNK_SIZE;
+
+
+macro @g(#state, a, b, c, d, x, y) @local
+{
+	#state[a] += #state[b] + x;
+	#state[d] = (#state[d] ^ #state[a]).rotr(16);
+	#state[c] += #state[d];
+	#state[b] = (#state[b] ^ #state[c]).rotr(12);
+	#state[a] += #state[b] + y;
+	#state[d] = (#state[d] ^ #state[a]).rotr(8);
+	#state[c] += #state[d];
+	#state[b] = (#state[b] ^ #state[c]).rotr(7);
+}
+
+macro @round(uint[] state, uint* msg, usz round) @local
+{
+	char* schedule = &MESSAGE_SCHEDULE[round];
+	@g(state, 0, 4,  8, 12, msg[schedule[0] ], msg[schedule[1] ]);
+	@g(state, 1, 5,  9, 13, msg[schedule[2] ], msg[schedule[3] ]);
+	@g(state, 2, 6, 10, 14, msg[schedule[4] ], msg[schedule[5] ]);
+	@g(state, 3, 7, 11, 15, msg[schedule[6] ], msg[schedule[7] ]);
+	@g(state, 0, 5, 10, 15, msg[schedule[8] ], msg[schedule[9] ]);
+	@g(state, 1, 6, 11, 12, msg[schedule[10]], msg[schedule[11]]);
+	@g(state, 2, 7,  8, 13, msg[schedule[12]], msg[schedule[13]]);
+	@g(state, 3, 4,  9, 14, msg[schedule[14]], msg[schedule[15]]);
+}
+
+fn void compress_pre(uint[] state, uint[] cv, char[BLOCK_SIZE] block, usz block_len, ulong counter, char flags) @local @noinline
+{
+	uint[16] block_words @noinit;
+	foreach (i, &b : block_words) *b = mem::load((uint*)&block[i * 4], 1);
+	state[0:8] = cv[0:8];
+	state[8:4] = IV[0:4];
+	state[12] = (uint)counter;
+	state[13] = (uint)(counter >> 32);
+	state[14] = (uint)block_len;
+	state[15] = (uint)flags;
+	@round(state, &block_words[0], 0);
+	@round(state, &block_words[0], 1);
+	@round(state, &block_words[0], 2);
+	@round(state, &block_words[0], 3);
+	@round(state, &block_words[0], 4);
+	@round(state, &block_words[0], 5);
+	@round(state, &block_words[0], 6);
+}
+
+macro compress_in_place(uint[] cv, char[BLOCK_SIZE] block, usz block_len, ulong counter, char flags) @local
+{
+	uint[16] state @noinit;
+	compress_pre(state[..], cv, block, block_len, counter, flags);
+	for (usz i = 0; i < 8; i++) cv[i] = state[i] ^ state[i + 8];
+}
+
+macro compress_xof(uint[] cv, char[BLOCK_SIZE] block, usz block_len, ulong counter, char flags, char[] out) @local
+{
+	uint[16] state @noinit;
+	compress_pre(state[..], cv, block, block_len, counter, flags);
+	$for usz $i = 0; $i < 8; $i++: mem::store((uint*)&out[4 * $i], state[$i] ^ state[$i + 8], 1); $endfor
+	$for usz $i = 0; $i < 8; $i++: mem::store((uint*)&out[4 * (8 + $i)], state[$i + 8] ^ cv[$i], 1); $endfor
+}
+
+macro @xof_many(uint[] cv, char[BLOCK_SIZE] block, usz block_len, ulong counter, char flags, char[] out, usz out_blocks) @local
+{
+	for (usz i = 0; i < out_blocks; i++, out = out[BLOCK_SIZE..]) compress_xof(cv, block, block_len, counter + i, flags, out);
+}
+
+macro hash_one(char* input, usz blocks, uint[] key, ulong counter, char flags, char flags_start, char flags_end, char[] out) @local
+{
+	uint[8] cv;
+	cv[..] = key[..];
+	char block_flags = flags | flags_start;
+	for (; blocks > 0; input += BLOCK_SIZE, blocks--, block_flags = flags)
+	{
+		if (blocks == 1) block_flags |= flags_end;
+		compress_in_place(cv[..], input[:BLOCK_SIZE], BLOCK_SIZE, counter, block_flags);
+	}
+	foreach (i, c : cv) mem::store((uint*)&out[i * 4], c, 1);
+}
+
+macro hash_many(char*[] inputs, usz num_inputs, usz blocks, uint[] key, ulong counter, bool $increment_counter, char flags, char flags_start, char flags_end, char* out) @local
+{
+	for (; num_inputs > 0; num_inputs--, inputs = inputs[1..], out += OUT_SIZE)
+	{
+		hash_one(inputs[0], blocks, key, counter, flags, flags_start, flags_end, out[:OUT_SIZE]);
+		$if $increment_counter: counter++; $endif
+	}
+}
+
+
+fn void compress_subtree_to_parent_node(char[] input, uint[] key, ulong chunk_counter, char flags, char[] out, bool use_tbb) @local @noinline
+{
+	char[MAX_SIMD_DEGREE_OR_2 * OUT_SIZE] cv_array;
+
+	usz num_cvs = compress_subtree_wide(input, key, chunk_counter, flags, cv_array[..], use_tbb);
+	assert(num_cvs <= 2);
+
+	$if MAX_SIMD_DEGREE_OR_2 > 2:
+		char[MAX_SIMD_DEGREE_OR_2 * OUT_SIZE / 2] out_array;
+		while (num_cvs > 2) num_cvs = compress_parents_parallel(cv_array[..], num_cvs, key, flags, &out_array);
+	$endif
+
+	out[..] = cv_array[:2 * OUT_SIZE];
+}
+
+fn usz compress_subtree_wide(char[] input, uint[] key, ulong chunk_counter, char flags, char* out, bool use_tbb) @local @noinline
+{
+	if (input.len <= @simd_degree() * CHUNK_SIZE) return compress_chunks_parallel(input, key, chunk_counter, flags, out);
+
+	usz left_input_len = left_subtree_len(input.len);
+	usz right_input_len = input.len - left_input_len;
+	char* right_input = &input[left_input_len];
+	ulong right_chunk_counter = chunk_counter + (ulong)(left_input_len / CHUNK_SIZE);
+
+	char[2 * MAX_SIMD_DEGREE_OR_2 * OUT_SIZE] cv_array;
+	usz degree = @simd_degree();
+	if (left_input_len > CHUNK_SIZE && degree == 1) degree = 2;
+	char* right_cvs = &cv_array[degree * OUT_SIZE];
+
+	usz left_n = compress_subtree_wide(input[:left_input_len], key, chunk_counter, flags, &cv_array, use_tbb);
+	usz right_n = compress_subtree_wide(right_input[:right_input_len], key, right_chunk_counter, flags, right_cvs, use_tbb);
+
+	if (left_n == 1)
+	{
+		out[:2 * OUT_SIZE] = cv_array[:2 * OUT_SIZE];
+		return 2;
+	}
+
+	return compress_parents_parallel(cv_array[..], left_n + right_n, key, flags, out);
+}
+
+fn usz compress_parents_parallel(char[] child_chaining_values, usz num_chaining_values, uint[] key, char flags, char* out) @local @noinline
+{
+	char*[MAX_SIMD_DEGREE_OR_2] parents_array;
+	usz parents_array_len = 0;
+
+	while (num_chaining_values - (2 * parents_array_len) >= 2)
+	{
+		parents_array[parents_array_len++] = &child_chaining_values[2 * parents_array_len * OUT_SIZE];
+	}
+
+	hash_many(parents_array[:parents_array_len], parents_array_len, 1, key, 0, false, flags | Blake3Flags.PARENT, 0, 0, out);
+
+	if (num_chaining_values > 2 * parents_array_len)
+	{
+		out[parents_array_len * OUT_SIZE : OUT_SIZE] = child_chaining_values[2 * parents_array_len * OUT_SIZE : OUT_SIZE];
+		return parents_array_len + 1;
+	}
+
+	return parents_array_len;
+}
+
+fn usz compress_chunks_parallel(char[] input, uint[] key, ulong chunk_counter, char flags, char* out) @local @noinline
+{
+	char*[MAX_SIMD_DEGREE] chunks_array;
+	usz input_position = 0;
+	usz chunks_array_len = 0;
+
+	for (; input.len - input_position >= CHUNK_SIZE; input_position += CHUNK_SIZE)
+	{
+		chunks_array[chunks_array_len++] = &input[input_position];
+	}
+
+	hash_many(chunks_array[:chunks_array_len], chunks_array_len, CHUNK_SIZE / BLOCK_SIZE, key, chunk_counter, true, flags, Blake3Flags.CHUNK_START, Blake3Flags.CHUNK_END, out);
+
+	if (input.len <= input_position) return chunks_array_len;
+
+	ulong counter = chunk_counter + (ulong)chunks_array_len;
+	Blake3ChunkState chunk_state;
+	chunk_state.init(key, flags);
+	chunk_state.chunk_counter = counter;
+	chunk_state.update(input[input_position : input.len - input_position]);
+	Blake3Output o = chunk_state.output();
+	o.chaining_value(&out[chunks_array_len * OUT_SIZE]);
+
+	return chunks_array_len + 1;
+}

lib/std/hash/gost/streebog.c3

@@ -0,0 +1,169 @@
+// Copyright (c) 2025 Zack Puhl <github@xmit.xyz>. 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.
+//
+// An implementation of Streebog-256 and Streebog-512, defined in the
+//   Russian standard GOST R 34.11-2012. Also known as "GOST-12".
+//
+module std::hash::streebog;
+
+
+enum StreebogLength : const inline uint
+{
+	SIZE_256 = 32,
+	SIZE_512 = 64,
+}
+
+struct Streebog
+{
+	ulong[8]    h;
+	ulong[8]    n;
+	ulong[8]    s;
+	ulong[8]    message;
+	usz 	    index;
+	usz			hash_size;
+}
+
+
+<*
+ @require $defined(data[0]) &&& $typeof(data[0]) == char : "Input data must be a char slice, char array, or string value."
+*>
+macro char[*] hash(StreebogLength $hash_size, data)
+{
+	// TODO: The speed of this hash hinges on use of >>> SIMD instructions <<<. It would be nice to have some added.
+	Streebog s @noinit;
+	s.init($hash_size);
+	s.update(data);
+	return s.final($hash_size);
+}
+
+macro char[*] hash_256(char[] data) => hash(SIZE_256, data);
+macro char[*] hash_512(char[] data) => hash(SIZE_512, data);
+
+
+macro @xor_512(#x, #y, #z) @local
+{
+	$for var $i = 0; $i < 8; $i++:
+		#z[$i] = #x[$i] ^ #y[$i];
+	$endfor
+}
+
+macro @add_512(#sum, #x) @local
+{
+	ulong carry = 0;
+	$for usz $i = 0; $i < 8; $i++:
+		#sum[$i] += #x[$i] + carry;
+		carry = #sum[$i] < #x[$i] ? 1 : (#sum[$i] == #x[$i] ? carry : 0);
+	$endfor
+}
+
+macro @lpsx(#a, #b, #result) @local
+{
+	ulong[8] z @noinit;
+	@xor_512(#a, #b, z);
+	// Do not unroll these loops - produces far too much code at compile-time.
+	for (usz i = 0; i < 8; i++)
+	{
+		#result[i] = TR[0][(z[0] >> (i << 3)) & 0xff];
+		for (usz j = 1; j < 8; j++) #result[i] ^= TR[j][(z[j] >> (i << 3)) & 0xff];
+	}
+}
+
+macro @g_n(#n, #h, #m) @local
+{
+	ulong[8] k_i;
+	ulong[8] state;
+
+	@lpsx(#h, #n, k_i);
+	@lpsx(k_i, #m, state);
+	// Do not unroll this loop - produces far too much code at compile-time.
+	for (usz i = 0; i < 11; i++)
+	{
+		@lpsx(k_i, ITERATION_CONSTANTS[i], k_i);
+		@lpsx(k_i, state, state);
+	}
+	@lpsx(k_i, ITERATION_CONSTANTS[11], k_i);
+	@xor_512(k_i, state, state);
+	@xor_512(state, #h, state);
+	@xor_512(state, #m, #h);
+}
+
+
+macro Streebog.@stage2(&self, #m) @local
+{
+	@g_n(self.n, self.h, #m);
+	@add_512(self.n, STAGE2_512);
+	@add_512(self.s, #m);
+}
+
+macro void Streebog.init(&self, StreebogLength $hash_size)
+{
+	mem::zero_volatile(@as_char_view(*self)); // explicitly initialize the entire state to 0
+	self.hash_size = $hash_size;
+	$if $hash_size != SIZE_512:
+		@as_char_view(self.h)[..] = (char[*]){ [0..63] = 0x01 }[..];
+	$endif
+}
+
+fn void Streebog.update(&self, char[] data)
+{
+	if (self.index)
+	{
+		usz rest = BLOCK_SIZE - self.index;
+		usz size = data.len;
+		usz len = size < rest ? size : rest;
+
+		@as_char_view(self.message)[self.index:len] = data[:len];
+		self.index += size;
+		if (size < rest) return;
+
+		self.@stage2(self.message);
+		data = data[rest..];
+		self.index = 0;
+	}
+	bool aligned = 0 == (usz)data.ptr % ulong.sizeof;
+	for (; data.len >= BLOCK_SIZE; data = data[BLOCK_SIZE..])
+	{
+		if (aligned)
+		{
+			self.@stage2((ulong*)data.ptr);
+		}
+		else
+		{
+			@as_char_view(self.message)[:BLOCK_SIZE] = data[:BLOCK_SIZE];
+			self.@stage2(self.message);
+		}
+	}
+	if (data.len)
+	{
+		self.index = data.len;
+		@as_char_view(self.message)[:data.len] = data[..];
+	}
+}
+
+<*
+ @require $hash_size == self.hash_size : "You must use the same output hash size as was initialized with the context."
+*>
+macro char[*] Streebog.final(&self, StreebogLength $hash_size)
+{
+	char[$hash_size] result;
+	ulong[8] unprocessed_bits_count;
+	usz index = self.index >> 3;
+	usz shift = (self.index & 0b111) * 8;
+
+	unprocessed_bits_count[0] = self.index * 8;
+	self.message[index]   &= ~(ulong.max << shift);
+	self.message[index++] ^= 1ul << shift;
+	if (index < 8) self.message[index..] = {};
+
+	@g_n(self.n, self.h, self.message);
+	@add_512(self.n, unprocessed_bits_count);
+	@add_512(self.s, self.message);
+	@g_n(ZERO_512, self.h, self.n);
+	@g_n(ZERO_512, self.h, self.s);
+
+	defer mem::zero_volatile(@as_char_view(*self)); // implicitly clear the structure when finalized
+
+	result[..] = @as_char_view(self.h)[(BLOCK_SIZE - $hash_size)..];
+	return result;
+}

lib/std/hash/gost/streebog_internal.c3

@@ -0,0 +1,305 @@
+// Copyright (c) 2025 Zack Puhl <github@xmit.xyz>. 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.
+//
+// An implementation of Streebog-256 and Streebog-512, defined in the
+//   Russian standard GOST R 34.11-2012. Also known as "GOST-12".
+//
+module std::hash::streebog @private;
+
+
+const usz BLOCK_SIZE = 64;
+
+const ulong[8] ZERO_512 @align(ulong.sizeof) = {};
+
+const ulong[8] STAGE2_512 @align(ulong.sizeof) = { [0] = 512ul, [1..7] = 0 };
+
+const ulong[8][12] ITERATION_CONSTANTS @align(ulong.sizeof) = {
+	{ 0xdd806559f2a64507, 0x05767436cc744d23, 0xa2422a08a460d315, 0x4b7ce09192676901, 0x714eb88d7585c4fc, 0x2f6a76432e45d016, 0xebcb2f81c0657c1f, 0xb1085bda1ecadae9 },
+	{ 0xe679047021b19bb7, 0x55dda21bd7cbcd56, 0x5cb561c2db0aa7ca, 0x9ab5176b12d69958, 0x61d55e0f16b50131, 0xf3feea720a232b98, 0x4fe39d460f70b5d7, 0x6fa3b58aa99d2f1a },
+	{ 0x991e96f50aba0ab2, 0xc2b6f443867adb31, 0xc1c93a376062db09, 0xd3e20fe490359eb1, 0xf2ea7514b1297b7b, 0x06f15e5f529c1f8b, 0x0a39fc286a3d8435, 0xf574dcac2bce2fc7 },
+	{ 0x220cbebc84e3d12e, 0x3453eaa193e837f1, 0xd8b71333935203be, 0xa9d72c82ed03d675, 0x9d721cad685e353f, 0x488e857e335c3c7d, 0xf948e1a05d71e4dd, 0xef1fdfb3e81566d2 },
+	{ 0x601758fd7c6cfe57, 0x7a56a27ea9ea63f5, 0xdfff00b723271a16, 0xbfcd1747253af5a3, 0x359e35d7800fffbd, 0x7f151c1f1686104a, 0x9a3f410c6ca92363, 0x4bea6bacad474799 },
+	{ 0xfa68407a46647d6e, 0xbf71c57236904f35, 0x0af21f66c2bec6b6, 0xcffaa6b71c9ab7b4, 0x187f9ab49af08ec6, 0x2d66c4f95142a46c, 0x6fa4c33b7a3039c0, 0xae4faeae1d3ad3d9 },
+	{ 0x8886564d3a14d493, 0x3517454ca23c4af3, 0x06476983284a0504, 0x0992abc52d822c37, 0xd3473e33197a93c9, 0x399ec6c7e6bf87c9, 0x51ac86febf240954, 0xf4c70e16eeaac5ec },
+	{ 0xa47f0dd4bf02e71e, 0x36acc2355951a8d9, 0x69d18d2bd1a5c42f, 0xf4892bcb929b0690, 0x89b4443b4ddbc49a, 0x4eb7f8719c36de1e, 0x03e7aa020c6e4141, 0x9b1f5b424d93c9a7 },
+	{ 0x7261445183235adb, 0x0e38dc92cb1f2a60, 0x7b2b8a9aa6079c54, 0x800a440bdbb2ceb1, 0x3cd955b7e00d0984, 0x3a7d3a1b25894224, 0x944c9ad8ec165fde, 0x378f5a541631229b },
+	{ 0x74b4c7fb98459ced, 0x3698fad1153bb6c3, 0x7a1e6c303b7652f4, 0x9fe76702af69334b, 0x1fffe18a1b336103, 0x8941e71cff8a78db, 0x382ae548b2e4f3f3, 0xabbedea680056f52 },
+	{ 0x6bcaa4cd81f32d1b, 0xdea2594ac06fd85d, 0xefbacd1d7d476e98, 0x8a1d71efea48b9ca, 0x2001802114846679, 0xd8fa6bbbebab0761, 0x3002c6cd635afe94, 0x7bcd9ed0efc889fb },
+	{ 0x48bc924af11bd720, 0xfaf417d5d9b21b99, 0xe71da4aa88e12852, 0x5d80ef9d1891cc86, 0xf82012d430219f9b, 0xcda43c32bcdf1d77, 0xd21380b00449b17a, 0x378ee767f11631ba }
+};
+
+const ulong[256][8] TR @align(ulong.sizeof) = {
+	{
+		0xd01f715b5c7ef8e6, 0x16fa240980778325, 0xa8a42e857ee049c8, 0x6ac1068fa186465b, 0x6e417bd7a2e9320b, 0x665c8167a437daab, 0x7666681aa89617f6, 0x4b959163700bdcf5,
+		0xf14be6b78df36248, 0xc585bd689a625cff, 0x9557d7fca67d82cb, 0x89f0b969af6dd366, 0xb0833d48749f6c35, 0xa1998c23b1ecbc7c, 0x8d70c431ac02a736, 0xd6dfbc2fd0a8b69e,
+		0x37aeb3e551fa198b, 0x0b7d128a40b5cf9c, 0x5a8f2008b5780cbc, 0xedec882284e333e5, 0xd25fc177d3c7c2ce, 0x5e0f5d50b61778ec, 0x1d873683c0c24cb9, 0xad040bcbb45d208c,
+		0x2f89a0285b853c76, 0x5732fff6791b8d58, 0x3e9311439ef6ec3f, 0xc9183a809fd3c00f, 0x83adf3f5260a01ee, 0xa6791941f4e8ef10, 0x103ae97d0ca1cd5d, 0x2ce948121dee1b4a,
+		0x39738421dbf2bf53, 0x093da2a6cf0cf5b4, 0xcd9847d89cbcb45f, 0xf9561c078b2d8ae8, 0x9c6a755a6971777f, 0xbc1ebaa0712ef0c5, 0x72e61542abf963a6, 0x78bb5fde229eb12e,
+		0x14ba94250fceb90d, 0x844d6697630e5282, 0x98ea08026a1e032f, 0xf06bbea144217f5c, 0xdb6263d11ccb377a, 0x641c314b2b8ee083, 0x320e96ab9b4770cf, 0x1ee7deb986a96b85,
+		0xe96cf57a878c47b5, 0xfdd6615f8842feb8, 0xc83862965601dd1b, 0x2ea9f83e92572162, 0xf876441142ff97fc, 0xeb2c455608357d9d, 0x5612a7e0b0c9904c, 0x6c01cbfb2d500823,
+		0x4548a6a7fa037a2d, 0xabc4c6bf388b6ef4, 0xbade77d4fdf8bebd, 0x799b07c8eb4cac3a, 0x0c9d87e805b19cf0, 0xcb588aac106afa27, 0xea0c1d40c1e76089, 0x2869354a1e816f1a,
+		0xff96d17307fbc490, 0x9f0a9d602f1a5043, 0x96373fc6e016a5f7, 0x5292dab8b3a6e41c, 0x9b8ae0382c752413, 0x4f15ec3b7364a8a5, 0x3fb349555724f12b, 0xc7c50d4415db66d7,
+		0x92b7429ee379d1a7, 0xd37f99611a15dfda, 0x231427c05e34a086, 0xa439a96d7b51d538, 0xb403401077f01865, 0xdda2aea5901d7902, 0x0a5d4a9c8967d288, 0xc265280adf660f93,
+		0x8bb0094520d4e94e, 0x2a29856691385532, 0x42a833c5bf072941, 0x73c64d54622b7eb2, 0x07e095624504536c, 0x8a905153e906f45a, 0x6f6123c16b3b2f1f, 0xc6e55552dc097bc3,
+		0x4468feb133d16739, 0xe211e7f0c7398829, 0xa2f96419f7879b40, 0x19074bdbc3ad38e9, 0xf4ebc3f9474e0b0c, 0x43886bd376d53455, 0xd8028beb5aa01046, 0x51f23282f5cdc320,
+		0xe7b1c2be0d84e16d, 0x081dfab006dee8a0, 0x3b33340d544b857b, 0x7f5bcabc679ae242, 0x0edd37c48a08a6d8, 0x81ed43d9a9b33bc6, 0xb1a3655ebd4d7121, 0x69a1eeb5e7ed6167,
+		0xf6ab73d5c8f73124, 0x1a67a3e185c61fd5, 0x2dc91004d43c065e, 0x0240b02c8fb93a28, 0x90f7f2b26cc0eb8f, 0x3cd3a16f114fd617, 0xaae49ea9f15973e0, 0x06c0cd748cd64e78,
+		0xda423bc7d5192a6e, 0xc345701c16b41287, 0x6d2193ede4821537, 0xfcf639494190e3ac, 0x7c3b228621f1c57e, 0xfb16ac2b0494b0c0, 0xbf7e529a3745d7f9, 0x6881b6a32e3f7c73,
+		0xca78d2bad9b8e733, 0xbbfe2fc2342aa3a9, 0x0dbddffecc6381e4, 0x70a6a56e2440598e, 0xe4d12a844befc651, 0x8c509c2765d0ba22, 0xee8c6018c28814d9, 0x17da7c1f49a59e31,
+		0x609c4c1328e194d3, 0xb3e3d57232f44b09, 0x91d7aaa4a512f69b, 0x0ffd6fd243dabbcc, 0x50d26a943c1fde34, 0x6be15e9968545b4f, 0x94778fea6faf9fdf, 0x2b09dd7058ea4826,
+		0x677cd9716de5c7bf, 0x49d5214fffb2e6dd, 0x0360e83a466b273c, 0x1fc786af4f7b7691, 0xa0b9d435783ea168, 0xd49f0c035f118cb6, 0x01205816c9d21d14, 0xac2453dd7d8f3d98,
+		0x545217cc3f70aa64, 0x26b4028e9489c9c2, 0xdec2469fd6765e3e, 0x04807d58036f7450, 0xe5f17292823ddb45, 0xf30b569b024a5860, 0x62dcfc3fa758aefb, 0xe84cad6c4e5e5aa1,
+		0xccb81fce556ea94b, 0x53b282ae7a74f908, 0x1b47fbf74c1402c1, 0x368eebf39828049f, 0x7afbeff2ad278b06, 0xbe5e0a8cfe97caed, 0xcfd8f7f413058e77, 0xf78b2bc301252c30,
+		0x4d555c17fcdd928d, 0x5f2f05467fc565f8, 0x24f4b2a21b30f3ea, 0x860dd6bbecb768aa, 0x4c750401350f8f99, 0x0000000000000000, 0xecccd0344d312ef1, 0xb5231806be220571,
+		0xc105c030990d28af, 0x653c695de25cfd97, 0x159acc33c61ca419, 0xb89ec7f872418495, 0xa9847693b73254dc, 0x58cf90243ac13694, 0x59efc832f3132b80, 0x5c4fed7c39ae42c4,
+		0x828dabe3efd81cfa, 0xd13f294d95ace5f2, 0x7d1b7a90e823d86a, 0xb643f03cf849224d, 0x3df3f979d89dcb03, 0x7426d836272f2dde, 0xdfe21e891fa4432a, 0x3a136c1b9d99986f,
+		0xfa36f43dcd46add4, 0xc025982650df35bb, 0x856d3e81aadc4f96, 0xc4a5e57e53b041eb, 0x4708168b75ba4005, 0xaf44bbe73be41aa4, 0x971767d029c4b8e3, 0xb9be9feebb939981,
+		0x215497ecd18d9aae, 0x316e7e91dd2c57f3, 0xcef8afe2dad79363, 0x3853dc371220a247, 0x35ee03c9de4323a3, 0xe6919aa8c456fc79, 0xe05157dc4880b201, 0x7bdbb7e464f59612,
+		0x127a59518318f775, 0x332ecebd52956ddb, 0x8f30741d23bb9d1e, 0xd922d3fd93720d52, 0x7746300c61440ae2, 0x25d4eab4d2e2eefe, 0x75068020eefd30ca, 0x135a01474acaea61,
+		0x304e268714fe4ae7, 0xa519f17bb283c82c, 0xdc82f6b359cf6416, 0x5baf781e7caa11a8, 0xb2c38d64fb26561d, 0x34ce5bdf17913eb7, 0x5d6fb56af07c5fd0, 0x182713cd0a7f25fd,
+		0x9e2ac576e6c84d57, 0x9aaab82ee5a73907, 0xa3d93c0f3e558654, 0x7e7b92aaae48ff56, 0x872d8ead256575be, 0x41c8dbfff96c0e7d, 0x99ca5014a3cc1e3b, 0x40e883e930be1369,
+		0x1ca76e95091051ad, 0x4e35b42dbab6b5b1, 0x05a0254ecabd6944, 0xe1710fca8152af15, 0xf22b0e8dcb984574, 0xb763a82a319b3f59, 0x63fca4296e8ab3ef, 0x9d4a2d4ca0a36a6b,
+		0xe331bfe60eeb953d, 0xd5bf541596c391a2, 0xf5cb9bef8e9c1618, 0x46284e9dbc685d11, 0x2074cffa185f87ba, 0xbd3ee2b6b8fcedd1, 0xae64e3f1f23607b0, 0xfeb68965ce29d984,
+		0x55724fdaf6a2b770, 0x29496d5cd753720e, 0xa75941573d3af204, 0x8e102c0bea69800a, 0x111ab16bc573d049, 0xd7ffe439197aab8a, 0xefac380e0b5a09cd, 0x48f579593660fbc9,
+		0x22347fd697e6bd92, 0x61bc1405e13389c7, 0x4ab5c975b9d9c1e1, 0x80cd1bcf606126d2, 0x7186fd78ed92449a, 0x93971a882aabccb3, 0x88d0e17f66bfce72, 0x27945a985d5bd4d6
+	},
+	{
+		0xde553f8c05a811c8, 0x1906b59631b4f565, 0x436e70d6b1964ff7, 0x36d343cb8b1e9d85, 0x843dfacc858aab5a, 0xfdfc95c299bfc7f9, 0x0f634bdea1d51fa2, 0x6d458b3b76efb3cd,
+		0x85c3f77cf8593f80, 0x3c91315fbe737cb2, 0x2148b03366ace398, 0x18f8b8264c6761bf, 0xc830c1c495c9fb0f, 0x981a76102086a0aa, 0xaa16012142f35760, 0x35cc54060c763cf6,
+		0x42907d66cc45db2d, 0x8203d44b965af4bc, 0x3d6f3cefc3a0e868, 0xbc73ff69d292bda7, 0x8722ed0102e20a29, 0x8f8185e8cd34deb7, 0x9b0561dda7ee01d9, 0x5335a0193227fad6,
+		0xc9cecc74e81a6fd5, 0x54f5832e5c2431ea, 0x99e47ba05d553470, 0xf7bee756acd226ce, 0x384e05a5571816fd, 0xd1367452a47d0e6a, 0xf29fde1c386ad85b, 0x320c77316275f7ca,
+		0xd0c879e2d9ae9ab0, 0xdb7406c69110ef5d, 0x45505e51a2461011, 0xfc029872e46c5323, 0xfa3cb6f5f7bc0cc5, 0x031f17cd8768a173, 0xbd8df2d9af41297d, 0x9d3b4f5ab43e5e3f,
+		0x4071671b36feee84, 0x716207e7d3e3b83d, 0x48d20ff2f9283a1a, 0x27769eb4757cbc7e, 0x5c56ebc793f2e574, 0xa48b474f9ef5dc18, 0x52cbada94ff46e0c, 0x60c7da982d8199c6,
+		0x0e9d466edc068b78, 0x4eec2175eaf865fc, 0x550b8e9e21f7a530, 0x6b7ba5bc653fec2b, 0x5eb7f1ba6949d0dd, 0x57ea94e3db4c9099, 0xf640eae6d101b214, 0xdd4a284182c0b0bb,
+		0xff1d8fbf6304f250, 0xb8accb933bf9d7e8, 0xe8867c478eb68c4d, 0x3f8e2692391bddc1, 0xcb2fd60912a15a7c, 0xaec935dbab983d2f, 0xf55ffd2b56691367, 0x80e2ce366ce1c115,
+		0x179bf3f8edb27e1d, 0x01fe0db07dd394da, 0xda8a0b76ecc37b87, 0x44ae53e1df9584cb, 0xb310b4b77347a205, 0xdfab323c787b8512, 0x3b511268d070b78e, 0x65e6e3d2b9396753,
+		0x6864b271e2574d58, 0x259784c98fc789d7, 0x02e11a7dfabb35a9, 0x8841a6dfa337158b, 0x7ade78c39b5dcdd0, 0xb7cf804d9a2cc84a, 0x20b6bd831b7f7742, 0x75bd331d3a88d272,
+		0x418f6aab4b2d7a5e, 0xd9951cbb6babdaf4, 0xb6318dfde7ff5c90, 0x1f389b112264aa83, 0x492c024284fbaec0, 0xe33a0363c608f9a0, 0x2688930408af28a4, 0xc7538a1a341ce4ad,
+		0x5da8e677ee2171ae, 0x8c9e92254a5c7fc4, 0x63d8cd55aae938b5, 0x29ebd8daa97a3706, 0x959827b37be88aa1, 0x1484e4356adadf6e, 0xa7945082199d7d6b, 0xbf6ce8a455fa1cd4,
+		0x9cc542eac9edcae5, 0x79c16f0e1c356ca3, 0x89bfab6fdee48151, 0xd4174d1830c5f0ff, 0x9258048415eb419d, 0x6139d72850520d1c, 0x6a85a80c18ec78f1, 0xcd11f88e0171059a,
+		0xcceff53e7ca29140, 0xd229639f2315af19, 0x90b91ef9ef507434, 0x5977d28d074a1be1, 0x311360fce51d56b9, 0xc093a92d5a1f2f91, 0x1a19a25bb6dc5416, 0xeb996b8a09de2d3e,
+		0xfee3820f1ed7668a, 0xd7085ad5b7ad518c, 0x7fff41890fe53345, 0xec5948bd67dde602, 0x2fd5f65dbaaa68e0, 0xa5754affe32648c2, 0xf8ddac880d07396c, 0x6fa491468c548664,
+		0x0c7c5c1326bdbed1, 0x4a33158f03930fb3, 0x699abfc19f84d982, 0xe4fa2054a80b329c, 0x6707f9af438252fa, 0x08a368e9cfd6d49e, 0x47b1442c58fd25b8, 0xbbb3dc5ebc91769b,
+		0x1665fe489061eac7, 0x33f27a811fa66310, 0x93a609346838d547, 0x30ed6d4c98cec263, 0x1dd9816cd8df9f2a, 0x94662a03063b1e7b, 0x83fdd9fbeb896066, 0x7b207573e68e590a,
+		0x5f49fc0a149a4407, 0x343259b671a5a82c, 0xfbc2bb458a6f981f, 0xc272b350a0a41a38, 0x3aaf1fd8ada32354, 0x6cbb868b0b3c2717, 0xa2b569c88d2583fe, 0xf180c9d1bf027928,
+		0xaf37386bd64ba9f5, 0x12bacab2790a8088, 0x4c0d3b0810435055, 0xb2eeb9070e9436df, 0xc5b29067cea7d104, 0xdcb425f1ff132461, 0x4f122cc5972bf126, 0xac282fa651230886,
+		0xe7e537992f6393ef, 0xe61b3a2952b00735, 0x709c0a57ae302ce7, 0xe02514ae416058d3, 0xc44c9dd7b37445de, 0x5a68c5408022ba92, 0x1c278cdca50c0bf0, 0x6e5a9cf6f18712be,
+		0x86dce0b17f319ef3, 0x2d34ec2040115d49, 0x4bcd183f7e409b69, 0x2815d56ad4a9a3dc, 0x24698979f2141d0d, 0x0000000000000000, 0x1ec696a15fb73e59, 0xd86b110b16784e2e,
+		0x8e7f8858b0e74a6d, 0x063e2e8713d05fe6, 0xe2c40ed3bbdb6d7a, 0xb1f1aeca89fc97ac, 0xe1db191e3cb3cc09, 0x6418ee62c4eaf389, 0xc6ad87aa49cf7077, 0xd6f65765ca7ec556,
+		0x9afb6c6dda3d9503, 0x7ce05644888d9236, 0x8d609f95378feb1e, 0x23a9aa4e9c17d631, 0x6226c0e5d73aac6f, 0x56149953a69f0443, 0xeeb852c09d66d3ab, 0x2b0ac2a753c102af,
+		0x07c023376e03cb3c, 0x2ccae1903dc2c993, 0xd3d76e2f5ec63bc3, 0x9e2458973356ff4c, 0xa66a5d32644ee9b1, 0x0a427294356de137, 0x783f62be61e6f879, 0x1344c70204d91452,
+		0x5b96c8f0fdf12e48, 0xa90916ecc59bf613, 0xbe92e5142829880e, 0x727d102a548b194e, 0x1be7afebcb0fc0cc, 0x3e702b2244c8491b, 0xd5e940a84d166425, 0x66f9f41f3e51c620,
+		0xabe80c913f20c3ba, 0xf07ec461c2d1edf2, 0xf361d3ac45b94c81, 0x0521394a94b8fe95, 0xadd622162cf09c5c, 0xe97871f7f3651897, 0xf4a1f09b2bba87bd, 0x095d6559b2054044,
+		0x0bbc7f2448be75ed, 0x2af4cf172e129675, 0x157ae98517094bb4, 0x9fda55274e856b96, 0x914713499283e0ee, 0xb952c623462a4332, 0x74433ead475b46a8, 0x8b5eb112245fb4f8,
+		0xa34b6478f0f61724, 0x11a5dd7ffe6221fb, 0xc16da49d27ccbb4b, 0x76a224d0bde07301, 0x8aa0bca2598c2022, 0x4df336b86d90c48f, 0xea67663a740db9e4, 0xef465f70e0b54771,
+		0x39b008152acb8227, 0x7d1e5bf4f55e06ec, 0x105bd0cf83b1b521, 0x775c2960c033e7db, 0x7e014c397236a79f, 0x811cc386113255cf, 0xeda7450d1a0e72d8, 0x5889df3d7a998f3b,
+		0x2e2bfbedc779fc3a, 0xce0eef438619a4e9, 0x372d4e7bf6cd095f, 0x04df34fae96b6a4f, 0xf923a13870d4adb6, 0xa1aa7e050a4d228d, 0xa8f71b5cb84862c9, 0xb52e9a306097fde3,
+		0x0d8251a35b6e2a0b, 0x2257a7fee1c442eb, 0x73831d9a29588d94, 0x51d4ba64c89ccf7f, 0x502ab7d4b54f5ba5, 0x97793dce8153bf08, 0xe5042de4d5d8a646, 0x9687307efc802bd2,
+		0xa05473b5779eb657, 0xb4d097801d446939, 0xcff0e2f3fbca3033, 0xc38cbee0dd778ee2, 0x464f499c252eb162, 0xcad1dbb96f72cea6, 0xba4dd1eec142e241, 0xb00fa37af42f0376
+	},
+	{
+		0xcce4cd3aa968b245, 0x089d5484e80b7faf, 0x638246c1b3548304, 0xd2fe0ec8c2355492, 0xa7fbdf7ff2374eee, 0x4df1600c92337a16, 0x84e503ea523b12fb, 0x0790bbfd53ab0c4a,
+		0x198a780f38f6ea9d, 0x2ab30c8f55ec48cb, 0xe0f7fed6b2c49db5, 0xb6ecf3f422cadbdc, 0x409c9a541358df11, 0xd3ce8a56dfde3fe3, 0xc3e9224312c8c1a0, 0x0d6dfa58816ba507,
+		0xddf3e1b179952777, 0x04c02a42748bb1d9, 0x94c2abff9f2decb8, 0x4f91752da8f8acf4, 0x78682befb169bf7b, 0xe1c77a48af2ff6c4, 0x0c5d7ec69c80ce76, 0x4cc1e4928fd81167,
+		0xfeed3d24d9997b62, 0x518bb6dfc3a54a23, 0x6dbf2d26151f9b90, 0xb5bc624b05ea664f, 0xe86aaa525acfe21a, 0x4801ced0fb53a0be, 0xc91463e6c00868ed, 0x1027a815cd16fe43,
+		0xf67069a0319204cd, 0xb04ccc976c8abce7, 0xc0b9b3fc35e87c33, 0xf380c77c58f2de65, 0x50bb3241de4e2152, 0xdf93f490435ef195, 0xf1e0d25d62390887, 0xaf668bfb1a3c3141,
+		0xbc11b251f00a7291, 0x73a5eed47e427d47, 0x25bee3f6ee4c3b2e, 0x43cc0beb34786282, 0xc824e778dde3039c, 0xf97d86d98a327728, 0xf2b043e24519b514, 0xe297ebf7880f4b57,
+		0x3a94a49a98fab688, 0x868516cb68f0c419, 0xeffa11af0964ee50, 0xa4ab4ec0d517f37d, 0xa9c6b498547c567a, 0x8e18424f80fbbbb6, 0x0bcdc53bcf2bc23c, 0x137739aaea3643d0,
+		0x2c1333ec1bac2ff0, 0x8d48d3f0a7db0625, 0x1e1ac3f26b5de6d7, 0xf520f81f16b2b95e, 0x9f0f6ec450062e84, 0x0130849e1deb6b71, 0xd45e31ab8c7533a9, 0x652279a2fd14e43f,
+		0x3209f01e70f1c927, 0xbe71a770cac1a473, 0x0e3d6be7a64b1894, 0x7ec8148cff29d840, 0xcb7476c7fac3be0f, 0x72956a4a63a91636, 0x37f95ec21991138f, 0x9e3fea5a4ded45f5,
+		0x7b38ba50964902e8, 0x222e580bbde73764, 0x61e253e0899f55e6, 0xfc8d2805e352ad80, 0x35994be3235ac56d, 0x09add01af5e014de, 0x5e8659a6780539c6, 0xb17c48097161d796,
+		0x026015213acbd6e2, 0xd1ae9f77e515e901, 0xb7dc776a3f21b0ad, 0xaba6a1b96eb78098, 0x9bcf4486248d9f5d, 0x582666c536455efd, 0xfdbdac9bfeb9c6f1, 0xc47999be4163cdea,
+		0x765540081722a7ef, 0x3e548ed8ec710751, 0x3d041f67cb51bac2, 0x7958af71ac82d40a, 0x36c9da5c047a78fe, 0xed9a048e33af38b2, 0x26ee7249c96c86bd, 0x900281bdeba65d61,
+		0x11172c8bd0fd9532, 0xea0abf73600434f8, 0x42fc8f75299309f3, 0x34a9cf7d3eb1ae1c, 0x2b838811480723ba, 0x5ce64c8742ceef24, 0x1adae9b01fd6570e, 0x3c349bf9d6bad1b3,
+		0x82453c891c7b75c0, 0x97923a40b80d512b, 0x4a61dbf1c198765c, 0xb48ce6d518010d3e, 0xcfb45c858e480fd6, 0xd933cbf30d1e96ae, 0xd70ea014ab558e3a, 0xc189376228031742,
+		0x9262949cd16d8b83, 0xeb3a3bed7def5f89, 0x49314a4ee6b8cbcf, 0xdcc3652f647e4c06, 0xda635a4c2a3e2b3d, 0x470c21a940f3d35b, 0x315961a157d174b4, 0x6672e81dda3459ac,
+		0x5b76f77a1165e36e, 0x445cb01667d36ec8, 0xc5491d205c88a69b, 0x456c34887a3805b9, 0xffddb9bac4721013, 0x99af51a71e4649bf, 0xa15be01cbc7729d5, 0x52db2760e485f7b0,
+		0x8c78576eba306d54, 0xae560f6507d75a30, 0x95f22f6182c687c9, 0x71c5fbf54489aba5, 0xca44f259e728d57e, 0x88b87d2ccebbdc8d, 0xbab18d32be4a15aa, 0x8be8ec93e99b611e,
+		0x17b713e89ebdf209, 0xb31c5d284baa0174, 0xeeca9531148f8521, 0xb8d198138481c348, 0x8988f9b2d350b7fc, 0xb9e11c8d996aa839, 0x5a4673e40c8e881f, 0x1687977683569978,
+		0xbf4123eed72acf02, 0x4ea1f1b3b513c785, 0xe767452be16f91ff, 0x7505d1b730021a7c, 0xa59bca5ec8fc980c, 0xad069eda20f7e7a3, 0x38f4b1bba231606a, 0x60d2d77e94743e97,
+		0x9affc0183966f42c, 0x248e6768f3a7505f, 0xcdd449a4b483d934, 0x87b59255751baf68, 0x1bea6d2e023d3c7f, 0x6b1f12455b5ffcab, 0x743555292de9710d, 0xd8034f6d10f5fddf,
+		0xc6198c9f7ba81b08, 0xbb8109aca3a17edb, 0xfa2d1766ad12cabb, 0xc729080166437079, 0x9c5fff7b77269317, 0x0000000000000000, 0x15d706c9a47624eb, 0x6fdf38072fd44d72,
+		0x5fb6dd3865ee52b7, 0xa33bf53d86bcff37, 0xe657c1b5fc84fa8e, 0xaa962527735cebe9, 0x39c43525bfda0b1b, 0x204e4d2a872ce186, 0x7a083ece8ba26999, 0x554b9c9db72efbfa,
+		0xb22cd9b656416a05, 0x96a2bedea5e63a5a, 0x802529a826b0a322, 0x8115ad363b5bc853, 0x8375b81701901eb1, 0x3069e53f4a3a1fc5, 0xbd2136cfede119e0, 0x18bafc91251d81ec,
+		0x1d4a524d4c7d5b44, 0x05f0aedc6960daa8, 0x29e39d3072ccf558, 0x70f57f6b5962c0d4, 0x989fd53903ad22ce, 0xf84d024797d91c59, 0x547b1803aac5908b, 0xf0d056c37fd263f6,
+		0xd56eb535919e58d8, 0x1c7ad6d351963035, 0x2e7326cd2167f912, 0xac361a443d1c8cd2, 0x697f076461942a49, 0x4b515f6fdc731d2d, 0x8ad8680df4700a6f, 0x41ac1eca0eb3b460,
+		0x7d988533d80965d3, 0xa8f6300649973d0b, 0x7765c4960ac9cc9e, 0x7ca801adc5e20ea2, 0xdea3700e5eb59ae4, 0xa06b6482a19c42a4, 0x6a2f96db46b497da, 0x27def6d7d487edcc,
+		0x463ca5375d18b82a, 0xa6cb5be1efdc259f, 0x53eba3fef96e9cc1, 0xce84d81b93a364a7, 0xf4107c810b59d22f, 0x333974806d1aa256, 0x0f0def79bba073e5, 0x231edc95a00c5c15,
+		0xe437d494c64f2c6c, 0x91320523f64d3610, 0x67426c83c7df32dd, 0x6eefbc99323f2603, 0x9d6f7be56acdf866, 0x5916e25b2bae358c, 0x7ff89012e2c2b331, 0x035091bf2720bd93,
+		0x561b0d22900e4669, 0x28d319ae6f279e29, 0x2f43a2533c8c9263, 0xd09e1be9f8fe8270, 0xf740ed3e2c796fbc, 0xdb53ded237d5404c, 0x62b2c25faebfe875, 0x0afd41a5d2c0a94d,
+		0x6412fd3ce0ff8f4e, 0xe3a76f6995e42026, 0x6c8fa9b808f4f0e1, 0xc2d9a6dd0f23aad1, 0x8f28c6d19d10d0c7, 0x85d587744fd0798a, 0xa20b71a39b579446, 0x684f83fa7c7f4138,
+		0xe507500adba4471d, 0x3f640a46f19a6c20, 0x1247bd34f7dd28a1, 0x2d23b77206474481, 0x93521002cc86e0f2, 0x572b89bc8de52d18, 0xfb1d93f8b0f9a1ca, 0xe95a2ecc4724896b,
+		0x3ba420048511ddf9, 0xd63e248ab6bee54b, 0x5dd6c8195f258455, 0x06a03f634e40673b, 0x1f2a476c76b68da6, 0x217ec9b49ac78af7, 0xecaa80102e4453c3, 0x14e78257b99d4f9a
+	},
+	{
+		0x20329b2cc87bba05, 0x4f5eb6f86546a531, 0xd4f44775f751b6b1, 0x8266a47b850dfa8b, 0xbb986aa15a6ca985, 0xc979eb08f9ae0f99, 0x2da6f447a2375ea1, 0x1e74275dcd7d8576,
+		0xbc20180a800bc5f8, 0xb4a2f701b2dc65be, 0xe726946f981b6d66, 0x48e6c453bf21c94c, 0x42cad9930f0a4195, 0xefa47b64aacccd20, 0x71180a8960409a42, 0x8bb3329bf6a44e0c,
+		0xd34c35de2d36dacc, 0xa92f5b7cbc23dc96, 0xb31a85aa68bb09c3, 0x13e04836a73161d2, 0xb24dfc4129c51d02, 0x8ae44b70b7da5acd, 0xe671ed84d96579a7, 0xa4bb3417d66f3832,
+		0x4572ab38d56d2de8, 0xb1b47761ea47215c, 0xe81c09cf70aba15d, 0xffbdb872ce7f90ac, 0xa8782297fd5dc857, 0x0d946f6b6a4ce4a4, 0xe4df1f4f5b995138, 0x9ebc71edca8c5762,
+		0x0a2c1dc0b02b88d9, 0x3b503c115d9d7b91, 0xc64376a8111ec3a2, 0xcec199a323c963e4, 0xdc76a87ec58616f7, 0x09d596e073a9b487, 0x14583a9d7d560daf, 0xf4c6dc593f2a0cb4,
+		0xdd21d19584f80236, 0x4a4836983ddde1d3, 0xe58866a41ae745f9, 0xf591a5b27e541875, 0x891dc05074586693, 0x5b068c651810a89e, 0xa30346bc0c08544f, 0x3dbf3751c684032d,
+		0x2a1e86ec785032dc, 0xf73f5779fca830ea, 0xb60c05ca30204d21, 0x0cc316802b32f065, 0x8770241bdd96be69, 0xb861e18199ee95db, 0xf805cad91418fcd1, 0x29e70dccbbd20e82,
+		0xc7140f435060d763, 0x0f3a9da0e8b0cc3b, 0xa2543f574d76408e, 0xbd7761e1c175d139, 0x4b1f4f737ca3f512, 0x6dc2df1f2fc137ab, 0xf1d05c3967b14856, 0xa742bf3715ed046c,
+		0x654030141d1697ed, 0x07b872abda676c7d, 0x3ce84eba87fa17ec, 0xc1fb0403cb79afdf, 0x3e46bc7105063f73, 0x278ae987121cd678, 0xa1adb4778ef47cd0, 0x26dd906c5362c2b9,
+		0x05168060589b44e2, 0xfbfc41f9d79ac08f, 0x0e6de44ba9ced8fa, 0x9feb08068bf243a3, 0x7b341749d06b129b, 0x229c69e74a87929a, 0xe09ee6c4427c011b, 0x5692e30e725c4c3a,
+		0xda99a33e5e9f6e4b, 0x353dd85af453a36b, 0x25241b4c90e0fee7, 0x5de987258309d022, 0xe230140fc0802984, 0x93281e86a0c0b3c6, 0xf229d719a4337408, 0x6f6c2dd4ad3d1f34,
+		0x8ea5b2fbae3f0aee, 0x8331dd90c473ee4a, 0x346aa1b1b52db7aa, 0xdf8f235e06042aa9, 0xcc6f6b68a1354b7b, 0x6c95a6f46ebf236a, 0x52d31a856bb91c19, 0x1a35ded6d498d555,
+		0xf37eaef2e54d60c9, 0x72e181a9a3c2a61c, 0x98537aad51952fde, 0x16f6c856ffaa2530, 0xd960281e9d1d5215, 0x3a0745fa1ce36f50, 0x0b7b642bf1559c18, 0x59a87eae9aec8001,
+		0x5e100c05408bec7c, 0x0441f98b19e55023, 0xd70dcc5534d38aef, 0x927f676de1bea707, 0x9769e70db925e3e5, 0x7a636ea29115065a, 0x468b201816ef11b6, 0xab81a9b73edff409,
+		0xc0ac7de88a07bb1e, 0x1f235eb68c0391b7, 0x6056b074458dd30f, 0xbe8eeac102f7ed67, 0xcd381283e04b5fba, 0x5cbefecec277c4e3, 0xd21b4c356c48ce0d, 0x1019c31664b35d8c,
+		0x247362a7d19eea26, 0xebe582efb3299d03, 0x02aef2cb82fc289f, 0x86275df09ce8aaa8, 0x28b07427faac1a43, 0x38a9b7319e1f47cf, 0xc82e92e3b8d01b58, 0x06ef0b409b1978bc,
+		0x62f842bfc771fb90, 0x9904034610eb3b1f, 0xded85ab5477a3e68, 0x90d195a663428f98, 0x5384636e2ac708d8, 0xcbd719c37b522706, 0xae9729d76644b0eb, 0x7c8c65e20a0c7ee6,
+		0x80c856b007f1d214, 0x8c0b40302cc32271, 0xdbcedad51fe17a8a, 0x740e8ae938dbdea0, 0xa615c6dc549310ad, 0x19cc55f6171ae90b, 0x49b1bdb8fe5fdd8d, 0xed0a89af2830e5bf,
+		0x6a7aadb4f5a65bd6, 0x7e22972988f05679, 0xf952b3325566e810, 0x39fecedadf61530e, 0x6101c99f04f3c7ce, 0x2e5f7f6761b562ff, 0xf08725d226cf5c97, 0x63af3b54860fef51,
+		0x8ff2cb10ef411e2f, 0x884ab9bb35267252, 0x4df04433e7ba8dae, 0x9afd8866d3690741, 0x66b9bb34de94abb3, 0x9baaf18d92171380, 0x543c11c5f0a064a5, 0x17a1b1bdbed431f1,
+		0xb5f58eeaf3a2717f, 0xc355f6c849858740, 0xec5df044694ef17e, 0xd83751f5dc6346d4, 0xfc4433520dfdacf2, 0x0000000000000000, 0x5a51f58e596ebc5f, 0x3285aaf12e34cf16,
+		0x8d5c39db6dbd36b0, 0x12b731dde64f7513, 0x94906c2d7aa7dfbb, 0x302b583aacc8e789, 0x9d45facd090e6b3c, 0x2165e2c78905aec4, 0x68d45f7f775a7349, 0x189b2c1d5664fdca,
+		0xe1c99f2f030215da, 0x6983269436246788, 0x8489af3b1e148237, 0xe94b702431d5b59c, 0x33d2d31a6f4adbd7, 0xbfd9932a4389f9a6, 0xb0e30e8aab39359d, 0xd1e2c715afcaf253,
+		0x150f43763c28196e, 0xc4ed846393e2eb3d, 0x03f98b20c3823c5e, 0xfd134ab94c83b833, 0x556b682eb1de7064, 0x36c4537a37d19f35, 0x7559f30279a5ca61, 0x799ae58252973a04,
+		0x9c12832648707ffd, 0x78cd9c6913e92ec5, 0x1d8dac7d0effb928, 0x439da0784e745554, 0x413352b3cc887dcb, 0xbacf134a1b12bd44, 0x114ebafd25cd494d, 0x2f08068c20cb763e,
+		0x76a07822ba27f63f, 0xeab2fb04f25789c2, 0xe3676de481fe3d45, 0x1b62a73d95e6c194, 0x641749ff5c68832c, 0xa5ec4dfc97112cf3, 0xf6682e92bdd6242b, 0x3f11c59a44782bb2,
+		0x317c21d1edb6f348, 0xd65ab5be75ad9e2e, 0x6b2dd45fb4d84f17, 0xfaab381296e4d44e, 0xd0b5befeeeb4e692, 0x0882ef0b32d7a046, 0x512a91a5a83b2047, 0x963e9ee6f85bf724,
+		0x4e09cf132438b1f0, 0x77f701c9fb59e2fe, 0x7ddb1c094b726a27, 0x5f4775ee01f5f8bd, 0x9186ec4d223c9b59, 0xfeeac1998f01846d, 0xac39db1ce4b89874, 0xb75b7c21715e59e0,
+		0xafc0503c273aa42a, 0x6e3b543fec430bf5, 0x704f7362213e8e83, 0x58ff0745db9294c0, 0x67eec2df9feabf72, 0xa0facd9ccf8a6811, 0xb936986ad890811a, 0x95c715c63bd9cb7a,
+		0xca8060283a2c33c7, 0x507de84ee9453486, 0x85ded6d05f6a96f6, 0x1cdad5964f81ade9, 0xd5a33e9eb62fa270, 0x40642b588df6690a, 0x7f75eec2c98e42b8, 0x2cf18dace3494a60,
+		0x23cb100c0bf9865b, 0xeef3028febb2d9e1, 0x4425d2d394133929, 0xaad6d05c7fa1e0c8, 0xad6ea2f7a5c68cb5, 0xc2028f2308fb9381, 0x819f2f5b468fc6d5, 0xc5bafd88d29cfffc,
+		0x47dc59f357910577, 0x2b49ff07392e261d, 0x57c59ae5332258fb, 0x73b6f842e2bcb2dd, 0xcf96e04862b77725, 0x4ca73dd8a6c4996f, 0x015779eb417e14c1, 0x37932a9176af8bf4
+	},
+	{
+		0x190a2c9b249df23e, 0x2f62f8b62263e1e9, 0x7a7f754740993655, 0x330b7ba4d5564d9f, 0x4c17a16a46672582, 0xb22f08eb7d05f5b8, 0x535f47f40bc148cc, 0x3aec5d27d4883037,
+		0x10ed0a1825438f96, 0x516101f72c233d17, 0x13cc6f949fd04eae, 0x739853c441474bfd, 0x653793d90d3f5b1b, 0x5240647b96b0fc2f, 0x0c84890ad27623e0, 0xd7189b32703aaea3,
+		0x2685de3523bd9c41, 0x99317c5b11bffefa, 0x0d9baa854f079703, 0x70b93648fbd48ac5, 0xa80441fce30bc6be, 0x7287704bdc36ff1e, 0xb65384ed33dc1f13, 0xd36417343ee34408,
+		0x39cd38ab6e1bf10f, 0x5ab861770a1f3564, 0x0ebacf09f594563b, 0xd04572b884708530, 0x3cae9722bdb3af47, 0x4a556b6f2f5cbaf2, 0xe1704f1f76c4bd74, 0x5ec4ed7144c6dfcf,
+		0x16afc01d4c7810e6, 0x283f113cd629ca7a, 0xaf59a8761741ed2d, 0xeed5a3991e215fac, 0x3bf37ea849f984d4, 0xe413e096a56ce33c, 0x2c439d3a98f020d1, 0x637559dc6404c46b,
+		0x9e6c95d1e5f5d569, 0x24bb9836045fe99a, 0x44efa466dac8ecc9, 0xc6eab2a5c80895d6, 0x803b50c035220cc4, 0x0321658cba93c138, 0x8f9ebc465dc7ee1c, 0xd15a5137190131d3,
+		0x0fa5ec8668e5e2d8, 0x91c979578d1037b1, 0x0642ca05693b9f70, 0xefca80168350eb4f, 0x38d21b24f36a45ec, 0xbeab81e1af73d658, 0x8cbfd9cae7542f24, 0xfd19cc0d81f11102,
+		0x0ac6430fbb4dbc90, 0x1d76a09d6a441895, 0x2a01573ff1cbbfa1, 0xb572e161894fde2b, 0x8124734fa853b827, 0x614b1fdf43e6b1b0, 0x68ac395c4238cc18, 0x21d837bfd7f7b7d2,
+		0x20c714304a860331, 0x5cfaab726324aa14, 0x74c5ba4eb50d606e, 0xf3a3030474654739, 0x23e671bcf015c209, 0x45f087e947b9582a, 0xd8bd77b418df4c7b, 0xe06f6c90ebb50997,
+		0x0bd96080263c0873, 0x7e03f9410e40dcfe, 0xb8e94be4c6484928, 0xfb5b0608e8ca8e72, 0x1a2b49179e0e3306, 0x4e29e76961855059, 0x4f36c4e6fcf4e4ba, 0x49740ee395cf7bca,
+		0xc2963ea386d17f7d, 0x90d65ad810618352, 0x12d34c1b02a1fa4d, 0xfa44258775bb3a91, 0x18150f14b9ec46dd, 0x1491861e6b9a653d, 0x9a1019d7ab2c3fc2, 0x3668d42d06fe13d7,
+		0xdcc1fbb25606a6d0, 0x969490dd795a1c22, 0x3549b1a1bc6dd2ef, 0xc94f5e23a0ed770e, 0xb9f6686b5b39fdcb, 0xc4d4f4a6efeae00d, 0xe732851a1fff2204, 0x94aad6de5eb869f9,
+		0x3f8ff2ae07206e7f, 0xfe38a9813b62d03a, 0xa7a1ad7a8bee2466, 0x7b6056c8dde882b6, 0x302a1e286fc58ca7, 0x8da0fa457a259bc7, 0xb3302b64e074415b, 0x5402ae7eff8b635f,
+		0x08f8050c9cafc94b, 0xae468bf98a3059ce, 0x88c355cca98dc58f, 0xb10e6d67c7963480, 0xbad70de7e1aa3cf3, 0xbfb4a26e320262bb, 0xcb711820870f02d5, 0xce12b7a954a75c9d,
+		0x563ce87dd8691684, 0x9f73b65e7884618a, 0x2b1e74b06cba0b42, 0x47cec1ea605b2df1, 0x1c698312f735ac76, 0x5fdbcefed9b76b2c, 0x831a354c8fb1cdfc, 0x820516c312c0791f,
+		0xb74ca762aeadabf0, 0xfc06ef821c80a5e1, 0x5723cbf24518a267, 0x9d4df05d5f661451, 0x588627742dfd40bf, 0xda8331b73f3d39a0, 0x17b0e392d109a405, 0xf965400bcf28fba9,
+		0x7c3dbf4229a2a925, 0x023e460327e275db, 0x6cd0b55a0ce126b3, 0xe62da695828e96e7, 0x42ad6e63b3f373b9, 0xe50cc319381d57df, 0xc5cbd729729b54ee, 0x46d1e265fd2a9912,
+		0x6428b056904eeff8, 0x8be23040131e04b7, 0x6709d5da2add2ec0, 0x075de98af44a2b93, 0x8447dcc67bfbe66f, 0x6616f655b7ac9a23, 0xd607b8bded4b1a40, 0x0563af89d3a85e48,
+		0x3db1b4ad20c21ba4, 0x11f22997b8323b75, 0x292032b34b587e99, 0x7f1cdace9331681d, 0x8e819fc9c0b65aff, 0xa1e3677fe2d5bb16, 0xcd33d225ee349da5, 0xd9a2543b85aef898,
+		0x795e10cbfa0af76d, 0x25a4bbb9992e5d79, 0x78413344677b438e, 0xf0826688cef68601, 0xd27b34bba392f0eb, 0x551d8df162fad7bc, 0x1e57c511d0d7d9ad, 0xdeffbdb171e4d30b,
+		0xf4feea8e802f6caa, 0xa480c8f6317de55e, 0xa0fc44f07fa40ff5, 0x95b5f551c3c9dd1a, 0x22f952336d6476ea, 0x0000000000000000, 0xa6be8ef5169f9085, 0xcc2cf1aa73452946,
+		0x2e7ddb39bf12550a, 0xd526dd3157d8db78, 0x486b2d6c08becf29, 0x9b0f3a58365d8b21, 0xac78cdfaadd22c15, 0xbc95c7e28891a383, 0x6a927f5f65dab9c3, 0xc3891d2c1ba0cb9e,
+		0xeaa92f9f50f8b507, 0xcf0d9426c9d6e87e, 0xca6e3baf1a7eb636, 0xab25247059980786, 0x69b31ad3df4978fb, 0xe2512a93cc577c4c, 0xff278a0ea61364d9, 0x71a615c766a53e26,
+		0x89dc764334fc716c, 0xf87a638452594f4a, 0xf2bc208be914f3da, 0x8766b94ac1682757, 0xbbc82e687cdb8810, 0x626a7a53f9757088, 0xa2c202f358467a2e, 0x4d0882e5db169161,
+		0x09e7268301de7da8, 0xe897699c771ac0dc, 0xc8507dac3d9cc3ed, 0xc0a878a0a1330aa6, 0x978bb352e42ba8c1, 0xe9884a13ea6b743f, 0x279afdbabecc28a2, 0x047c8c064ed9eaab,
+		0x507e2278b15289f4, 0x599904fbb08cf45c, 0xbd8ae46d15e01760, 0x31353da7f2b43844, 0x8558ff49e68a528c, 0x76fbfc4d92ef15b5, 0x3456922e211c660c, 0x86799ac55c1993b4,
+		0x3e90d1219a51da9c, 0x2d5cbeb505819432, 0x982e5fd48cce4a19, 0xdb9c1238a24c8d43, 0xd439febecaa96f9b, 0x418c0bef0960b281, 0x158ea591f6ebd1de, 0x1f48e69e4da66d4e,
+		0x8afd13cf8e6fb054, 0xf5e1c9011d5ed849, 0xe34e091c5126c8af, 0xad67ee7530a398f6, 0x43b24dec2e82c75a, 0x75da99c1287cd48d, 0x92e81cdb3783f689, 0xa3dd217cc537cecd,
+		0x60543c50de970553, 0x93f73f54aaf2426a, 0xa91b62737e7a725d, 0xf19d4507538732e2, 0x77e4dfc20f9ea156, 0x7d229ccdb4d31dc6, 0x1b346a98037f87e5, 0xedf4c615a4b29e94,
+		0x4093286094110662, 0xb0114ee85ae78063, 0x6ff1d0d6b672e78b, 0x6dcf96d591909250, 0xdfe09e3eec9567e8, 0x3214582b4827f97c, 0xb46dc2ee143e6ac8, 0xf6c0ac8da7cd1971,
+		0xebb60c10cd8901e4, 0xf7df8f023abcad92, 0x9c52d3d2c217a0b2, 0x6b8d5cd0f8ab0d20, 0x3777f7a29b8fa734, 0x011f238f9d71b4e3, 0xc1b75b2f3c42be45, 0x5de588fdfe551ef7,
+		0x6eeef3592b035368, 0xaa3a07ffc4e9b365, 0xecebe59a39c32a77, 0x5ba742f8976e8187, 0x4b4a48e0b22d0e11, 0xddded83dcb771233, 0xa59feb79ac0c51bd, 0xc7f5912a55792135
+	},
+	{
+		0x6d6ae04668a9b08a, 0x3ab3f04b0be8c743, 0xe51e166b54b3c908, 0xbe90a9eb35c2f139, 0xb2c7066637f2bec1, 0xaa6945613392202c, 0x9a28c36f3b5201eb, 0xddce5a93ab536994,
+		0x0e34133ef6382827, 0x52a02ba1ec55048b, 0xa2f88f97c4b2a177, 0x8640e513ca2251a5, 0xcdf1d36258137622, 0xfe6cb708dedf8ddb, 0x8a174a9ec8121e5d, 0x679896036b81560e,
+		0x59ed033395795fee, 0x1dd778ab8b74edaf, 0xee533ef92d9f926d, 0x2a8c79baf8a8d8f5, 0x6bcf398e69b119f6, 0xe20491742fafdd95, 0x276488e0809c2aec, 0xea955b82d88f5cce,
+		0x7102c63a99d9e0c4, 0xf9763017a5c39946, 0x429fa2501f151b3d, 0x4659c72bea05d59e, 0x984b7fdccf5a6634, 0xf742232953fbb161, 0x3041860e08c021c7, 0x747bfd9616cd9386,
+		0x4bb1367192312787, 0x1b72a1638a6c44d3, 0x4a0e68a6e8359a66, 0x169a5039f258b6ca, 0xb98a2ef44edee5a4, 0xd9083fe85e43a737, 0x967f6ce239624e13, 0x8874f62d3c1a7982,
+		0x3c1629830af06e3f, 0x9165ebfd427e5a8e, 0xb5dd81794ceeaa5c, 0x0de8f15a7834f219, 0x70bd98ede3dd5d25, 0xaccc9ca9328a8950, 0x56664eda1945ca28, 0x221db34c0f8859ae,
+		0x26dbd637fa98970d, 0x1acdffb4f068f932, 0x4585254f64090fa0, 0x72de245e17d53afa, 0x1546b25d7c546cf4, 0x207e0ffffb803e71, 0xfaaad2732bcf4378, 0xb462dfae36ea17bd,
+		0xcf926fd1ac1b11fd, 0xe0672dc7dba7ba4a, 0xd3fa49ad5d6b41b3, 0x8ba81449b216a3bc, 0x14f9ec8a0650d115, 0x40fc1ee3eb1d7ce2, 0x23a2ed9b758ce44f, 0x782c521b14fddc7e,
+		0x1c68267cf170504e, 0xbcf31558c1ca96e6, 0xa781b43b4ba6d235, 0xf6fd7dfe29ff0c80, 0xb0a4bad5c3fad91e, 0xd199f51ea963266c, 0x414340349119c103, 0x5405f269ed4dadf7,
+		0xabd61bb649969dcd, 0x6813dbeae7bdc3c8, 0x65fb2ab09f8931d1, 0xf1e7fae152e3181d, 0xc1a67cef5a2339da, 0x7a4feea8e0f5bba1, 0x1e0b9acf05783791, 0x5b8ebf8061713831,
+		0x80e53cdbcb3af8d9, 0x7e898bd315e57502, 0xc6bcfbf0213f2d47, 0x95a38e86b76e942d, 0x092e94218d243cba, 0x8339debf453622e7, 0xb11be402b9fe64ff, 0x57d9100d634177c9,
+		0xcc4e8db52217cbc3, 0x3b0cae9c71ec7aa2, 0xfb158ca451cbfe99, 0x2b33276d82ac6514, 0x01bf5ed77a04bde1, 0xc5601994af33f779, 0x75c4a3416cc92e67, 0xf3844652a6eb7fc2,
+		0x3487e375fdd0ef64, 0x18ae430704609eed, 0x4d14efb993298efb, 0x815a620cb13e4538, 0x125c354207487869, 0x9eeea614ce42cf48, 0xce2d3106d61fac1c, 0xbbe99247bad6827b,
+		0x071a871f7b1c149d, 0x2e4a1cc10db81656, 0x77a71ff298c149b8, 0x06a5d9c80118a97c, 0xad73c27e488e34b1, 0x443a7b981e0db241, 0xe3bbcfa355ab6074, 0x0af276450328e684,
+		0x73617a896dd1871b, 0x58525de4ef7de20f, 0xb7be3dcab8e6cd83, 0x19111dd07e64230c, 0x842359a03e2a367a, 0x103f89f1f3401fb6, 0xdc710444d157d475, 0xb835702334da5845,
+		0x4320fc876511a6dc, 0xd026abc9d3679b8d, 0x17250eee885c0b2b, 0x90dab52a387ae76f, 0x31fed8d972c49c26, 0x89cba8fa461ec463, 0x2ff5421677bcabb7, 0x396f122f85e41d7d,
+		0xa09b332430bac6a8, 0xc888e8ced7070560, 0xaeaf201ac682ee8f, 0x1180d7268944a257, 0xf058a43628e7a5fc, 0xbd4c4b8fbbce2b07, 0xa1246df34abe7b49, 0x7d5569b79be9af3c,
+		0xa9b5a705bd9efa12, 0xdb6b835baa4bc0e8, 0x05793bac8f147342, 0x21c1512881848390, 0xfdb0556c50d357e5, 0x613d4fcb6a99ff72, 0x03dce2648e0cda3e, 0xe949b9e6568386f0,
+		0xfc0f0bbb2ad7ea04, 0x6a70675913b5a417, 0x7f36d5046fe1c8e3, 0x0c57af8d02304ff8, 0x32223abdfcc84618, 0x0891caf6f720815b, 0xa63eeaec31a26fd4, 0x2507345374944d33,
+		0x49d28ac266394058, 0xf5219f9aa7f3d6be, 0x2d96fea583b4cc68, 0x5a31e1571b7585d0, 0x8ed12fe53d02d0fe, 0xdfade6205f5b0e4b, 0x4cabb16ee92d331a, 0x04c6657bf510cea3,
+		0xd73c2cd6a87b8f10, 0xe1d87310a1a307ab, 0x6cd5be9112ad0d6b, 0x97c032354366f3f2, 0xd4e0ceb22677552e, 0x0000000000000000, 0x29509bde76a402cb, 0xc27a9e8bd42fe3e4,
+		0x5ef7842cee654b73, 0xaf107ecdbc86536e, 0x3fcacbe784fcb401, 0xd55f90655c73e8cf, 0xe6c2f40fdabf1336, 0xe8f6e7312c873b11, 0xeb2a0555a28be12f, 0xe4a148bc2eb774e9,
+		0x9b979db84156bc0a, 0x6eb60222e6a56ab4, 0x87ffbbc4b026ec44, 0xc703a5275b3b90a6, 0x47e699fc9001687f, 0x9c8d1aa73a4aa897, 0x7cea3760e1ed12dd, 0x4ec80ddd1d2554c5,
+		0x13e36b957d4cc588, 0x5d2b66486069914d, 0x92b90999cc7280b0, 0x517cc9c56259deb5, 0xc937b619ad03b881, 0xec30824ad997f5b2, 0xa45d565fc5aa080b, 0xd6837201d27f32f1,
+		0x635ef3789e9198ad, 0x531f75769651b96a, 0x4f77530a6721e924, 0x486dd4151c3dfdb9, 0x5f48dafb9461f692, 0x375b011173dc355a, 0x3da9775470f4d3de, 0x8d0dcd81b30e0ac0,
+		0x36e45fc609d888bb, 0x55baacbe97491016, 0x8cb29356c90ab721, 0x76184125e2c5f459, 0x99f4210bb55edbd5, 0x6f095cf59ca1d755, 0x9f51f8c3b44672a9, 0x3538bda287d45285,
+		0x50c39712185d6354, 0xf23b1885dcefc223, 0x79930ccc6ef9619f, 0xed8fdc9da3934853, 0xcb540aaa590bdf5e, 0x5c94389f1a6d2cac, 0xe77daad8a0bbaed7, 0x28efc5090ca0bf2a,
+		0xbf2ff73c4fc64cd8, 0xb37858b14df60320, 0xf8c96ec0dfc724a7, 0x828680683f329f06, 0x941cd051cd6a29cc, 0xc3c5c05cae2b5e05, 0xb601631dc2e27062, 0xc01922382027843b,
+		0x24b86a840e90f0d2, 0xd245177a276ffc52, 0x0f8b4de98c3c95c6, 0x3e759530fef809e0, 0x0b4d2892792c5b65, 0xc4df4743d5374a98, 0xa5e20888bfaeb5ea, 0xba56cc90c0d23f9a,
+		0x38d04cf8ffe0a09c, 0x62e1adafe495254c, 0x0263bcb3f40867df, 0xcaeb547d230f62bf, 0x6082111c109d4293, 0xdad4dd8cd04f7d09, 0xefec602e579b2f8c, 0x1fb4c4187f7c8a70,
+		0xffd3e9dfa4db303a, 0x7bf0b07f9af10640, 0xf49ec14dddf76b5f, 0x8f6e713247066d1f, 0x339d646a86ccfbf9, 0x64447467e58d8c30, 0x2c29a072f9b07189, 0xd8b7613f24471ad6,
+		0x6627c8d41185ebef, 0xa347d140beb61c96, 0xde12b8f7255fb3aa, 0x9d324470404e1576, 0x9306574eb6763d51, 0xa80af9d2c79a47f3, 0x859c0777442e8b9b, 0x69ac853d9db97e29
+	},
+	{
+		0xc3407dfc2de6377e, 0x5b9e93eea4256f77, 0xadb58fdd50c845e0, 0x5219ff11a75bed86, 0x356b61cfd90b1de9, 0xfb8f406e25abe037, 0x7a5a0231c0f60796, 0x9d3cd216e1f5020b,
+		0x0c6550fb6b48d8f3, 0xf57508c427ff1c62, 0x4ad35ffa71cb407d, 0x6290a2da1666aa6d, 0xe284ec2349355f9f, 0xb3c307c53d7c84ec, 0x05e23c0468365a02, 0x190bac4d6c9ebfa8,
+		0x94bbbee9e28b80fa, 0xa34fc777529cb9b5, 0xcc7b39f095bcd978, 0x2426addb0ce532e3, 0x7e79329312ce4fc7, 0xab09a72eebec2917, 0xf8d15499f6b9d6c2, 0x1a55b8babf8c895d,
+		0xdb8add17fb769a85, 0xb57f2f368658e81b, 0x8acd36f18f3f41f6, 0x5ce3b7bba50f11d3, 0x114dcc14d5ee2f0a, 0xb91a7fcded1030e8, 0x81d5425fe55de7a1, 0xb6213bc1554adeee,
+		0x80144ef95f53f5f2, 0x1e7688186db4c10c, 0x3b912965db5fe1bc, 0xc281715a97e8252d, 0x54a5d7e21c7f8171, 0x4b12535ccbc5522e, 0x1d289cefbea6f7f9, 0x6ef5f2217d2e729e,
+		0xe6a7dc819b0d17ce, 0x1b94b41c05829b0e, 0x33d7493c622f711e, 0xdcf7f942fa5ce421, 0x600fba8b7f7a8ecb, 0x46b60f011a83988e, 0x235b898e0dcf4c47, 0x957ab24f588592a9,
+		0x4354330572b5c28c, 0xa5f3ef84e9b8d542, 0x8c711e02341b2d01, 0x0b1874ae6a62a657, 0x1213d8e306fc19ff, 0xfe6d7c6a4d9dba35, 0x65ed868f174cd4c9, 0x88522ea0e6236550,
+		0x899322065c2d7703, 0xc01e690bfef4018b, 0x915982ed8abddaf8, 0xbe675b98ec3a4e4c, 0xa996bf7f82f00db1, 0xe1daf8d49a27696a, 0x2effd5d3dc8986e7, 0xd153a51f2b1a2e81,
+		0x18caa0ebd690adfb, 0x390e3134b243c51a, 0x2778b92cdff70416, 0x029f1851691c24a6, 0x5e7cafeacc133575, 0xfa4e4cc89fa5f264, 0x5a5f9f481e2b7d24, 0x484c47ab18d764db,
+		0x400a27f2a1a7f479, 0xaeeb9b2a83da7315, 0x721c626879869734, 0x042330a2d2384851, 0x85f672fd3765aff0, 0xba446b3a3e02061d, 0x73dd6ecec3888567, 0xffac70ccf793a866,
+		0xdfa9edb5294ed2d4, 0x6c6aea7014325638, 0x834a5a0e8c41c307, 0xcdba35562fb2cb2b, 0x0ad97808d06cb404, 0x0f3b440cb85aee06, 0xe5f9c876481f213b, 0x98deee1289c35809,
+		0x59018bbfcd394bd1, 0xe01bf47220297b39, 0xde68e1139340c087, 0x9fa3ca4788e926ad, 0xbb85679c840c144e, 0x53d8f3b71d55ffd5, 0x0da45c5dd146caa0, 0x6f34fe87c72060cd,
+		0x57fbc315cf6db784, 0xcee421a1fca0fdde, 0x3d2d0196607b8d4b, 0x642c8a29ad42c69a, 0x14aff010bdd87508, 0xac74837beac657b3, 0x3216459ad821634d, 0x3fb219c70967a9ed,
+		0x06bc28f3bb246cf7, 0xf2082c9126d562c6, 0x66b39278c45ee23c, 0xbd394f6f3f2878b9, 0xfd33689d9e8f8cc0, 0x37f4799eb017394f, 0x108cc0b26fe03d59, 0xda4bd1b1417888d6,
+		0xb09d1332ee6eb219, 0x2f3ed975668794b4, 0x58c0871977375982, 0x7561463d78ace990, 0x09876cff037e82f1, 0x7fb83e35a8c05d94, 0x26b9b58a65f91645, 0xef20b07e9873953f,
+		0x3148516d0b3355b8, 0x41cb2b541ba9e62a, 0x790416c613e43163, 0xa011d380818e8f40, 0x3a5025c36151f3ef, 0xd57095bdf92266d0, 0x498d4b0da2d97688, 0x8b0c3a57353153a5,
+		0x21c491df64d368e1, 0x8f2f0af5e7091bf4, 0x2da1c1240f9bb012, 0xc43d59a92ccc49da, 0xbfa6573e56345c1f, 0x828b56a8364fd154, 0x9a41f643e0df7caf, 0xbcf843c985266aea,
+		0x2b1de9d7b4bfdce5, 0x20059d79dedd7ab2, 0x6dabe6d6ae3c446b, 0x45e81bf6c991ae7b, 0x6351ae7cac68b83e, 0xa432e32253b6c711, 0xd092a9b991143cd2, 0xcac711032e98b58f,
+		0xd8d4c9e02864ac70, 0xc5fc550f96c25b89, 0xd7ef8dec903e4276, 0x67729ede7e50f06f, 0xeac28c7af045cf3d, 0xb15c1f945460a04a, 0x9cfddeb05bfb1058, 0x93c69abce3a1fe5e,
+		0xeb0380dc4a4bdd6e, 0xd20db1e8f8081874, 0x229a8528b7c15e14, 0x44291750739fbc28, 0xd3ccbd4e42060a27, 0xf62b1c33f4ed2a97, 0x86a8660ae4779905, 0xd62e814a2a305025,
+		0x477703a7a08d8add, 0x7b9b0e977af815c5, 0x78c51a60a9ea2330, 0xa6adfb733aaae3b7, 0x97e5aa1e3199b60f, 0x0000000000000000, 0xf4b404629df10e31, 0x5564db44a6719322,
+		0x9207961a59afec0d, 0x9624a6b88b97a45c, 0x363575380a192b1c, 0x2c60cd82b595a241, 0x7d272664c1dc7932, 0x7142769faa94a1c1, 0xa1d0df263b809d13, 0x1630e841d4c451ae,
+		0xc1df65ad44fa13d8, 0x13d2d445bcf20bac, 0xd915c546926abe23, 0x38cf3d92084dd749, 0xe766d0272103059d, 0xc7634d5effde7f2f, 0x077d2455012a7ea4, 0xedbfa82ff16fb199,
+		0xaf2a978c39d46146, 0x42953fa3c8bbd0df, 0xcb061da59496a7dc, 0x25e7a17db6eb20b0, 0x34aa6d6963050fba, 0xa76cf7d580a4f1e4, 0xf7ea10954ee338c4, 0xfcf2643b24819e93,
+		0xcf252d0746aeef8d, 0x4ef06f58a3f3082c, 0x563acfb37563a5d7, 0x5086e740ce47c920, 0x2982f186dda3f843, 0x87696aac5e798b56, 0x5d22bb1d1f010380, 0x035e14f7d31236f5,
+		0x3cec0d30da759f18, 0xf3c920379cdb7095, 0xb8db736b571e22bb, 0xdd36f5e44052f672, 0xaac8ab8851e23b44, 0xa857b3d938fe1fe2, 0x17f1e4e76eca43fd, 0xec7ea4894b61a3ca,
+		0x9e62c6e132e734fe, 0xd4b1991b432c7483, 0x6ad6c283af163acf, 0x1ce9904904a8e5aa, 0x5fbda34c761d2726, 0xf910583f4cb7c491, 0xc6a241f845d06d7c, 0x4f3163fe19fd1a7f,
+		0xe99c988d2357f9c8, 0x8eee06535d0709a7, 0x0efa48aa0254fc55, 0xb4be23903c56fa48, 0x763f52caabbedf65, 0xeee1bcd8227d876c, 0xe345e085f33b4dcc, 0x3e731561b369bbbe,
+		0x2843fd2067adea10, 0x2adce5710eb1ceb6, 0xb7e03767ef44ccbd, 0x8db012a48e153f52, 0x61ceb62dc5749c98, 0xe85d942b9959eb9b, 0x4c6f7709caef2c8a, 0x84377e5b8d6bbda3,
+		0x30895dcbb13d47eb, 0x74a04a9bc2a2fbc3, 0x6b17ce251518289c, 0xe438c4d0f2113368, 0x1fb784bed7bad35f, 0x9b80fae55ad16efc, 0x77fe5e6c11b0cd36, 0xc858095247849129,
+		0x08466059b97090a2, 0x01c10ca6ba0e1253, 0x6988d6747c040c3a, 0x6849dad2c60a1e69, 0x5147ebe67449db73, 0xc99905f4fd8a837a, 0x991fe2b433cd4a5a, 0xf09734c04fc94660,
+		0xa28ecbd1e892abe6, 0xf1563866f5c75433, 0x4dae7baf70e13ed9, 0x7ce62ac27bd26b61, 0x70837a39109ab392, 0x90988e4b30b3c8ab, 0xb2020b63877296bf, 0x156efcb607d6675b
+	},
+	{
+		0xe63f55ce97c331d0, 0x25b506b0015bba16, 0xc8706e29e6ad9ba8, 0x5b43d3775d521f6a, 0x0bfa3d577035106e, 0xab95fc172afb0e66, 0xf64b63979e7a3276, 0xf58b4562649dad4b,
+		0x48f7c3dbae0c83f1, 0xff31916642f5c8c5, 0xcbb048dc1c4a0495, 0x66b8f83cdf622989, 0x35c130e908e2b9b0, 0x7c761a61f0b34fa1, 0x3601161cf205268d, 0x9e54ccfe2219b7d6,
+		0x8b7d90a538940837, 0x9cd403588ea35d0b, 0xbc3c6fea9ccc5b5a, 0xe5ff733b6d24aeed, 0xceed22de0f7eb8d2, 0xec8581cab1ab545e, 0xb96105e88ff8e71d, 0x8ca03501871a5ead,
+		0x76ccce65d6db2a2f, 0x5883f582a7b58057, 0x3f7be4ed2e8adc3e, 0x0fe7be06355cd9c9, 0xee054e6c1d11be83, 0x1074365909b903a6, 0x5dde9f80b4813c10, 0x4a770c7d02b6692c,
+		0x5379c8d5d7809039, 0xb4067448161ed409, 0x5f5e5026183bd6cd, 0xe898029bf4c29df9, 0x7fb63c940a54d09c, 0xc5171f897f4ba8bc, 0xa6f28db7b31d3d72, 0x2e4f3be7716eaa78,
+		0x0d6771a099e63314, 0x82076254e41bf284, 0x2f0fd2b42733df98, 0x5c9e76d3e2dc49f0, 0x7aeb569619606cdb, 0x83478b07b2468764, 0xcfadcb8d5923cd32, 0x85dac7f05b95a41e,
+		0xb5469d1b4043a1e9, 0xb821ecbbd9a592fd, 0x1b8e0b0e798c13c8, 0x62a57b6d9a0be02e, 0xfcf1b793b81257f8, 0x9d94ea0bd8fe28eb, 0x4cea408aeb654a56, 0x23284a47e888996c,
+		0x2d8f1d128b893545, 0xf4cbac3132c0d8ab, 0xbd7c86b9ca912eba, 0x3a268eef3dbe6079, 0xf0d62f6077a9110c, 0x2735c916ade150cb, 0x89fd5f03942ee2ea, 0x1acee25d2fd16628,
+		0x90f39bab41181bff, 0x430dfe8cde39939f, 0xf70b8ac4c8274796, 0x1c53aeaac6024552, 0x13b410acf35e9c9b, 0xa532ab4249faa24f, 0x2b1251e5625a163f, 0xd7e3e676da4841c7,
+		0xa7b264e4e5404892, 0xda8497d643ae72d3, 0x861ae105a1723b23, 0x38a6414991048aa4, 0x6578dec92585b6b4, 0x0280cfa6acbaeadd, 0x88bdb650c273970a, 0x9333bd5ebbff84c2,
+		0x4e6a8f2c47dfa08b, 0x321c954db76cef2a, 0x418d312a72837942, 0xb29b38bfffcdf773, 0x6c022c38f90a4c07, 0x5a033a240b0f6a8a, 0x1f93885f3ce5da6f, 0xc38a537e96988bc6,
+		0x39e6a81ac759ff44, 0x29929e43cee0fce2, 0x40cdd87924de0ca2, 0xe9d8ebc8a29fe819, 0x0c2798f3cfbb46f4, 0x55e484223e53b343, 0x4650948ecd0d2fd8, 0x20e86cb2126f0651,
+		0x6d42c56baf5739e7, 0xa06fc1405ace1e08, 0x7babbfc54f3d193b, 0x424d17df8864e67f, 0xd8045870ef14980e, 0xc6d7397c85ac3781, 0x21a885e1443273b1, 0x67f8116f893f5c69,
+		0x24f5efe35706cff6, 0xd56329d076f2ab1a, 0x5e1eb9754e66a32d, 0x28d2771098bd8902, 0x8f6013f47dfdc190, 0x17a993fdb637553c, 0xe0a219397e1012aa, 0x786b9930b5da8606,
+		0x6e82e39e55b0a6da, 0x875a0856f72f4ec3, 0x3741ff4fa458536d, 0xac4859b3957558fc, 0x7ef6d5c75c09a57c, 0xc04a758b6c7f14fb, 0xf9acdd91ab26ebbf, 0x7391a467c5ef9668,
+		0x335c7c1ee1319aca, 0xa91533b18641e4bb, 0xe4bf9a683b79db0d, 0x8e20faa72ba0b470, 0x51f907737b3a7ae4, 0x2268a314bed5ec8c, 0xd944b123b949edee, 0x31dcb3b84d8b7017,
+		0xd3fe65279f218860, 0x097af2f1dc8ffab3, 0x9b09a6fc312d0b91, 0xcc6ded78a3c4520f, 0x3481d9ba5ebfcc50, 0x4f2a667f1182d56b, 0xdfd9fdd4509ace94, 0x26752045fbbc252b,
+		0xbffc491f662bc467, 0xdd593272fc202449, 0x3cbbc218d46d4303, 0x91b372f817456e1f, 0x681faf69bc6385a0, 0xb686bbeebaa43ed4, 0x1469b5084cd0ca01, 0x98c98009cbca94ac,
+		0x6438379a73d8c354, 0xc2caba2dc0c5fe26, 0x3e3b0dbe78d7a9de, 0x50b9ee202d670f04, 0x4590b27b37eab0e5, 0x6025b4cb36b10af3, 0xfb2c1237079c0162, 0xa12f28130c936be8,
+		0x4b37e52e54eb1ccc, 0x083a1ba28ad28f53, 0xc10a9cd83a22611b, 0x9f1425ad7444c236, 0x069d4cf7e9d3237a, 0xedc56899e7f621be, 0x778c273680865fcf, 0x309c5aeb1bd605f7,
+		0x8de0dc52d1472b4d, 0xf8ec34c2fd7b9e5f, 0xea18cd3d58787724, 0xaad515447ca67b86, 0x9989695a9d97e14c, 0x0000000000000000, 0xf196c63321f464ec, 0x71116bc169557cb5,
+		0xaf887f466f92c7c1, 0x972e3e0ffe964d65, 0x190ec4a8d536f915, 0x95aef1a9522ca7b8, 0xdc19db21aa7d51a9, 0x94ee18fa0471d258, 0x8087adf248a11859, 0xc457f6da2916dd5c,
+		0xfa6cfb6451c17482, 0xf256e0c6db13fbd1, 0x6a9f60cf10d96f7d, 0x4daaa9d9bd383fb6, 0x03c026f5fae79f3d, 0xde99148706c7bb74, 0x2a52b8b6340763df, 0x6fc20acd03edd33a,
+		0xd423c08320afdefa, 0xbbe1ca4e23420dc0, 0x966ed75ca8cb3885, 0xeb58246e0e2502c4, 0x055d6a021334bc47, 0xa47242111fa7d7af, 0xe3623fcc84f78d97, 0x81c744a11efc6db9,
+		0xaec8961539cfb221, 0xf31609958d4e8e31, 0x63e5923ecc5695ce, 0x47107ddd9b505a38, 0xa3afe7b5a0298135, 0x792b7063e387f3e6, 0x0140e953565d75e0, 0x12f4f9ffa503e97b,
+		0x750ce8902c3cb512, 0xdbc47e8515f30733, 0x1ed3610c6ab8af8f, 0x5239218681dde5d9, 0xe222d69fd2aaf877, 0xfe71783514a8bd25, 0xcaf0a18f4a177175, 0x61655d9860ec7f13,
+		0xe77fbc9dc19e4430, 0x2ccff441ddd440a5, 0x16e97aaee06a20dc, 0xa855dae2d01c915b, 0x1d1347f9905f30b2, 0xb7c652bdecf94b34, 0xd03e43d265c6175d, 0xfdb15ec0ee4f2218,
+		0x57644b8492e9599e, 0x07dda5a4bf8e569a, 0x54a46d71680ec6a3, 0x5624a2d7c4b42c7e, 0xbebca04c3076b187, 0x7d36f332a6ee3a41, 0x3b6667bc6be31599, 0x695f463aea3ef040,
+		0xad08b0e0c3282d1c, 0xb15b1e4a052a684e, 0x44d05b2861b7c505, 0x15295c5b1a8dbfe1, 0x744c01c37a61c0f2, 0x59c31cd1f1e8f5b7, 0xef45a73f4b4ccb63, 0x6bdf899c46841a9d,
+		0x3dfb2b4b823036e3, 0xa2ef0ee6f674f4d5, 0x184e2dfb836b8cf5, 0x1134df0a5fe47646, 0xbaa1231d751f7820, 0xd17eaa81339b62bd, 0xb01bf71953771dae, 0x849a2ea30dc8d1fe,
+		0x705182923f080955, 0x0ea757556301ac29, 0x041d83514569c9a7, 0x0abad4042668658e, 0x49b72a88f851f611, 0x8a3d79f66ec97dd7, 0xcd2d042bf59927ef, 0xc930877ab0f0ee48,
+		0x9273540deda2f122, 0xc797d02fd3f14261, 0xe1e2f06a284d674a, 0xd2be8c74c97cfd80, 0x9a494faf67707e71, 0xb3dbd1eca9908293, 0x72d14d3493b2e388, 0xd6a30f258c153427
+	}
+};

lib/std/hash/hmac.c3

@@ -1,4 +1,4 @@
-module std::hash::hmac{HashAlg, HASH_BYTES, BLOCK_BYTES};
+module std::hash::hmac <HashAlg, HASH_BYTES, BLOCK_BYTES>;
 import std::crypto;
 
 struct Hmac

lib/std/hash/komi.c3

@@ -58,38 +58,38 @@ fn ulong hash(char[] data, ulong seed = 0)
 
 		if (@likely(data.len >= 8))
 		{
-			r1h ^= @unaligned_load(*(ulong*)data.ptr, 1);
+			r1h ^= mem::load((ulong*)data.ptr, 1);
 
 			r2h ^= (data.len < 12)
 				? ((data[data.len - 3] | ((ulong)data[data.len - 2] << 8) | ((ulong)data[data.len - 1] << 16) | ((ulong)1 << 24)) >> ((data.len * 8) ^ 88))
-				: (((@unaligned_load(*(uint*)&data[^4], 1) | ((ulong)1 << 32)) >> (128 - data.len * 8)) << 32 | @unaligned_load(*(uint*)&data[8], 1));
+				: (((mem::load((uint*)&data[^4], 1) | ((ulong)1 << 32)) >> (128 - data.len * 8)) << 32 | mem::load((uint*)&data[8], 1));
 		}
 		else if (data.len != 0)
 		{
 			r1h ^= (data.len < 4)
 				? (((ulong)1 << (data.len * 8)) ^ data[0] ^ (data.len > 1 ? (ulong)data[1] << 8 : 0) ^ (data.len > 2 ? (ulong)data[2] << 16 : 0))
-				: (((@unaligned_load(*(uint*)&data[^4], 1) | ((ulong)1 << 32)) >> (64 - data.len * 8)) << 32 | @unaligned_load(*(uint*)&data[0], 1));
+				: (((mem::load((uint*)&data[^4], 1) | ((ulong)1 << 32)) >> (64 - data.len * 8)) << 32 | mem::load((uint*)&data[0], 1));
 		}
 	}
 	else if (data.len < 32)
 	{
 		// HASH16
 		@komimul(
-			@unaligned_load(*(ulong*)&data[0], 1) ^ seed1,
-			@unaligned_load(*(ulong*)&data[8], 1) ^ seed5,
+			mem::load((ulong*)&data[0], 1) ^ seed1,
+			mem::load((ulong*)&data[8], 1) ^ seed5,
 			seed1, seed5
 		);
 		seed1 ^= seed5;
 
 		if (data.len < 24)
 		{
-			r1h = (((@unaligned_load(*(ulong*)&data[^8], 1) >> 8) | ((ulong)1 << 56)) >> (((int)(data.len * 8) ^ 184))) ^ seed1;
+			r1h = (((mem::load((ulong*)&data[^8], 1) >> 8) | ((ulong)1 << 56)) >> (((int)(data.len * 8) ^ 184))) ^ seed1;
 			r2h = seed5;
 		}
 		else
 		{
-			r1h = @unaligned_load(*(ulong*)&data[16], 1) ^ seed1;
-			r2h = (((@unaligned_load(*(ulong*)&data[^8], 1) >> 8) | ((ulong)1 << 56)) >> (((int)(data.len * 8) ^ 248))) ^ seed5;
+			r1h = mem::load((ulong*)&data[16], 1) ^ seed1;
+			r2h = (((mem::load((ulong*)&data[^8], 1) >> 8) | ((ulong)1 << 56)) >> (((int)(data.len * 8) ^ 248))) ^ seed5;
 		}
 	}
 	else
@@ -106,8 +106,8 @@ fn ulong hash(char[] data, ulong seed = 0)
 			{
 				$for var $x = 0; $x < 4; ++$x :
 					@komimul(
-						@unaligned_load(*(ulong*)&data[0 + ($x * 8)], 1) ^ seeds[$x],
-						@unaligned_load(*(ulong*)&data[32 + ($x * 8)], 1) ^ seeds[4 + $x],
+						mem::load((ulong*)&data[0 + ($x * 8)], 1) ^ seeds[$x],
+						mem::load((ulong*)&data[32 + ($x * 8)], 1) ^ seeds[4 + $x],
 						seeds[$x], seeds[4 + $x]
 					);
 				$endfor
@@ -125,8 +125,8 @@ fn ulong hash(char[] data, ulong seed = 0)
 		for (; data.len >= 16; data = data[16:^16])
 		{
 			@komimul(
-				@unaligned_load(*(ulong*)&data[0], 1) ^ seed1,
-				@unaligned_load(*(ulong*)&data[8], 1) ^ seed5,
+				mem::load((ulong*)&data[0], 1) ^ seed1,
+				mem::load((ulong*)&data[8], 1) ^ seed5,
 				seed1, seed5
 			);
 			seed1 ^= seed5;
@@ -137,13 +137,13 @@ fn ulong hash(char[] data, ulong seed = 0)
 			// NOTE: This is translated from the original code. It grabs the last ulong off the buffer even though the
 			//        data slice is less than 8 bytes. This is possible because this branch only occurs in a loop where
 			//        the original data slice length is >= 32.
-			r1h = (((@unaligned_load(*(ulong*)(data.ptr + data.len - 8), 1) >> 8) | ((ulong)1 << 56)) >> ((data.len * 8) ^ 0x38)) ^ seed1;
+			r1h = (((mem::load((ulong*)(data.ptr + data.len - 8), 1) >> 8) | ((ulong)1 << 56)) >> ((data.len * 8) ^ 0x38)) ^ seed1;
 			r2h = seed5;
 		}
 		else
 		{
-			r1h = @unaligned_load(*(ulong*)data.ptr, 1) ^ seed1;
-			r2h = (((@unaligned_load(*(ulong*)&data[^8], 1) >> 8) | ((ulong)1 << 56)) >> ((data.len * 8) ^ 0x78)) ^ seed5;
+			r1h = mem::load((ulong*)data.ptr, 1) ^ seed1;
+			r2h = (((mem::load((ulong*)&data[^8], 1) >> 8) | ((ulong)1 << 56)) >> ((data.len * 8) ^ 0x78)) ^ seed5;
 		}
 	}
 

lib/std/hash/md5.c3

@@ -108,7 +108,7 @@ macro @i(x, y, z) => y ^ (x | ~z);
 
 macro void @step(#f, a, b, c, d, ptr, n, t, s)
 {
-	*a += #f(b, c, d) + @unaligned_load(*(uint *)&ptr[n * 4], 2) + t;
+	*a += #f(b, c, d) + mem::load((uint *)&ptr[n * 4], 2) + t;
 	*a = (*a << s) | ((*a & 0xffffffff) >> (32 - s));
 	*a += b;
 }

lib/std/hash/poly1305.c3

@@ -0,0 +1,171 @@
+// Copyright (c) 2025 Zack Puhl <github@xmit.xyz>. 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.
+//
+// Poly1305 code dedicated from repo: https://github.com/NotsoanoNimus/chacha20_aead.c3l (but massively cleaned)
+module std::hash::poly1305;
+
+
+<* The fixed output length of Poly1305 tags (hashes).  *>
+const TAG_SIZE = 16;
+
+<* Fixed length of a Poly1305 block. *>
+const BLOCK_SIZE = 16;
+
+<* Fixed length of the required Poly1305 key, *>
+const KEY_SIZE = 32;
+
+
+struct Poly1305
+{
+    ulong[3]        h; // hash internal state
+    uint128         r; // secret portion of key
+    uint128         nonce; // initialization vector, derived from key
+    char[TAG_SIZE]  temp; // last partial ingestion state
+    usz             num; // index into last partial ingestion
+    // Additional, cached state information:
+    ulong           r0;
+    ulong           r1;
+    ulong           s1;
+}
+
+
+<* Constant-time carrying computation for block permutations. *>
+macro ulong constant_time_carry(ulong a, ulong b) @local
+{
+    return (a ^ ((a ^ b) | ((a - b) ^ b))) >> (bitsizeof(ulong) - 1);
+}
+
+
+<*
+ Compute a Poly1305 message authentication code for the input with the given key (secret + nonce) value.
+ Note that this construct SHOULD NOT be used with the 'Hmac' module; it is its own MAC package.
+
+ @param[in] input
+ @param key
+*>
+fn char[TAG_SIZE] hash(char[] input, char[KEY_SIZE] key)
+{
+    Poly1305 p @noinit;
+    p.init(key);
+    p.update(input);
+    return p.final();
+}
+
+<* Alias for the `hash` function; Message Authentication Code. *>
+alias mac = hash;
+
+<* Alias for the `hash` function; "tag" generation. *>
+alias tag = hash;
+
+
+fn void Poly1305.init(&self, char[KEY_SIZE] key)
+{
+    *self = { // implicitly clears state as well
+        .r      = mem::load((uint128*)&key[ 0], 1) & 0x0ffffffc_0ffffffc_0ffffffc_0fffffff, // clamped per spec
+        .nonce  = mem::load((uint128*)&key[16], 1)
+    };
+
+    self.r0 = @unaligned_load(((ulong*)&self.r)[0], 1);
+    self.r1 = @unaligned_load(((ulong*)&self.r)[1], 1);
+
+    self.s1 = self.r1 + (self.r1 >> 2);
+}
+
+
+fn void Poly1305.update(&self, char[] input)
+{
+    if (self.num) // currently between consuming full blocks?
+    {
+        usz rem = BLOCK_SIZE - self.num;
+        if (input.len < rem)
+        {
+            self.temp[self.num:input.len] = input[..]; // saving another partial block
+            self.num += input.len; // move index forward
+            return;
+        }
+        // ingest up to a block size to finish the partial, then advance the slice ptr
+        self.temp[self.num:rem] = input[:rem];
+        self.blocks(self.temp[..]);
+        input = input[rem..];
+    }
+
+    usz even_length = input.len - (input.len % BLOCK_SIZE);
+    if (even_length >= BLOCK_SIZE)
+    {
+        self.blocks(input[:even_length]); // consume blocks
+        input = input[even_length..]; // scroll to end (remainder)
+    }
+
+    if (input.len) self.temp[:input.len] = input[..]; // keep remainder (uneven block sizes)
+
+    self.num = input.len;
+}
+
+
+fn char[TAG_SIZE] Poly1305.final(&self)
+{
+    if (self.num) // consume any leftovers
+    {
+        self.temp[self.num++] = 1; // partial blocks must end with 0x01
+        self.temp[self.num..] = {}; // explicit zeros on the rest
+        self.blocks(self.temp[..], 0); // chomp
+    }
+
+    uint128 t = (uint128)self.h[0] + 5;
+    ulong g0 = (ulong)t;
+
+    t = (uint128)self.h[1] + (t >> 64);
+    ulong g1 = (ulong)t;
+
+    ulong mask = 0 - ((self.h[2] + (ulong)(t >> 64)) >> 2);
+    self.h[0] = (self.h[0] & ~mask) | (g0 & mask);
+    self.h[1] = (self.h[1] & ~mask) | (g1 & mask);
+
+    t = (uint128)self.h[0] + (ulong)self.nonce;
+    self.h[0] = (ulong)t;
+
+    t = (uint128)self.h[1] + (ulong)(self.nonce >> 64) + (t >> 64);
+    self.h[1] = (ulong)t;
+
+    // Store, clear context, return.
+    uint128 result = ((uint128)self.h[1] << 64) + self.h[0];
+    *self = {};
+    return @as_char_view(result)[:TAG_SIZE];
+}
+
+
+fn void Poly1305.blocks(&self, char[] input, ulong pad_bit = 1) @local
+{
+    for (; input.len >= BLOCK_SIZE; input = input[BLOCK_SIZE..])
+    {
+        ulong i0 = mem::load((ulong*)&input[0], 1);
+        ulong i1 = mem::load((ulong*)&input[8], 1);
+
+        uint128 d0 = (uint128)self.h[0] + i0;
+        self.h[0] = (ulong)d0;
+
+        uint128 d1 = (uint128)self.h[1] + (d0 >> 64) + i1;
+        self.h[1] = (ulong)d1;
+
+        self.h[2] += (ulong)(d1 >> 64) + pad_bit;
+
+        d0 = ((uint128)self.h[0] * self.r0) + ((uint128)self.h[1] * self.s1);
+        d1 = ((uint128)self.h[0] * self.r1) + ((uint128)self.h[1] * self.r0) + ((uint128)self.h[2] * self.s1);
+        self.h[2] = (self.h[2] * self.r0);
+
+        self.h[0] = (ulong)d0;
+
+        d1 = d1 + (d0 >> 64);
+        self.h[1] = (ulong)d1;
+        self.h[2] = self.h[2] + (ulong)(d1 >> 64);
+
+        ulong c = (self.h[2] >> 2) + (self.h[2] & ~(ulong)3);
+
+        self.h[2] &= 3;
+        self.h[0] += c;
+        c = constant_time_carry(self.h[0], c);
+        self.h[1] += c;
+        self.h[2] += constant_time_carry(self.h[1], c);
+    }
+}

lib/std/hash/ripemd.c3

@@ -0,0 +1,577 @@
+// Copyright (c) 2025 Zack Puhl <github@xmit.xyz>. 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::ripemd;
+const usz[4] PERMISSIBLE_SIZES_BITS = { 128, 160, 256, 320 };
+
+<* Unchanging block size. *>
+const BLOCK_SIZE = 64;
+
+
+<* RIPE-MD initial values. *>
+const uint[10] H = {
+	0x67452301, 0xefcdab89, 0x98badcfe, 0x10325476, 0xc3d2e1f0,
+	0x76543210, 0xfedcba98, 0x89abcdef, 0x01234567, 0x3c2d1e0f
+};
+
+<* RIPE-MD constant values. *>
+const uint[9] K = {
+	0x00000000, 0x5a827999, 0x6ed9eba1, 0x8f1bbcdc, 0xa953fd4e,
+	0x50a28be6, 0x5c4dd124, 0x6d703ef3, 0x7a6d76e9
+};
+
+<*
+ @require @in(DIGEST_BITS, ...PERMISSIBLE_SIZES_BITS) : "Invalid DIGEST_BITS; must be one of {128, 160, 256, 320}."
+*>
+module std::hash::ripemd <DIGEST_BITS>;
+
+<* Ultimate digest's size in bytes. *>
+const DIGEST_BYTES = DIGEST_BITS / 8;
+
+
+struct RipeMd
+{
+	uint[10]	state;
+	uint[16]	buffer;
+	ulong		byte_count;
+}
+
+
+<*
+ Compute the RIPE-MD hash of the given data.
+
+ @param [in] data : "The data to hash."
+*>
+macro char[*] hash(char[] data)
+{
+	RipeMd r @noinit;
+	r.init();
+	r.update(data);
+	return r.final();
+}
+
+<*
+ Initialize the RIPE-MD hash context.
+*>
+fn void RipeMd.init(&self)
+{
+	mem::zero_volatile(@as_char_view(*self));
+	self.state[..] = H[..];
+	// I don't care about the reason, I want to strangle whoever thought this was a good idea.
+	//   I get it, you have to set the lane initializers correctly, but damn I want my hours back.
+	$if DIGEST_BITS == 256: self.state[4:4] = self.state[5:4]; $endif
+}
+
+<*
+ Update the RIPE-MD hash context with the given data.
+
+ @param [in] data : "The data to digest."
+*>
+fn void RipeMd.update(&self, char[] data)
+{
+	usz $bufsz = $sizeof(self.buffer);
+	uint avail = $bufsz - (uint)(self.byte_count & 0x3f);
+	self.byte_count += data.len;
+
+	if (avail > data.len)
+	{
+		@as_char_view(self.buffer)[($bufsz - avail) : data.len] = data[..];
+		return;
+	}
+
+	@as_char_view(self.buffer)[($bufsz - avail) : avail] = data[:avail];
+	self.transform(self.buffer);
+	data = data[avail..];
+
+	for (bool is_aligned = 0 == (usz)data.ptr % uint.sizeof; data.len >= $bufsz; data = data[$bufsz..])
+	{
+		if (is_aligned) // when aligned, this optimization is a ~10-20% performance boost
+		{
+			self.transform(((uint*)data.ptr)[:$bufsz / uint.sizeof]);
+		}
+		else
+		{
+			@as_char_view(self.buffer)[:$bufsz] = data[:$bufsz];
+			self.transform(self.buffer);
+		}
+	}
+
+	@as_char_view(self.buffer)[:data.len] = data[..];
+}
+
+<*
+ Finalize the RIPE-MD hash context and return the final hash value.
+ This implicitly destroys the underlying hash structure for the sake of security.
+*>
+fn char[DIGEST_BYTES] RipeMd.final(&self)
+{
+	char[DIGEST_BYTES] result @align(uint.sizeof);
+	static char[64] padding = { [0] = 0x80, [1..63] = 0x00 };
+
+	// Napkin maffs.
+	ulong bits = (ulong)self.byte_count << 3;
+	uint index = (uint)self.byte_count & 0x3f;
+	uint padlen = (index < 56) ? (56 - index) : ((64 + 56) - index);
+
+	// Update with padding, then append bit-length of digested message.
+	self.update(padding[:padlen]);
+	self.update(@as_char_view(bits));
+
+	// Clean up implicitly.
+	defer mem::zero_volatile(@as_char_view(*self));
+
+	// Copy and return.
+	((uint*)&result)[:DIGEST_BYTES / uint.sizeof] = self.state[:DIGEST_BYTES / uint.sizeof];
+	return result;
+}
+
+
+// State-based permutation functions given as a function pointer to each computation lane.
+fn uint f1(uint x, uint y, uint z) @inline @local => x ^ y ^ z;
+fn uint f2(uint x, uint y, uint z) @inline @local => z ^ (x & (y ^ z));
+fn uint f3(uint x, uint y, uint z) @inline @local => (x | ~y) ^ z;
+fn uint f4(uint x, uint y, uint z) @inline @local => y ^ (z & (x ^ y));
+fn uint f5(uint x, uint y, uint z) @inline @local => x ^ (y | ~z);
+
+// The primary workhorse of the digest function. Behavior only changes slightly.
+//   All 128/256 inputs will expand with 'e' as 0 - they don't use it anyway.
+macro @round(#a, b, #c, d, e, #func, k, x, s) @local
+{
+	#a += #func(b, #c, d) + x + k;
+	$switch:
+		$case DIGEST_BITS == 128 ||| DIGEST_BITS == 256:
+			#a = #a.rotl(s);
+		$case DIGEST_BITS == 160 ||| DIGEST_BITS == 320:
+			#a = #a.rotl(s) + e;
+			#c = #c.rotl(10);
+		$default:
+			$error "Invalid digest bits";
+	$endswitch
+}
+
+<*
+ Apply a transformation to some input data, but DO NOT inline the contents of the macro within the parent call.
+
+ @param in : "Data or message block to process."
+*>
+fn void RipeMd.transform(&self, uint[BLOCK_SIZE / uint.sizeof] in) @noinline => self.@transform(in);
+
+<*
+ @param in : "Data or message block to process."
+*>
+macro RipeMd.@transform(&self, uint[BLOCK_SIZE / uint.sizeof] in) @local
+{
+	uint aa, bb, cc, dd, ee, aaa, bbb, ccc, ddd, eee;
+	uint[] state = self.state[..];
+
+	// Run computations lanes based on the size of the resulting digest.
+	//   The behavior is different enough for each that this can simply stay broken out into separate cases.
+	//   Otherwise, lacing them all together is just not worth the effort and isn't as clear to other developers.
+	//   (No shade, but see 'RustCrypto/hashes' RIPE-MD macros for an example of what I mean...)
+	$switch DIGEST_BITS:
+		$case 128:
+			aa = state[0];
+			bb = state[1];
+			cc = state[2];
+			dd = state[3];
+			aaa = state[0];
+			bbb = state[1];
+			ccc = state[2];
+			ddd = state[3];
+
+			@round(aa, bb, cc, dd, 0, f1, K[0], in[0],  11);  @round(aaa, bbb, ccc, ddd, 0, f4, K[5], in[5],   8);
+			@round(dd, aa, bb, cc, 0, f1, K[0], in[1],  14);  @round(ddd, aaa, bbb, ccc, 0, f4, K[5], in[14],  9);
+			@round(cc, dd, aa, bb, 0, f1, K[0], in[2],  15);  @round(ccc, ddd, aaa, bbb, 0, f4, K[5], in[7],   9);
+			@round(bb, cc, dd, aa, 0, f1, K[0], in[3],  12);  @round(bbb, ccc, ddd, aaa, 0, f4, K[5], in[0],  11);
+			@round(aa, bb, cc, dd, 0, f1, K[0], in[4],   5);  @round(aaa, bbb, ccc, ddd, 0, f4, K[5], in[9],  13);
+			@round(dd, aa, bb, cc, 0, f1, K[0], in[5],   8);  @round(ddd, aaa, bbb, ccc, 0, f4, K[5], in[2],  15);
+			@round(cc, dd, aa, bb, 0, f1, K[0], in[6],   7);  @round(ccc, ddd, aaa, bbb, 0, f4, K[5], in[11], 15);
+			@round(bb, cc, dd, aa, 0, f1, K[0], in[7],   9);  @round(bbb, ccc, ddd, aaa, 0, f4, K[5], in[4],   5);
+			@round(aa, bb, cc, dd, 0, f1, K[0], in[8],  11);  @round(aaa, bbb, ccc, ddd, 0, f4, K[5], in[13],  7);
+			@round(dd, aa, bb, cc, 0, f1, K[0], in[9],  13);  @round(ddd, aaa, bbb, ccc, 0, f4, K[5], in[6],   7);
+			@round(cc, dd, aa, bb, 0, f1, K[0], in[10], 14);  @round(ccc, ddd, aaa, bbb, 0, f4, K[5], in[15],  8);
+			@round(bb, cc, dd, aa, 0, f1, K[0], in[11], 15);  @round(bbb, ccc, ddd, aaa, 0, f4, K[5], in[8],  11);
+			@round(aa, bb, cc, dd, 0, f1, K[0], in[12],  6);  @round(aaa, bbb, ccc, ddd, 0, f4, K[5], in[1],  14);
+			@round(dd, aa, bb, cc, 0, f1, K[0], in[13],  7);  @round(ddd, aaa, bbb, ccc, 0, f4, K[5], in[10], 14);
+			@round(cc, dd, aa, bb, 0, f1, K[0], in[14],  9);  @round(ccc, ddd, aaa, bbb, 0, f4, K[5], in[3],  12);
+			@round(bb, cc, dd, aa, 0, f1, K[0], in[15],  8);  @round(bbb, ccc, ddd, aaa, 0, f4, K[5], in[12],  6);
+
+			@round(aa, bb, cc, dd, 0, f2, K[1], in[7],   7);  @round(aaa, bbb, ccc, ddd, 0, f3, K[6], in[6],   9);
+			@round(dd, aa, bb, cc, 0, f2, K[1], in[4],   6);  @round(ddd, aaa, bbb, ccc, 0, f3, K[6], in[11], 13);
+			@round(cc, dd, aa, bb, 0, f2, K[1], in[13],  8);  @round(ccc, ddd, aaa, bbb, 0, f3, K[6], in[3],  15);
+			@round(bb, cc, dd, aa, 0, f2, K[1], in[1],  13);  @round(bbb, ccc, ddd, aaa, 0, f3, K[6], in[7],   7);
+			@round(aa, bb, cc, dd, 0, f2, K[1], in[10], 11);  @round(aaa, bbb, ccc, ddd, 0, f3, K[6], in[0],  12);
+			@round(dd, aa, bb, cc, 0, f2, K[1], in[6],   9);  @round(ddd, aaa, bbb, ccc, 0, f3, K[6], in[13],  8);
+			@round(cc, dd, aa, bb, 0, f2, K[1], in[15],  7);  @round(ccc, ddd, aaa, bbb, 0, f3, K[6], in[5],   9);
+			@round(bb, cc, dd, aa, 0, f2, K[1], in[3],  15);  @round(bbb, ccc, ddd, aaa, 0, f3, K[6], in[10], 11);
+			@round(aa, bb, cc, dd, 0, f2, K[1], in[12],  7);  @round(aaa, bbb, ccc, ddd, 0, f3, K[6], in[14],  7);
+			@round(dd, aa, bb, cc, 0, f2, K[1], in[0],  12);  @round(ddd, aaa, bbb, ccc, 0, f3, K[6], in[15],  7);
+			@round(cc, dd, aa, bb, 0, f2, K[1], in[9],  15);  @round(ccc, ddd, aaa, bbb, 0, f3, K[6], in[8],  12);
+			@round(bb, cc, dd, aa, 0, f2, K[1], in[5],   9);  @round(bbb, ccc, ddd, aaa, 0, f3, K[6], in[12],  7);
+			@round(aa, bb, cc, dd, 0, f2, K[1], in[2],  11);  @round(aaa, bbb, ccc, ddd, 0, f3, K[6], in[4],   6);
+			@round(dd, aa, bb, cc, 0, f2, K[1], in[14],  7);  @round(ddd, aaa, bbb, ccc, 0, f3, K[6], in[9],  15);
+			@round(cc, dd, aa, bb, 0, f2, K[1], in[11], 13);  @round(ccc, ddd, aaa, bbb, 0, f3, K[6], in[1],  13);
+			@round(bb, cc, dd, aa, 0, f2, K[1], in[8],  12);  @round(bbb, ccc, ddd, aaa, 0, f3, K[6], in[2],  11);
+
+			@round(aa, bb, cc, dd, 0, f3, K[2], in[3],  11);  @round(aaa, bbb, ccc, ddd, 0, f2, K[7], in[15],  9);
+			@round(dd, aa, bb, cc, 0, f3, K[2], in[10], 13);  @round(ddd, aaa, bbb, ccc, 0, f2, K[7], in[5],   7);
+			@round(cc, dd, aa, bb, 0, f3, K[2], in[14],  6);  @round(ccc, ddd, aaa, bbb, 0, f2, K[7], in[1],  15);
+			@round(bb, cc, dd, aa, 0, f3, K[2], in[4],   7);  @round(bbb, ccc, ddd, aaa, 0, f2, K[7], in[3],  11);
+			@round(aa, bb, cc, dd, 0, f3, K[2], in[9],  14);  @round(aaa, bbb, ccc, ddd, 0, f2, K[7], in[7],   8);
+			@round(dd, aa, bb, cc, 0, f3, K[2], in[15],  9);  @round(ddd, aaa, bbb, ccc, 0, f2, K[7], in[14],  6);
+			@round(cc, dd, aa, bb, 0, f3, K[2], in[8],  13);  @round(ccc, ddd, aaa, bbb, 0, f2, K[7], in[6],   6);
+			@round(bb, cc, dd, aa, 0, f3, K[2], in[1],  15);  @round(bbb, ccc, ddd, aaa, 0, f2, K[7], in[9],  14);
+			@round(aa, bb, cc, dd, 0, f3, K[2], in[2],  14);  @round(aaa, bbb, ccc, ddd, 0, f2, K[7], in[11], 12);
+			@round(dd, aa, bb, cc, 0, f3, K[2], in[7],   8);  @round(ddd, aaa, bbb, ccc, 0, f2, K[7], in[8],  13);
+			@round(cc, dd, aa, bb, 0, f3, K[2], in[0],  13);  @round(ccc, ddd, aaa, bbb, 0, f2, K[7], in[12],  5);
+			@round(bb, cc, dd, aa, 0, f3, K[2], in[6],   6);  @round(bbb, ccc, ddd, aaa, 0, f2, K[7], in[2],  14);
+			@round(aa, bb, cc, dd, 0, f3, K[2], in[13],  5);  @round(aaa, bbb, ccc, ddd, 0, f2, K[7], in[10], 13);
+			@round(dd, aa, bb, cc, 0, f3, K[2], in[11], 12);  @round(ddd, aaa, bbb, ccc, 0, f2, K[7], in[0],  13);
+			@round(cc, dd, aa, bb, 0, f3, K[2], in[5],   7);  @round(ccc, ddd, aaa, bbb, 0, f2, K[7], in[4],   7);
+			@round(bb, cc, dd, aa, 0, f3, K[2], in[12],  5);  @round(bbb, ccc, ddd, aaa, 0, f2, K[7], in[13],  5);
+
+			@round(aa, bb, cc, dd, 0, f4, K[3], in[1],  11);  @round(aaa, bbb, ccc, ddd, 0, f1, 0, in[8],  15);
+			@round(dd, aa, bb, cc, 0, f4, K[3], in[9],  12);  @round(ddd, aaa, bbb, ccc, 0, f1, 0, in[6],   5);
+			@round(cc, dd, aa, bb, 0, f4, K[3], in[11], 14);  @round(ccc, ddd, aaa, bbb, 0, f1, 0, in[4],   8);
+			@round(bb, cc, dd, aa, 0, f4, K[3], in[10], 15);  @round(bbb, ccc, ddd, aaa, 0, f1, 0, in[1],  11);
+			@round(aa, bb, cc, dd, 0, f4, K[3], in[0],  14);  @round(aaa, bbb, ccc, ddd, 0, f1, 0, in[3],  14);
+			@round(dd, aa, bb, cc, 0, f4, K[3], in[8],  15);  @round(ddd, aaa, bbb, ccc, 0, f1, 0, in[11], 14);
+			@round(cc, dd, aa, bb, 0, f4, K[3], in[12],  9);  @round(ccc, ddd, aaa, bbb, 0, f1, 0, in[15],  6);
+			@round(bb, cc, dd, aa, 0, f4, K[3], in[4],   8);  @round(bbb, ccc, ddd, aaa, 0, f1, 0, in[0],  14);
+			@round(aa, bb, cc, dd, 0, f4, K[3], in[13],  9);  @round(aaa, bbb, ccc, ddd, 0, f1, 0, in[5],   6);
+			@round(dd, aa, bb, cc, 0, f4, K[3], in[3],  14);  @round(ddd, aaa, bbb, ccc, 0, f1, 0, in[12],  9);
+			@round(cc, dd, aa, bb, 0, f4, K[3], in[7],   5);  @round(ccc, ddd, aaa, bbb, 0, f1, 0, in[2],  12);
+			@round(bb, cc, dd, aa, 0, f4, K[3], in[15],  6);  @round(bbb, ccc, ddd, aaa, 0, f1, 0, in[13],  9);
+			@round(aa, bb, cc, dd, 0, f4, K[3], in[14],  8);  @round(aaa, bbb, ccc, ddd, 0, f1, 0, in[9],  12);
+			@round(dd, aa, bb, cc, 0, f4, K[3], in[5],   6);  @round(ddd, aaa, bbb, ccc, 0, f1, 0, in[7],   5);
+			@round(cc, dd, aa, bb, 0, f4, K[3], in[6],   5);  @round(ccc, ddd, aaa, bbb, 0, f1, 0, in[10], 15);
+			@round(bb, cc, dd, aa, 0, f4, K[3], in[2],  12);  @round(bbb, ccc, ddd, aaa, 0, f1, 0, in[14],  8);
+
+			ddd += cc + state[1];
+			state[1] = state[2] + dd + aaa;
+			state[2] = state[3] + aa + bbb;
+			state[3] = state[0] + bb + ccc;
+			state[0] = ddd;
+
+		$case 160:
+			aa = state[0];
+			bb = state[1];
+			cc = state[2];
+			dd = state[3];
+			ee = state[4];
+			aaa = state[0];
+			bbb = state[1];
+			ccc = state[2];
+			ddd = state[3];
+			eee = state[4];
+
+			@round(aa, bb, cc, dd, ee, f1, K[0], in[0],  11);  @round(aaa, bbb, ccc, ddd, eee, f5, K[5], in[5],   8);
+			@round(ee, aa, bb, cc, dd, f1, K[0], in[1],  14);  @round(eee, aaa, bbb, ccc, ddd, f5, K[5], in[14],  9);
+			@round(dd, ee, aa, bb, cc, f1, K[0], in[2],  15);  @round(ddd, eee, aaa, bbb, ccc, f5, K[5], in[7],   9);
+			@round(cc, dd, ee, aa, bb, f1, K[0], in[3],  12);  @round(ccc, ddd, eee, aaa, bbb, f5, K[5], in[0],  11);
+			@round(bb, cc, dd, ee, aa, f1, K[0], in[4],   5);  @round(bbb, ccc, ddd, eee, aaa, f5, K[5], in[9],  13);
+			@round(aa, bb, cc, dd, ee, f1, K[0], in[5],   8);  @round(aaa, bbb, ccc, ddd, eee, f5, K[5], in[2],  15);
+			@round(ee, aa, bb, cc, dd, f1, K[0], in[6],   7);  @round(eee, aaa, bbb, ccc, ddd, f5, K[5], in[11], 15);
+			@round(dd, ee, aa, bb, cc, f1, K[0], in[7],   9);  @round(ddd, eee, aaa, bbb, ccc, f5, K[5], in[4],   5);
+			@round(cc, dd, ee, aa, bb, f1, K[0], in[8],  11);  @round(ccc, ddd, eee, aaa, bbb, f5, K[5], in[13],  7);
+			@round(bb, cc, dd, ee, aa, f1, K[0], in[9],  13);  @round(bbb, ccc, ddd, eee, aaa, f5, K[5], in[6],   7);
+			@round(aa, bb, cc, dd, ee, f1, K[0], in[10], 14);  @round(aaa, bbb, ccc, ddd, eee, f5, K[5], in[15],  8);
+			@round(ee, aa, bb, cc, dd, f1, K[0], in[11], 15);  @round(eee, aaa, bbb, ccc, ddd, f5, K[5], in[8],  11);
+			@round(dd, ee, aa, bb, cc, f1, K[0], in[12],  6);  @round(ddd, eee, aaa, bbb, ccc, f5, K[5], in[1],  14);
+			@round(cc, dd, ee, aa, bb, f1, K[0], in[13],  7);  @round(ccc, ddd, eee, aaa, bbb, f5, K[5], in[10], 14);
+			@round(bb, cc, dd, ee, aa, f1, K[0], in[14],  9);  @round(bbb, ccc, ddd, eee, aaa, f5, K[5], in[3],  12);
+			@round(aa, bb, cc, dd, ee, f1, K[0], in[15],  8);  @round(aaa, bbb, ccc, ddd, eee, f5, K[5], in[12],  6);
+
+			@round(ee, aa, bb, cc, dd, f2, K[1], in[7],   7);  @round(eee, aaa, bbb, ccc, ddd, f4, K[6], in[6],   9);
+			@round(dd, ee, aa, bb, cc, f2, K[1], in[4],   6);  @round(ddd, eee, aaa, bbb, ccc, f4, K[6], in[11], 13);
+			@round(cc, dd, ee, aa, bb, f2, K[1], in[13],  8);  @round(ccc, ddd, eee, aaa, bbb, f4, K[6], in[3],  15);
+			@round(bb, cc, dd, ee, aa, f2, K[1], in[1],  13);  @round(bbb, ccc, ddd, eee, aaa, f4, K[6], in[7],   7);
+			@round(aa, bb, cc, dd, ee, f2, K[1], in[10], 11);  @round(aaa, bbb, ccc, ddd, eee, f4, K[6], in[0],  12);
+			@round(ee, aa, bb, cc, dd, f2, K[1], in[6],   9);  @round(eee, aaa, bbb, ccc, ddd, f4, K[6], in[13],  8);
+			@round(dd, ee, aa, bb, cc, f2, K[1], in[15],  7);  @round(ddd, eee, aaa, bbb, ccc, f4, K[6], in[5],   9);
+			@round(cc, dd, ee, aa, bb, f2, K[1], in[3],  15);  @round(ccc, ddd, eee, aaa, bbb, f4, K[6], in[10], 11);
+			@round(bb, cc, dd, ee, aa, f2, K[1], in[12],  7);  @round(bbb, ccc, ddd, eee, aaa, f4, K[6], in[14],  7);
+			@round(aa, bb, cc, dd, ee, f2, K[1], in[0],  12);  @round(aaa, bbb, ccc, ddd, eee, f4, K[6], in[15],  7);
+			@round(ee, aa, bb, cc, dd, f2, K[1], in[9],  15);  @round(eee, aaa, bbb, ccc, ddd, f4, K[6], in[8],  12);
+			@round(dd, ee, aa, bb, cc, f2, K[1], in[5],   9);  @round(ddd, eee, aaa, bbb, ccc, f4, K[6], in[12],  7);
+			@round(cc, dd, ee, aa, bb, f2, K[1], in[2],  11);  @round(ccc, ddd, eee, aaa, bbb, f4, K[6], in[4],   6);
+			@round(bb, cc, dd, ee, aa, f2, K[1], in[14],  7);  @round(bbb, ccc, ddd, eee, aaa, f4, K[6], in[9],  15);
+			@round(aa, bb, cc, dd, ee, f2, K[1], in[11], 13);  @round(aaa, bbb, ccc, ddd, eee, f4, K[6], in[1],  13);
+			@round(ee, aa, bb, cc, dd, f2, K[1], in[8],  12);  @round(eee, aaa, bbb, ccc, ddd, f4, K[6], in[2],  11);
+
+			@round(dd, ee, aa, bb, cc, f3, K[2], in[3],  11);  @round(ddd, eee, aaa, bbb, ccc, f3, K[7], in[15],  9);
+			@round(cc, dd, ee, aa, bb, f3, K[2], in[10], 13);  @round(ccc, ddd, eee, aaa, bbb, f3, K[7], in[5],   7);
+			@round(bb, cc, dd, ee, aa, f3, K[2], in[14],  6);  @round(bbb, ccc, ddd, eee, aaa, f3, K[7], in[1],  15);
+			@round(aa, bb, cc, dd, ee, f3, K[2], in[4],   7);  @round(aaa, bbb, ccc, ddd, eee, f3, K[7], in[3],  11);
+			@round(ee, aa, bb, cc, dd, f3, K[2], in[9],  14);  @round(eee, aaa, bbb, ccc, ddd, f3, K[7], in[7],   8);
+			@round(dd, ee, aa, bb, cc, f3, K[2], in[15],  9);  @round(ddd, eee, aaa, bbb, ccc, f3, K[7], in[14],  6);
+			@round(cc, dd, ee, aa, bb, f3, K[2], in[8],  13);  @round(ccc, ddd, eee, aaa, bbb, f3, K[7], in[6],   6);
+			@round(bb, cc, dd, ee, aa, f3, K[2], in[1],  15);  @round(bbb, ccc, ddd, eee, aaa, f3, K[7], in[9],  14);
+			@round(aa, bb, cc, dd, ee, f3, K[2], in[2],  14);  @round(aaa, bbb, ccc, ddd, eee, f3, K[7], in[11], 12);
+			@round(ee, aa, bb, cc, dd, f3, K[2], in[7],   8);  @round(eee, aaa, bbb, ccc, ddd, f3, K[7], in[8],  13);
+			@round(dd, ee, aa, bb, cc, f3, K[2], in[0],  13);  @round(ddd, eee, aaa, bbb, ccc, f3, K[7], in[12],  5);
+			@round(cc, dd, ee, aa, bb, f3, K[2], in[6],   6);  @round(ccc, ddd, eee, aaa, bbb, f3, K[7], in[2],  14);
+			@round(bb, cc, dd, ee, aa, f3, K[2], in[13],  5);  @round(bbb, ccc, ddd, eee, aaa, f3, K[7], in[10], 13);
+			@round(aa, bb, cc, dd, ee, f3, K[2], in[11], 12);  @round(aaa, bbb, ccc, ddd, eee, f3, K[7], in[0],  13);
+			@round(ee, aa, bb, cc, dd, f3, K[2], in[5],   7);  @round(eee, aaa, bbb, ccc, ddd, f3, K[7], in[4],   7);
+			@round(dd, ee, aa, bb, cc, f3, K[2], in[12],  5);  @round(ddd, eee, aaa, bbb, ccc, f3, K[7], in[13],  5);
+
+			@round(cc, dd, ee, aa, bb, f4, K[3], in[1],  11);  @round(ccc, ddd, eee, aaa, bbb, f2, K[8], in[8],  15);
+			@round(bb, cc, dd, ee, aa, f4, K[3], in[9],  12);  @round(bbb, ccc, ddd, eee, aaa, f2, K[8], in[6],   5);
+			@round(aa, bb, cc, dd, ee, f4, K[3], in[11], 14);  @round(aaa, bbb, ccc, ddd, eee, f2, K[8], in[4],   8);
+			@round(ee, aa, bb, cc, dd, f4, K[3], in[10], 15);  @round(eee, aaa, bbb, ccc, ddd, f2, K[8], in[1],  11);
+			@round(dd, ee, aa, bb, cc, f4, K[3], in[0],  14);  @round(ddd, eee, aaa, bbb, ccc, f2, K[8], in[3],  14);
+			@round(cc, dd, ee, aa, bb, f4, K[3], in[8],  15);  @round(ccc, ddd, eee, aaa, bbb, f2, K[8], in[11], 14);
+			@round(bb, cc, dd, ee, aa, f4, K[3], in[12],  9);  @round(bbb, ccc, ddd, eee, aaa, f2, K[8], in[15],  6);
+			@round(aa, bb, cc, dd, ee, f4, K[3], in[4],   8);  @round(aaa, bbb, ccc, ddd, eee, f2, K[8], in[0],  14);
+			@round(ee, aa, bb, cc, dd, f4, K[3], in[13],  9);  @round(eee, aaa, bbb, ccc, ddd, f2, K[8], in[5],   6);
+			@round(dd, ee, aa, bb, cc, f4, K[3], in[3],  14);  @round(ddd, eee, aaa, bbb, ccc, f2, K[8], in[12],  9);
+			@round(cc, dd, ee, aa, bb, f4, K[3], in[7],   5);  @round(ccc, ddd, eee, aaa, bbb, f2, K[8], in[2],  12);
+			@round(bb, cc, dd, ee, aa, f4, K[3], in[15],  6);  @round(bbb, ccc, ddd, eee, aaa, f2, K[8], in[13],  9);
+			@round(aa, bb, cc, dd, ee, f4, K[3], in[14],  8);  @round(aaa, bbb, ccc, ddd, eee, f2, K[8], in[9],  12);
+			@round(ee, aa, bb, cc, dd, f4, K[3], in[5],   6);  @round(eee, aaa, bbb, ccc, ddd, f2, K[8], in[7],   5);
+			@round(dd, ee, aa, bb, cc, f4, K[3], in[6],   5);  @round(ddd, eee, aaa, bbb, ccc, f2, K[8], in[10], 15);
+			@round(cc, dd, ee, aa, bb, f4, K[3], in[2],  12);  @round(ccc, ddd, eee, aaa, bbb, f2, K[8], in[14],  8);
+
+			@round(bb, cc, dd, ee, aa, f5, K[4], in[4],   9);  @round(bbb, ccc, ddd, eee, aaa, f1, 0, in[12],  8);
+			@round(aa, bb, cc, dd, ee, f5, K[4], in[0],  15);  @round(aaa, bbb, ccc, ddd, eee, f1, 0, in[15],  5);
+			@round(ee, aa, bb, cc, dd, f5, K[4], in[5],   5);  @round(eee, aaa, bbb, ccc, ddd, f1, 0, in[10], 12);
+			@round(dd, ee, aa, bb, cc, f5, K[4], in[9],  11);  @round(ddd, eee, aaa, bbb, ccc, f1, 0, in[4],   9);
+			@round(cc, dd, ee, aa, bb, f5, K[4], in[7],   6);  @round(ccc, ddd, eee, aaa, bbb, f1, 0, in[1],  12);
+			@round(bb, cc, dd, ee, aa, f5, K[4], in[12],  8);  @round(bbb, ccc, ddd, eee, aaa, f1, 0, in[5],   5);
+			@round(aa, bb, cc, dd, ee, f5, K[4], in[2],  13);  @round(aaa, bbb, ccc, ddd, eee, f1, 0, in[8],  14);
+			@round(ee, aa, bb, cc, dd, f5, K[4], in[10], 12);  @round(eee, aaa, bbb, ccc, ddd, f1, 0, in[7],   6);
+			@round(dd, ee, aa, bb, cc, f5, K[4], in[14],  5);  @round(ddd, eee, aaa, bbb, ccc, f1, 0, in[6],   8);
+			@round(cc, dd, ee, aa, bb, f5, K[4], in[1],  12);  @round(ccc, ddd, eee, aaa, bbb, f1, 0, in[2],  13);
+			@round(bb, cc, dd, ee, aa, f5, K[4], in[3],  13);  @round(bbb, ccc, ddd, eee, aaa, f1, 0, in[13],  6);
+			@round(aa, bb, cc, dd, ee, f5, K[4], in[8],  14);  @round(aaa, bbb, ccc, ddd, eee, f1, 0, in[14],  5);
+			@round(ee, aa, bb, cc, dd, f5, K[4], in[11], 11);  @round(eee, aaa, bbb, ccc, ddd, f1, 0, in[0],  15);
+			@round(dd, ee, aa, bb, cc, f5, K[4], in[6],   8);  @round(ddd, eee, aaa, bbb, ccc, f1, 0, in[3],  13);
+			@round(cc, dd, ee, aa, bb, f5, K[4], in[15],  5);  @round(ccc, ddd, eee, aaa, bbb, f1, 0, in[9],  11);
+			@round(bb, cc, dd, ee, aa, f5, K[4], in[13],  6);  @round(bbb, ccc, ddd, eee, aaa, f1, 0, in[11], 11);
+
+			ddd += cc + state[1];
+			state[1] = state[2] + dd + eee;
+			state[2] = state[3] + ee + aaa;
+			state[3] = state[4] + aa + bbb;
+			state[4] = state[0] + bb + ccc;
+			state[0] = ddd;
+
+		$case 256:
+			aa = state[0];
+			bb = state[1];
+			cc = state[2];
+			dd = state[3];
+			aaa = state[4];
+			bbb = state[5];
+			ccc = state[6];
+			ddd = state[7];
+
+			@round(aa, bb, cc, dd, 0, f1, K[0], in[0],  11);  @round(aaa, bbb, ccc, ddd, 0, f4, K[5], in[5],   8);
+			@round(dd, aa, bb, cc, 0, f1, K[0], in[1],  14);  @round(ddd, aaa, bbb, ccc, 0, f4, K[5], in[14],  9);
+			@round(cc, dd, aa, bb, 0, f1, K[0], in[2],  15);  @round(ccc, ddd, aaa, bbb, 0, f4, K[5], in[7],   9);
+			@round(bb, cc, dd, aa, 0, f1, K[0], in[3],  12);  @round(bbb, ccc, ddd, aaa, 0, f4, K[5], in[0],  11);
+			@round(aa, bb, cc, dd, 0, f1, K[0], in[4],   5);  @round(aaa, bbb, ccc, ddd, 0, f4, K[5], in[9],  13);
+			@round(dd, aa, bb, cc, 0, f1, K[0], in[5],   8);  @round(ddd, aaa, bbb, ccc, 0, f4, K[5], in[2],  15);
+			@round(cc, dd, aa, bb, 0, f1, K[0], in[6],   7);  @round(ccc, ddd, aaa, bbb, 0, f4, K[5], in[11], 15);
+			@round(bb, cc, dd, aa, 0, f1, K[0], in[7],   9);  @round(bbb, ccc, ddd, aaa, 0, f4, K[5], in[4],   5);
+			@round(aa, bb, cc, dd, 0, f1, K[0], in[8],  11);  @round(aaa, bbb, ccc, ddd, 0, f4, K[5], in[13],  7);
+			@round(dd, aa, bb, cc, 0, f1, K[0], in[9],  13);  @round(ddd, aaa, bbb, ccc, 0, f4, K[5], in[6],   7);
+			@round(cc, dd, aa, bb, 0, f1, K[0], in[10], 14);  @round(ccc, ddd, aaa, bbb, 0, f4, K[5], in[15],  8);
+			@round(bb, cc, dd, aa, 0, f1, K[0], in[11], 15);  @round(bbb, ccc, ddd, aaa, 0, f4, K[5], in[8],  11);
+			@round(aa, bb, cc, dd, 0, f1, K[0], in[12],  6);  @round(aaa, bbb, ccc, ddd, 0, f4, K[5], in[1],  14);
+			@round(dd, aa, bb, cc, 0, f1, K[0], in[13],  7);  @round(ddd, aaa, bbb, ccc, 0, f4, K[5], in[10], 14);
+			@round(cc, dd, aa, bb, 0, f1, K[0], in[14],  9);  @round(ccc, ddd, aaa, bbb, 0, f4, K[5], in[3],  12);
+			@round(bb, cc, dd, aa, 0, f1, K[0], in[15],  8);  @round(bbb, ccc, ddd, aaa, 0, f4, K[5], in[12],  6);
+
+			@swap(aa, aaa);
+
+			@round(aa, bb, cc, dd, 0, f2, K[1], in[7],   7);  @round(aaa, bbb, ccc, ddd, 0, f3, K[6], in[6],   9);
+			@round(dd, aa, bb, cc, 0, f2, K[1], in[4],   6);  @round(ddd, aaa, bbb, ccc, 0, f3, K[6], in[11], 13);
+			@round(cc, dd, aa, bb, 0, f2, K[1], in[13],  8);  @round(ccc, ddd, aaa, bbb, 0, f3, K[6], in[3],  15);
+			@round(bb, cc, dd, aa, 0, f2, K[1], in[1],  13);  @round(bbb, ccc, ddd, aaa, 0, f3, K[6], in[7],   7);
+			@round(aa, bb, cc, dd, 0, f2, K[1], in[10], 11);  @round(aaa, bbb, ccc, ddd, 0, f3, K[6], in[0],  12);
+			@round(dd, aa, bb, cc, 0, f2, K[1], in[6],   9);  @round(ddd, aaa, bbb, ccc, 0, f3, K[6], in[13],  8);
+			@round(cc, dd, aa, bb, 0, f2, K[1], in[15],  7);  @round(ccc, ddd, aaa, bbb, 0, f3, K[6], in[5],   9);
+			@round(bb, cc, dd, aa, 0, f2, K[1], in[3],  15);  @round(bbb, ccc, ddd, aaa, 0, f3, K[6], in[10], 11);
+			@round(aa, bb, cc, dd, 0, f2, K[1], in[12],  7);  @round(aaa, bbb, ccc, ddd, 0, f3, K[6], in[14],  7);
+			@round(dd, aa, bb, cc, 0, f2, K[1], in[0],  12);  @round(ddd, aaa, bbb, ccc, 0, f3, K[6], in[15],  7);
+			@round(cc, dd, aa, bb, 0, f2, K[1], in[9],  15);  @round(ccc, ddd, aaa, bbb, 0, f3, K[6], in[8],  12);
+			@round(bb, cc, dd, aa, 0, f2, K[1], in[5],   9);  @round(bbb, ccc, ddd, aaa, 0, f3, K[6], in[12],  7);
+			@round(aa, bb, cc, dd, 0, f2, K[1], in[2],  11);  @round(aaa, bbb, ccc, ddd, 0, f3, K[6], in[4],   6);
+			@round(dd, aa, bb, cc, 0, f2, K[1], in[14],  7);  @round(ddd, aaa, bbb, ccc, 0, f3, K[6], in[9],  15);
+			@round(cc, dd, aa, bb, 0, f2, K[1], in[11], 13);  @round(ccc, ddd, aaa, bbb, 0, f3, K[6], in[1],  13);
+			@round(bb, cc, dd, aa, 0, f2, K[1], in[8],  12);  @round(bbb, ccc, ddd, aaa, 0, f3, K[6], in[2],  11);
+
+			@swap(bb, bbb);
+
+			@round(aa, bb, cc, dd, 0, f3, K[2], in[3],  11);  @round(aaa, bbb, ccc, ddd, 0, f2, K[7], in[15],  9);
+			@round(dd, aa, bb, cc, 0, f3, K[2], in[10], 13);  @round(ddd, aaa, bbb, ccc, 0, f2, K[7], in[5],   7);
+			@round(cc, dd, aa, bb, 0, f3, K[2], in[14],  6);  @round(ccc, ddd, aaa, bbb, 0, f2, K[7], in[1],  15);
+			@round(bb, cc, dd, aa, 0, f3, K[2], in[4],   7);  @round(bbb, ccc, ddd, aaa, 0, f2, K[7], in[3],  11);
+			@round(aa, bb, cc, dd, 0, f3, K[2], in[9],  14);  @round(aaa, bbb, ccc, ddd, 0, f2, K[7], in[7],   8);
+			@round(dd, aa, bb, cc, 0, f3, K[2], in[15],  9);  @round(ddd, aaa, bbb, ccc, 0, f2, K[7], in[14],  6);
+			@round(cc, dd, aa, bb, 0, f3, K[2], in[8],  13);  @round(ccc, ddd, aaa, bbb, 0, f2, K[7], in[6],   6);
+			@round(bb, cc, dd, aa, 0, f3, K[2], in[1],  15);  @round(bbb, ccc, ddd, aaa, 0, f2, K[7], in[9],  14);
+			@round(aa, bb, cc, dd, 0, f3, K[2], in[2],  14);  @round(aaa, bbb, ccc, ddd, 0, f2, K[7], in[11], 12);
+			@round(dd, aa, bb, cc, 0, f3, K[2], in[7],   8);  @round(ddd, aaa, bbb, ccc, 0, f2, K[7], in[8],  13);
+			@round(cc, dd, aa, bb, 0, f3, K[2], in[0],  13);  @round(ccc, ddd, aaa, bbb, 0, f2, K[7], in[12],  5);
+			@round(bb, cc, dd, aa, 0, f3, K[2], in[6],   6);  @round(bbb, ccc, ddd, aaa, 0, f2, K[7], in[2],  14);
+			@round(aa, bb, cc, dd, 0, f3, K[2], in[13],  5);  @round(aaa, bbb, ccc, ddd, 0, f2, K[7], in[10], 13);
+			@round(dd, aa, bb, cc, 0, f3, K[2], in[11], 12);  @round(ddd, aaa, bbb, ccc, 0, f2, K[7], in[0],  13);
+			@round(cc, dd, aa, bb, 0, f3, K[2], in[5],   7);  @round(ccc, ddd, aaa, bbb, 0, f2, K[7], in[4],   7);
+			@round(bb, cc, dd, aa, 0, f3, K[2], in[12],  5);  @round(bbb, ccc, ddd, aaa, 0, f2, K[7], in[13],  5);
+
+			@swap(cc, ccc);
+
+			@round(aa, bb, cc, dd, 0, f4, K[3], in[1],  11);  @round(aaa, bbb, ccc, ddd, 0, f1, 0, in[8],  15);
+			@round(dd, aa, bb, cc, 0, f4, K[3], in[9],  12);  @round(ddd, aaa, bbb, ccc, 0, f1, 0, in[6],   5);
+			@round(cc, dd, aa, bb, 0, f4, K[3], in[11], 14);  @round(ccc, ddd, aaa, bbb, 0, f1, 0, in[4],   8);
+			@round(bb, cc, dd, aa, 0, f4, K[3], in[10], 15);  @round(bbb, ccc, ddd, aaa, 0, f1, 0, in[1],  11);
+			@round(aa, bb, cc, dd, 0, f4, K[3], in[0],  14);  @round(aaa, bbb, ccc, ddd, 0, f1, 0, in[3],  14);
+			@round(dd, aa, bb, cc, 0, f4, K[3], in[8],  15);  @round(ddd, aaa, bbb, ccc, 0, f1, 0, in[11], 14);
+			@round(cc, dd, aa, bb, 0, f4, K[3], in[12],  9);  @round(ccc, ddd, aaa, bbb, 0, f1, 0, in[15],  6);
+			@round(bb, cc, dd, aa, 0, f4, K[3], in[4],   8);  @round(bbb, ccc, ddd, aaa, 0, f1, 0, in[0],  14);
+			@round(aa, bb, cc, dd, 0, f4, K[3], in[13],  9);  @round(aaa, bbb, ccc, ddd, 0, f1, 0, in[5],   6);
+			@round(dd, aa, bb, cc, 0, f4, K[3], in[3],  14);  @round(ddd, aaa, bbb, ccc, 0, f1, 0, in[12],  9);
+			@round(cc, dd, aa, bb, 0, f4, K[3], in[7],   5);  @round(ccc, ddd, aaa, bbb, 0, f1, 0, in[2],  12);
+			@round(bb, cc, dd, aa, 0, f4, K[3], in[15],  6);  @round(bbb, ccc, ddd, aaa, 0, f1, 0, in[13],  9);
+			@round(aa, bb, cc, dd, 0, f4, K[3], in[14],  8);  @round(aaa, bbb, ccc, ddd, 0, f1, 0, in[9],  12);
+			@round(dd, aa, bb, cc, 0, f4, K[3], in[5],   6);  @round(ddd, aaa, bbb, ccc, 0, f1, 0, in[7],   5);
+			@round(cc, dd, aa, bb, 0, f4, K[3], in[6],   5);  @round(ccc, ddd, aaa, bbb, 0, f1, 0, in[10], 15);
+			@round(bb, cc, dd, aa, 0, f4, K[3], in[2],  12);  @round(bbb, ccc, ddd, aaa, 0, f1, 0, in[14],  8);
+
+			@swap(dd, ddd);
+
+			state[0] += aa;
+			state[1] += bb;
+			state[2] += cc;
+			state[3] += dd;
+			state[4] += aaa;
+			state[5] += bbb;
+			state[6] += ccc;
+			state[7] += ddd;
+
+		$case 320:
+			aa = state[0];
+			bb = state[1];
+			cc = state[2];
+			dd = state[3];
+			ee = state[4];
+			aaa = state[5];
+			bbb = state[6];
+			ccc = state[7];
+			ddd = state[8];
+			eee = state[9];
+
+			@round(aa, bb, cc, dd, ee, f1, K[0], in[0],  11);  @round(aaa, bbb, ccc, ddd, eee, f5, K[5], in[5],   8);
+			@round(ee, aa, bb, cc, dd, f1, K[0], in[1],  14);  @round(eee, aaa, bbb, ccc, ddd, f5, K[5], in[14],  9);
+			@round(dd, ee, aa, bb, cc, f1, K[0], in[2],  15);  @round(ddd, eee, aaa, bbb, ccc, f5, K[5], in[7],   9);
+			@round(cc, dd, ee, aa, bb, f1, K[0], in[3],  12);  @round(ccc, ddd, eee, aaa, bbb, f5, K[5], in[0],  11);
+			@round(bb, cc, dd, ee, aa, f1, K[0], in[4],   5);  @round(bbb, ccc, ddd, eee, aaa, f5, K[5], in[9],  13);
+			@round(aa, bb, cc, dd, ee, f1, K[0], in[5],   8);  @round(aaa, bbb, ccc, ddd, eee, f5, K[5], in[2],  15);
+			@round(ee, aa, bb, cc, dd, f1, K[0], in[6],   7);  @round(eee, aaa, bbb, ccc, ddd, f5, K[5], in[11], 15);
+			@round(dd, ee, aa, bb, cc, f1, K[0], in[7],   9);  @round(ddd, eee, aaa, bbb, ccc, f5, K[5], in[4],   5);
+			@round(cc, dd, ee, aa, bb, f1, K[0], in[8],  11);  @round(ccc, ddd, eee, aaa, bbb, f5, K[5], in[13],  7);
+			@round(bb, cc, dd, ee, aa, f1, K[0], in[9],  13);  @round(bbb, ccc, ddd, eee, aaa, f5, K[5], in[6],   7);
+			@round(aa, bb, cc, dd, ee, f1, K[0], in[10], 14);  @round(aaa, bbb, ccc, ddd, eee, f5, K[5], in[15],  8);
+			@round(ee, aa, bb, cc, dd, f1, K[0], in[11], 15);  @round(eee, aaa, bbb, ccc, ddd, f5, K[5], in[8],  11);
+			@round(dd, ee, aa, bb, cc, f1, K[0], in[12],  6);  @round(ddd, eee, aaa, bbb, ccc, f5, K[5], in[1],  14);
+			@round(cc, dd, ee, aa, bb, f1, K[0], in[13],  7);  @round(ccc, ddd, eee, aaa, bbb, f5, K[5], in[10], 14);
+			@round(bb, cc, dd, ee, aa, f1, K[0], in[14],  9);  @round(bbb, ccc, ddd, eee, aaa, f5, K[5], in[3],  12);
+			@round(aa, bb, cc, dd, ee, f1, K[0], in[15],  8);  @round(aaa, bbb, ccc, ddd, eee, f5, K[5], in[12],  6);
+
+			@swap(aa, aaa);
+
+			@round(ee, aa, bb, cc, dd, f2, K[1], in[7],   7);  @round(eee, aaa, bbb, ccc, ddd, f4, K[6], in[6],   9);
+			@round(dd, ee, aa, bb, cc, f2, K[1], in[4],   6);  @round(ddd, eee, aaa, bbb, ccc, f4, K[6], in[11], 13);
+			@round(cc, dd, ee, aa, bb, f2, K[1], in[13],  8);  @round(ccc, ddd, eee, aaa, bbb, f4, K[6], in[3],  15);
+			@round(bb, cc, dd, ee, aa, f2, K[1], in[1],  13);  @round(bbb, ccc, ddd, eee, aaa, f4, K[6], in[7],   7);
+			@round(aa, bb, cc, dd, ee, f2, K[1], in[10], 11);  @round(aaa, bbb, ccc, ddd, eee, f4, K[6], in[0],  12);
+			@round(ee, aa, bb, cc, dd, f2, K[1], in[6],   9);  @round(eee, aaa, bbb, ccc, ddd, f4, K[6], in[13],  8);
+			@round(dd, ee, aa, bb, cc, f2, K[1], in[15],  7);  @round(ddd, eee, aaa, bbb, ccc, f4, K[6], in[5],   9);
+			@round(cc, dd, ee, aa, bb, f2, K[1], in[3],  15);  @round(ccc, ddd, eee, aaa, bbb, f4, K[6], in[10], 11);
+			@round(bb, cc, dd, ee, aa, f2, K[1], in[12],  7);  @round(bbb, ccc, ddd, eee, aaa, f4, K[6], in[14],  7);
+			@round(aa, bb, cc, dd, ee, f2, K[1], in[0],  12);  @round(aaa, bbb, ccc, ddd, eee, f4, K[6], in[15],  7);
+			@round(ee, aa, bb, cc, dd, f2, K[1], in[9],  15);  @round(eee, aaa, bbb, ccc, ddd, f4, K[6], in[8],  12);
+			@round(dd, ee, aa, bb, cc, f2, K[1], in[5],   9);  @round(ddd, eee, aaa, bbb, ccc, f4, K[6], in[12],  7);
+			@round(cc, dd, ee, aa, bb, f2, K[1], in[2],  11);  @round(ccc, ddd, eee, aaa, bbb, f4, K[6], in[4],   6);
+			@round(bb, cc, dd, ee, aa, f2, K[1], in[14],  7);  @round(bbb, ccc, ddd, eee, aaa, f4, K[6], in[9],  15);
+			@round(aa, bb, cc, dd, ee, f2, K[1], in[11], 13);  @round(aaa, bbb, ccc, ddd, eee, f4, K[6], in[1],  13);
+			@round(ee, aa, bb, cc, dd, f2, K[1], in[8],  12);  @round(eee, aaa, bbb, ccc, ddd, f4, K[6], in[2],  11);
+
+			@swap(bb, bbb);
+
+			@round(dd, ee, aa, bb, cc, f3, K[2], in[3],  11);  @round(ddd, eee, aaa, bbb, ccc, f3, K[7], in[15],  9);
+			@round(cc, dd, ee, aa, bb, f3, K[2], in[10], 13);  @round(ccc, ddd, eee, aaa, bbb, f3, K[7], in[5],   7);
+			@round(bb, cc, dd, ee, aa, f3, K[2], in[14],  6);  @round(bbb, ccc, ddd, eee, aaa, f3, K[7], in[1],  15);
+			@round(aa, bb, cc, dd, ee, f3, K[2], in[4],   7);  @round(aaa, bbb, ccc, ddd, eee, f3, K[7], in[3],  11);
+			@round(ee, aa, bb, cc, dd, f3, K[2], in[9],  14);  @round(eee, aaa, bbb, ccc, ddd, f3, K[7], in[7],   8);
+			@round(dd, ee, aa, bb, cc, f3, K[2], in[15],  9);  @round(ddd, eee, aaa, bbb, ccc, f3, K[7], in[14],  6);
+			@round(cc, dd, ee, aa, bb, f3, K[2], in[8],  13);  @round(ccc, ddd, eee, aaa, bbb, f3, K[7], in[6],   6);
+			@round(bb, cc, dd, ee, aa, f3, K[2], in[1],  15);  @round(bbb, ccc, ddd, eee, aaa, f3, K[7], in[9],  14);
+			@round(aa, bb, cc, dd, ee, f3, K[2], in[2],  14);  @round(aaa, bbb, ccc, ddd, eee, f3, K[7], in[11], 12);
+			@round(ee, aa, bb, cc, dd, f3, K[2], in[7],   8);  @round(eee, aaa, bbb, ccc, ddd, f3, K[7], in[8],  13);
+			@round(dd, ee, aa, bb, cc, f3, K[2], in[0],  13);  @round(ddd, eee, aaa, bbb, ccc, f3, K[7], in[12],  5);
+			@round(cc, dd, ee, aa, bb, f3, K[2], in[6],   6);  @round(ccc, ddd, eee, aaa, bbb, f3, K[7], in[2],  14);
+			@round(bb, cc, dd, ee, aa, f3, K[2], in[13],  5);  @round(bbb, ccc, ddd, eee, aaa, f3, K[7], in[10], 13);
+			@round(aa, bb, cc, dd, ee, f3, K[2], in[11], 12);  @round(aaa, bbb, ccc, ddd, eee, f3, K[7], in[0],  13);
+			@round(ee, aa, bb, cc, dd, f3, K[2], in[5],   7);  @round(eee, aaa, bbb, ccc, ddd, f3, K[7], in[4],   7);
+			@round(dd, ee, aa, bb, cc, f3, K[2], in[12],  5);  @round(ddd, eee, aaa, bbb, ccc, f3, K[7], in[13],  5);
+
+			@swap(cc, ccc);
+
+			@round(cc, dd, ee, aa, bb, f4, K[3], in[1],  11);  @round(ccc, ddd, eee, aaa, bbb, f2, K[8], in[8],  15);
+			@round(bb, cc, dd, ee, aa, f4, K[3], in[9],  12);  @round(bbb, ccc, ddd, eee, aaa, f2, K[8], in[6],   5);
+			@round(aa, bb, cc, dd, ee, f4, K[3], in[11], 14);  @round(aaa, bbb, ccc, ddd, eee, f2, K[8], in[4],   8);
+			@round(ee, aa, bb, cc, dd, f4, K[3], in[10], 15);  @round(eee, aaa, bbb, ccc, ddd, f2, K[8], in[1],  11);
+			@round(dd, ee, aa, bb, cc, f4, K[3], in[0],  14);  @round(ddd, eee, aaa, bbb, ccc, f2, K[8], in[3],  14);
+			@round(cc, dd, ee, aa, bb, f4, K[3], in[8],  15);  @round(ccc, ddd, eee, aaa, bbb, f2, K[8], in[11], 14);
+			@round(bb, cc, dd, ee, aa, f4, K[3], in[12],  9);  @round(bbb, ccc, ddd, eee, aaa, f2, K[8], in[15],  6);
+			@round(aa, bb, cc, dd, ee, f4, K[3], in[4],   8);  @round(aaa, bbb, ccc, ddd, eee, f2, K[8], in[0],  14);
+			@round(ee, aa, bb, cc, dd, f4, K[3], in[13],  9);  @round(eee, aaa, bbb, ccc, ddd, f2, K[8], in[5],   6);
+			@round(dd, ee, aa, bb, cc, f4, K[3], in[3],  14);  @round(ddd, eee, aaa, bbb, ccc, f2, K[8], in[12],  9);
+			@round(cc, dd, ee, aa, bb, f4, K[3], in[7],   5);  @round(ccc, ddd, eee, aaa, bbb, f2, K[8], in[2],  12);
+			@round(bb, cc, dd, ee, aa, f4, K[3], in[15],  6);  @round(bbb, ccc, ddd, eee, aaa, f2, K[8], in[13],  9);
+			@round(aa, bb, cc, dd, ee, f4, K[3], in[14],  8);  @round(aaa, bbb, ccc, ddd, eee, f2, K[8], in[9],  12);
+			@round(ee, aa, bb, cc, dd, f4, K[3], in[5],   6);  @round(eee, aaa, bbb, ccc, ddd, f2, K[8], in[7],   5);
+			@round(dd, ee, aa, bb, cc, f4, K[3], in[6],   5);  @round(ddd, eee, aaa, bbb, ccc, f2, K[8], in[10], 15);
+			@round(cc, dd, ee, aa, bb, f4, K[3], in[2],  12);  @round(ccc, ddd, eee, aaa, bbb, f2, K[8], in[14],  8);
+
+			@swap(dd, ddd);
+
+			@round(bb, cc, dd, ee, aa, f5, K[4], in[4],   9);  @round(bbb, ccc, ddd, eee, aaa, f1, 0, in[12],  8);
+			@round(aa, bb, cc, dd, ee, f5, K[4], in[0],  15);  @round(aaa, bbb, ccc, ddd, eee, f1, 0, in[15],  5);
+			@round(ee, aa, bb, cc, dd, f5, K[4], in[5],   5);  @round(eee, aaa, bbb, ccc, ddd, f1, 0, in[10], 12);
+			@round(dd, ee, aa, bb, cc, f5, K[4], in[9],  11);  @round(ddd, eee, aaa, bbb, ccc, f1, 0, in[4],   9);
+			@round(cc, dd, ee, aa, bb, f5, K[4], in[7],   6);  @round(ccc, ddd, eee, aaa, bbb, f1, 0, in[1],  12);
+			@round(bb, cc, dd, ee, aa, f5, K[4], in[12],  8);  @round(bbb, ccc, ddd, eee, aaa, f1, 0, in[5],   5);
+			@round(aa, bb, cc, dd, ee, f5, K[4], in[2],  13);  @round(aaa, bbb, ccc, ddd, eee, f1, 0, in[8],  14);
+			@round(ee, aa, bb, cc, dd, f5, K[4], in[10], 12);  @round(eee, aaa, bbb, ccc, ddd, f1, 0, in[7],   6);
+			@round(dd, ee, aa, bb, cc, f5, K[4], in[14],  5);  @round(ddd, eee, aaa, bbb, ccc, f1, 0, in[6],   8);
+			@round(cc, dd, ee, aa, bb, f5, K[4], in[1],  12);  @round(ccc, ddd, eee, aaa, bbb, f1, 0, in[2],  13);
+			@round(bb, cc, dd, ee, aa, f5, K[4], in[3],  13);  @round(bbb, ccc, ddd, eee, aaa, f1, 0, in[13],  6);
+			@round(aa, bb, cc, dd, ee, f5, K[4], in[8],  14);  @round(aaa, bbb, ccc, ddd, eee, f1, 0, in[14],  5);
+			@round(ee, aa, bb, cc, dd, f5, K[4], in[11], 11);  @round(eee, aaa, bbb, ccc, ddd, f1, 0, in[0],  15);
+			@round(dd, ee, aa, bb, cc, f5, K[4], in[6],   8);  @round(ddd, eee, aaa, bbb, ccc, f1, 0, in[3],  13);
+			@round(cc, dd, ee, aa, bb, f5, K[4], in[15],  5);  @round(ccc, ddd, eee, aaa, bbb, f1, 0, in[9],  11);
+			@round(bb, cc, dd, ee, aa, f5, K[4], in[13],  6);  @round(bbb, ccc, ddd, eee, aaa, f1, 0, in[11], 11);
+
+			@swap(ee, eee);
+
+			state[0] += aa;
+			state[1] += bb;
+			state[2] += cc;
+			state[3] += dd;
+			state[4] += ee;
+			state[5] += aaa;
+			state[6] += bbb;
+			state[7] += ccc;
+			state[8] += ddd;
+			state[9] += eee;
+
+		$default:
+			$error "Invalid digest size";
+	$endswitch
+}

lib/std/hash/sha256.c3

@@ -1,177 +1,175 @@
 module std::hash::sha256;
-
-import std::hash::hmac;
+import std::bits, 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;
-}
-
 alias HmacSha256 = Hmac{Sha256, HASH_SIZE, BLOCK_SIZE};
 alias hmac = hmac::hash{Sha256, HASH_SIZE, BLOCK_SIZE};
 alias pbkdf2 = hmac::pbkdf2{Sha256, HASH_SIZE, BLOCK_SIZE};
 
-fn char[HASH_SIZE] hash(char[] data)
+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
+};
+
+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 => x.rotr(2) ^ x.rotr(13) ^ x.rotr(22);
+macro uint @_sigma1(uint x) @local => x.rotr(6) ^ x.rotr(11) ^ x.rotr(25);
+macro uint @sigma0(uint x) @local => x.rotr(7) ^ x.rotr(18) ^ (x >> 3);
+macro uint @sigma1(uint x) @local => x.rotr(17) ^ x.rotr(19) ^ (x >> 10);
+
+struct Sha256
 {
-    Sha256 sha256 @noinit;
-    sha256.init();
-    sha256.update(data);
-    return sha256.final();
+	uint[8] state @align(usz.sizeof);
+	char[BLOCK_SIZE] buffer @align(ulong.sizeof);   // must align along bitcount sizeof - see `final`
+	ulong bitcount;
 }
 
-fn void Sha256.init(&self)
+<*
+ Compute and return a hash value.
+
+ @param [in] data : "The input data to hash."
+*>
+fn char[HASH_SIZE] hash(char[] data)
 {
-    // Sha256 initialization constants
-    *self = {
-        .state = {
-            0x6A09E667,
-            0xBB67AE85,
-            0x3C6EF372,
-            0xA54FF53A,
-            0x510E527F,
-            0x9B05688C,
-            0x1F83D9AB,
-            0x5BE0CD19
-        }
-    };
+	Sha256 sha256 @noinit;
+	sha256.init();
+	sha256.update(data);
+	return sha256.final();
 }
 
+fn void Sha256.init(&self) => *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);
+fn void Sha256.update(&self, char[] data)
+{
+	uint buffer_pos = (uint)(self.bitcount >> 3) % BLOCK_SIZE;
+	self.bitcount += (ulong)data.len << 3; // always record ingested bits count immediately
 
-    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
-        }
-    }
+	// Get the buffer position back to 0 if we're midway through consuming some data.
+	if (buffer_pos > 0 && buffer_pos < BLOCK_SIZE)
+	{
+		usz len = min(BLOCK_SIZE - buffer_pos, data.len);
+		self.buffer[buffer_pos:len] = data[:len];
+		data = data[len..];
+		if (buffer_pos + len == BLOCK_SIZE) self.transform();
+	}
+
+	// When the data pointer is aligned, we can disregard unaligned loading in the `transform` macro.
+	//   We do this here from the outer call to reduce the expense of checking alignment on every single block.
+	if (0 == (usz)data.ptr % usz.sizeof)
+	{
+		for (; data.len >= BLOCK_SIZE; data = data[BLOCK_SIZE..]) self.transform((uint*)data.ptr);
+	}
+	else
+	{
+		for (; data.len >= BLOCK_SIZE; data = data[BLOCK_SIZE..]) self.transform_unaligned((uint*)data.ptr);
+	}
+
+	// Leftover data just gets stored away for the next update or final.
+	if (data.len)
+	{
+		self.buffer[..] = 0;
+		self.buffer[:data.len] = data[..];
+	}
 }
 
-fn char[HASH_SIZE] Sha256.final(&self) {
-    char[HASH_SIZE] hash;
-    ulong i = (self.bitcount / 8) % BLOCK_SIZE;
+fn char[HASH_SIZE] Sha256.final(&self)
+{
+	char[HASH_SIZE] hash @align(uint.sizeof);
+	ulong i = (self.bitcount / 8) % BLOCK_SIZE;
 
-    // Append 0x80 to the buffer
-    self.buffer[i++] = 0x80;
+	// 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;
-    }
+	// Pad the buffer with zeros
+	if (i > BLOCK_SIZE - 8)
+	{
+		self.buffer[i..] = 0x00;
+		self.transform();
+		i = 0;  // Reset buffer index after transformation
+	}
 
-    // 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);
-    }
+	self.buffer[i..(BLOCK_SIZE - 8)] = 0x00;
 
-    sha256_transform(&self.state, &self.buffer);
+	// Append the bitcount in big-endian format
+	*(ulong*)(&self.buffer[BLOCK_SIZE - 8]) = env::BIG_ENDIAN ??? self.bitcount : bswap(self.bitcount);
 
-    // 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;
+	self.transform();
+
+	// Convert state to the final hash
+	foreach (x, s : self.state) *(uint*)(&hash[x * uint.sizeof]) = env::BIG_ENDIAN ??? s : bswap(s);
+	
+	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;
+// These wrappers are necessary to significantly reduce code generation from macro expansions.
+//   Note that transformations on `self.buffer` (when incoming == null) should always be aligned.
+fn void Sha256.transform(&self, uint* incoming = null) @local @noinline => self.do_transform(incoming, true);
+fn void Sha256.transform_unaligned(&self, uint* incoming = null) @local @noinline => self.do_transform(incoming, false);
 
-    // 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];
-    }
+macro Sha256.do_transform(&self, uint* incoming = null, bool $aligned = true) @local
+{
+	uint a, b, c, d, e, f, g, h, t1, t2 @noinit;
+	uint[64] m @noinit;
+	int i @noinit;
 
-    // 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];
+	if (!incoming) incoming = (uint*)&self.buffer;
 
-    // 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;
-    }
+	$if env::BIG_ENDIAN:
+		@as_char_view(m)[:BLOCK_SIZE] = @as_char_view(incoming)[:BLOCK_SIZE];
+	$else
+		// Unrolling this seems to make the hash slower.
+		for (i = 0; i < 16; ++i) m[i] = bswap($aligned ??? incoming[i] : @unaligned_load(incoming[i], 1));
+	$endif
 
-    // 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;
+	for (i = 16; i < 64; i++) m[i] = @sigma1(m[i - 2]) + m[i - 7] + @sigma0(m[i - 15]) + m[i - 16];
+
+	a = self.state[0];
+	b = self.state[1];
+	c = self.state[2];
+	d = self.state[3];
+	e = self.state[4];
+	f = self.state[5];
+	g = self.state[6];
+	h = self.state[7];
+
+	$for usz $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;
+	$endfor
+
+	self.state[0] += a;
+	self.state[1] += b;
+	self.state[2] += c;
+	self.state[3] += d;
+	self.state[4] += e;
+	self.state[5] += f;
+	self.state[6] += g;
+	self.state[7] += h;
 }
 

lib/std/hash/siphash.c3

@@ -48,7 +48,7 @@ alias hash = siphash::hash { uint128, 4, 8 };
 
  @require OutType.typeid == uint128.typeid || OutType.typeid == ulong.typeid : "Module OutType must be either uint128 or ulong."
 *>
-module std::hash::siphash { OutType, BLOCK_ROUNDS, FINALIZE_ROUNDS };
+module std::hash::siphash <OutType, BLOCK_ROUNDS, FINALIZE_ROUNDS>;
 
 
 struct SipHash

lib/std/hash/whirlpool/whirlpool.c3

@@ -161,7 +161,7 @@ fn void Whirlpool.process_block(&self, char* block) @local
 	// NOTE: These loops are unrolled with C3's Chad-tier compile-time evaluation.
 	$for var $round = 0; $round < 8; $round++:
 		k[$round] = self.hash[$round];
-		state[$round] = $$bswap(@unaligned_load(((ulong*)block)[$round], 1)) ^ self.hash[$round];
+		state[$round] = $$bswap(mem::load((ulong*)block + $round, 1)) ^ self.hash[$round];
 		self.hash[$round] = state[$round];
 	$endfor
 

lib/std/hash/wyhash2.c3

@@ -34,8 +34,8 @@ fn ulong hash(char[] input, ulong seed = 0)
 	{
 		if (@likely(input.len >= 4))
 		{
-			a = (ulong)@unaligned_load(*(uint*)input.ptr, 1);   // first 4 bytes widened to a u64
-			b = (ulong)@unaligned_load(*(uint*)&input[^4], 1);   // a walking 4-byte window based on input.len
+			a = (ulong)mem::load((uint*)input.ptr, 1);   // first 4 bytes widened to a u64
+			b = (ulong)mem::load((uint*)&input[^4], 1);   // a walking 4-byte window based on input.len
 		}
 		else if (input.len > 0)
 		{
@@ -44,8 +44,8 @@ fn ulong hash(char[] input, ulong seed = 0)
 	}
 	else
 	{
-		a = @unaligned_load(*(ulong*)input.ptr, 1);   // first 8 bytes
-		b = @unaligned_load(*(ulong*)&input[^8], 1);   // a walking 8-byte window based on input.len
+		a = mem::load((ulong*)input.ptr, 1);   // first 8 bytes
+		b = mem::load((ulong*)&input[^8], 1);   // a walking 8-byte window based on input.len
 	}
 
 	uint128 r = ((uint128)a ^ 0xe703_7ed1_a0b4_28db) * ((uint128)b ^ seed);

lib/std/io/file.c3

@@ -102,8 +102,8 @@ fn void? File.close(&self) @inline @dynamic
 		switch (libc::errno())
 		{
 			case errno::ECONNRESET:
-			case errno::EBADF: return io::FILE_NOT_VALID?;
-			case errno::EINTR: return io::INTERRUPTED?;
+			case errno::EBADF: return io::FILE_NOT_VALID~;
+			case errno::EINTR: return io::INTERRUPTED~;
 			case errno::EDQUOT:
 			case errno::EFAULT:
 			case errno::EAGAIN:
@@ -111,8 +111,8 @@ fn void? File.close(&self) @inline @dynamic
 			case errno::ENETDOWN:
 			case errno::ENETUNREACH:
 			case errno::ENOSPC:
-			case errno::EIO: return io::INCOMPLETE_WRITE?;
-			default: return io::UNKNOWN_ERROR?;
+			case errno::EIO: return io::INCOMPLETE_WRITE~;
+			default: return io::UNKNOWN_ERROR~;
 		}
 	}
 	self.file = null;
@@ -156,7 +156,7 @@ fn bool File.isatty(self) @if(env::LIBC)
 fn char? File.read_byte(&self) @dynamic
 {
 	int c = libc::fgetc(self.file);
-	if (c == -1) return io::EOF?;
+	if (c == -1) return io::EOF~;
 	return (char)c;
 }
 
@@ -172,7 +172,7 @@ 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 io::OVERFLOW?;
+	if (len > buffer.len) return io::OVERFLOW~;
 	file.seek(0, SET)!;
 	usz read = 0;
 	while (read < len)

lib/std/io/file_mmap.c3

@@ -26,8 +26,8 @@ fn void? FileMmap.destroy(&self) @maydiscard
 {
 	fault err1 = @catch(self.file.close());
 	fault err2 = @catch(self.vm.destroy());
-	if (err1) return err1?;
-	if (err2) return err2?;
+	if (err1) return err1~;
+	if (err2) return err2~;
 }
 
 module std::io::file @if(env::LIBC &&& env::POSIX);

lib/std/io/formatter.c3

@@ -2,6 +2,9 @@ module std::io;
 import std::collections::map;
 import libc;
 
+// It is vitally important that the formatter doesn't do any allocations,
+// or it ceases to be useful in constrained environments.
+
 const int PRINTF_NTOA_BUFFER_SIZE = 256;
 
 interface Printable
@@ -93,9 +96,9 @@ 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?;
+		if (self.first_fault) return self.first_fault~;
 		self.first_fault = err;
-		return err?;
+		return err~;
 	}
 	return 1;
 }
@@ -130,7 +133,7 @@ fn usz? Formatter.print_with_function(&self, Printable arg)
 		if (!arg) return self.out_substr("(null)");
 		return self.out_substr(arg.to_constant_string());
 	}
-	return NOT_FOUND?;
+	return NOT_FOUND~;
 }
 
 fn usz? Formatter.out_unknown(&self, String category, any arg) @private
@@ -337,7 +340,7 @@ 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 INVALID_FORMAT?;
+	return INVALID_FORMAT~;
 }
 
 macro usz? @wrap_bad(Formatter* f, #action)
@@ -349,11 +352,11 @@ macro usz? @wrap_bad(Formatter* f, #action)
 		{
 			case BUFFER_EXCEEDED:
         	case INTERNAL_BUFFER_EXCEEDED:
-        		return f.first_err(err)?;
+        		return f.first_err(err)~;
         	default:
 				err = f.first_err(INVALID_ARGUMENT);
         		f.out_substr("<INVALID>")!;
-        		return err?;
+        		return err~;
 		}
 	}
 	return len;
@@ -561,7 +564,7 @@ fn usz? Formatter.vprintf(&self, String format, any[] anys)
 	//	out((char)0, buffer, idx < maxlen ? idx : maxlen - 1U, maxlen);
 
 	// return written chars without terminating \0
-	if (self.first_fault) return self.first_fault?;
+	if (self.first_fault) return self.first_fault~;
 	return total_len;
 }
 

lib/std/io/formatter_private.c3

@@ -86,7 +86,7 @@ fn uint128? int_from_any(any arg, bool *is_neg) @private
 			double d = *(double*)arg;
 			return (uint128)((*is_neg = d < 0) ? -d : d);
 		default:
-			return BAD_FORMAT?;
+			return BAD_FORMAT~;
 	}
 }
 
@@ -128,7 +128,7 @@ fn FloatType? float_from_any(any arg) @private
 		case double:
 			return (FloatType)*(double*)arg;
 		default:
-			return BAD_FORMAT?;
+			return BAD_FORMAT~;
 	}
 }
 
@@ -283,7 +283,7 @@ fn usz? Formatter.floatformat(&self, FloatFormatting formatting, double y) @priv
 		} while (y);
 		isz outlen = s - buf;
 		isz explen = ebuf - estr;
-		if (p > int.max - 2 - explen - pl) return INTERNAL_BUFFER_EXCEEDED?;
+		if (p > int.max - 2 - explen - pl) return INTERNAL_BUFFER_EXCEEDED~;
 		usz len;
 		usz l = p && outlen - 2 < p
 				? p + 2 + explen
@@ -364,7 +364,10 @@ fn usz? Formatter.floatformat(&self, FloatFormatting formatting, double y) @priv
 	}
 
 	// Perform rounding: j is precision after the radix (possibly neg)
-	int j = (int)(p - (isz)(formatting == FLOAT ? 0 : e - (int)(formatting == ADAPTIVE && p)));
+	int j = (int)p;
+	if (formatting != FLOAT) j -= e;
+	if (formatting == ADAPTIVE && p != 0) j -= 1;
+
 	if (j < 9 * (z - r - 1))
 	{
 		uint x;
@@ -451,12 +454,12 @@ fn usz? Formatter.floatformat(&self, FloatFormatting formatting, double y) @priv
 			}
 		}
 	}
-	if (p > int.max - 1 - (isz)(p || self.flags.hash)) return INTERNAL_BUFFER_EXCEEDED?;
+	if (p > int.max - 1 - (isz)(p || self.flags.hash)) return INTERNAL_BUFFER_EXCEEDED~;
 	int l = (int)(1 + p + (isz)(p || self.flags.hash));
 	char* estr @noinit;
 	if (formatting == FLOAT)
 	{
-		if (e > int.max - l) return INTERNAL_BUFFER_EXCEEDED?;
+		if (e > int.max - l) return INTERNAL_BUFFER_EXCEEDED~;
 		if (e > 0) l += e;
 	}
 	else
@@ -465,10 +468,10 @@ fn usz? Formatter.floatformat(&self, FloatFormatting formatting, double y) @priv
 		while (ebuf - estr < 2) (--estr)[0] = '0';
 		*--estr = (e < 0 ? '-' : '+');
 		*--estr = self.flags.uppercase ? 'E' : 'e';
-		if (ebuf - estr > (isz)int.max - l) return INTERNAL_BUFFER_EXCEEDED?;
+		if (ebuf - estr > (isz)int.max - l) return INTERNAL_BUFFER_EXCEEDED~;
 		l += (int)(ebuf - estr);
 	}
-	if (l > int.max - pl) return INTERNAL_BUFFER_EXCEEDED?;
+	if (l > int.max - pl) return INTERNAL_BUFFER_EXCEEDED~;
 	usz len;
 	if (!self.flags.left && !self.flags.zeropad) len += self.pad(' ', self.width, pl + l)!;
 	if (is_neg || self.flags.plus) len += self.out(is_neg ? '-' : '+')!;
@@ -556,7 +559,7 @@ fn usz? Formatter.ntoa(&self, uint128 value, bool negative, uint base) @private
 			case 2:
 				do
 				{
-					if (len >= PRINTF_NTOA_BUFFER_SIZE) return INTERNAL_BUFFER_EXCEEDED?;
+					if (len >= PRINTF_NTOA_BUFFER_SIZE) return INTERNAL_BUFFER_EXCEEDED~;
 					buf[len++] = '0' + (char)value & 1;
 					value >>= 1;
 				}
@@ -564,13 +567,13 @@ fn usz? Formatter.ntoa(&self, uint128 value, bool negative, uint base) @private
 			case 10:
 				if (!value)
 				{
-					if (len >= PRINTF_NTOA_BUFFER_SIZE) return INTERNAL_BUFFER_EXCEEDED?;
+					if (len >= PRINTF_NTOA_BUFFER_SIZE) return INTERNAL_BUFFER_EXCEEDED~;
 					buf[len++] = '0';
 					break;
 				}
 				while (value >= 10)
 				{
-					if (len + 1 >= PRINTF_NTOA_BUFFER_SIZE) return INTERNAL_BUFFER_EXCEEDED?;
+					if (len + 1 >= PRINTF_NTOA_BUFFER_SIZE) return INTERNAL_BUFFER_EXCEEDED~;
 					char digit = (char)(value % 100);
 					buf[len:2] = DIGIT_PAIRS[2 * digit:2];
 					len += 2;
@@ -578,13 +581,13 @@ fn usz? Formatter.ntoa(&self, uint128 value, bool negative, uint base) @private
 				}
 				if (value > 0)
 				{
-					if (len >= PRINTF_NTOA_BUFFER_SIZE) return INTERNAL_BUFFER_EXCEEDED?;
+					if (len >= PRINTF_NTOA_BUFFER_SIZE) return INTERNAL_BUFFER_EXCEEDED~;
 					buf[len++] = '0' + (char)value;
 				}
 			case 16:
 				do
 				{
-					if (len >= PRINTF_NTOA_BUFFER_SIZE) return INTERNAL_BUFFER_EXCEEDED?;
+					if (len >= PRINTF_NTOA_BUFFER_SIZE) return INTERNAL_BUFFER_EXCEEDED~;
 					char digit = (char)value & 0xF;
 					buf[len++] = digit + (digit < 10 ? '0' : past_10);
 					value >>= 4;
@@ -593,7 +596,7 @@ fn usz? Formatter.ntoa(&self, uint128 value, bool negative, uint base) @private
 			case 8:
 				do
 				{
-					if (len >= PRINTF_NTOA_BUFFER_SIZE) return INTERNAL_BUFFER_EXCEEDED?;
+					if (len >= PRINTF_NTOA_BUFFER_SIZE) return INTERNAL_BUFFER_EXCEEDED~;
 					buf[len++] = '0' + (char)value & 0x7;
 					value >>= 3;
 				}
@@ -613,12 +616,12 @@ fn usz? Formatter.ntoa_format(&self, String buf, usz len, bool negative, uint ba
 		if (self.width && self.flags.zeropad && (negative || self.flags.plus || self.flags.space)) self.width--;
 		while (len < self.prec)
 		{
-			if (len >= buf.len) return INTERNAL_BUFFER_EXCEEDED?;
+			if (len >= buf.len) return INTERNAL_BUFFER_EXCEEDED~;
 			buf[len++] = '0';
 		}
 		while (self.flags.zeropad && len < self.width)
 		{
-			if (len >= buf.len) return INTERNAL_BUFFER_EXCEEDED?;
+			if (len >= buf.len) return INTERNAL_BUFFER_EXCEEDED~;
 			buf[len++] = '0';
 		}
 	}
@@ -633,7 +636,7 @@ fn usz? Formatter.ntoa_format(&self, String buf, usz len, bool negative, uint ba
 		}
 		if (base != 10)
 		{
-			if (len + 1 >= buf.len) return INTERNAL_BUFFER_EXCEEDED?;
+			if (len + 1 >= buf.len) return INTERNAL_BUFFER_EXCEEDED~;
 			switch (base)
 			{
 				case 16:
@@ -652,13 +655,13 @@ fn usz? Formatter.ntoa_format(&self, String buf, usz len, bool negative, uint ba
 	switch (true)
 	{
 		case negative:
-			if (len >= buf.len) return INTERNAL_BUFFER_EXCEEDED?;
+			if (len >= buf.len) return INTERNAL_BUFFER_EXCEEDED~;
 			buf[len++] = '-';
 		case self.flags.plus:
-			if (len >= buf.len) return INTERNAL_BUFFER_EXCEEDED?;
+			if (len >= buf.len) return INTERNAL_BUFFER_EXCEEDED~;
 			buf[len++] = '+';
 		case self.flags.space:
-			if (len >= buf.len) return INTERNAL_BUFFER_EXCEEDED?;
+			if (len >= buf.len) return INTERNAL_BUFFER_EXCEEDED~;
 			buf[len++] = ' ';
 	}
 	if (len) self.out_reverse(buf[:len])!;
@@ -738,10 +741,10 @@ fn int? printf_parse_format_field(
 	if (c.is_digit()) return simple_atoi(format_ptr, format_len, index_ptr);
 	if (c != '*') return 0;
 	usz len = ++(*index_ptr);
-	if (len >= format_len) return BAD_FORMAT?;
-	if (*args_index_ptr >= args_len) return BAD_FORMAT?;
+	if (len >= format_len) return BAD_FORMAT~;
+	if (*args_index_ptr >= args_len) return BAD_FORMAT~;
 	any val = args_ptr[(*args_index_ptr)++];
-	if (!val.type.kindof.is_int()) return BAD_FORMAT?;
+	if (!val.type.kindof.is_int()) return BAD_FORMAT~;
 	uint? intval = types::any_to_int(val, int);
-	return intval ?? BAD_FORMAT?;
+	return intval ?? BAD_FORMAT~;
 }

lib/std/io/io.c3

@@ -57,6 +57,11 @@ faultdef
  @return `The string containing the data read.`
 *>
 macro String? readline(Allocator allocator, stream = io::stdin())
+{
+	return readline_impl{$typeof(stream)}(allocator, stream);
+}
+
+fn String? readline_impl(Allocator allocator, Stream stream) <Stream> @private
 {
 	if (allocator == tmem)
 	{
@@ -99,16 +104,21 @@ macro String? treadline(stream = io::stdin())
 *>
 macro usz? readline_to_stream(out_stream, in_stream = io::stdin())
 {
-	bool $is_stream = $typeof(in_stream) == InStream;
+	return readline_to_stream_impl{$typeof(in_stream), $typeof(out_stream)}(out_stream, in_stream);
+}
+
+fn usz? readline_to_stream_impl(OStream out_stream, IStream in_stream) <IStream, OStream> @private
+{
+	bool $is_stream = IStream == InStream;
 	$if $is_stream:
-		var func = &in_stream.read_byte;
+		var func @safeinfer = &in_stream.read_byte;
 		char val = func((void*)in_stream)!;
 	$else
 		char val = in_stream.read_byte()!;
 	$endif
-	bool $is_out_stream = $typeof(out_stream) == OutStream;
+	bool $is_out_stream = OStream == OutStream;
 	$if $is_out_stream:
-		var out_func = &out_stream.write_byte;
+		var out_func @safeinfer = &out_stream.write_byte;
 	$endif
 
 	if (val == '\n') return 0;
@@ -132,7 +142,7 @@ macro usz? readline_to_stream(out_stream, in_stream = io::stdin())
 		if (catch err = c)
 		{
 			if (err == io::EOF) break;
-			return err?;
+			return err~;
 		}
 		if (c == '\r') continue;
 		if (c == '\n') break;
@@ -404,7 +414,7 @@ fn char[]? bprintf(char[] buffer, String format, args...) @maydiscard
 fn void? out_buffer_fn(void *data, char c) @private
 {
 	BufferData *buffer_data = data;
-	if (buffer_data.written >= buffer_data.buffer.len) return BUFFER_EXCEEDED?;
+	if (buffer_data.written >= buffer_data.buffer.len) return BUFFER_EXCEEDED~;
 	buffer_data.buffer[buffer_data.written++] = c;
 }
 

lib/std/io/os/chdir.c3

@@ -11,20 +11,20 @@ macro void? native_chdir(Path path)
 				{
 					switch (libc::errno())
 					{
-						case errno::EACCES: return io::NO_PERMISSION?;
-						case errno::ENAMETOOLONG: return io::NAME_TOO_LONG?;
-						case errno::ENOTDIR: return io::FILE_NOT_DIR?;
-						case errno::ENOENT: return io::FILE_NOT_FOUND?;
-						case errno::ELOOP: return io::SYMLINK_FAILED?;
-						default: return io::GENERAL_ERROR?;
+						case errno::EACCES: return io::NO_PERMISSION~;
+						case errno::ENAMETOOLONG: return io::NAME_TOO_LONG~;
+						case errno::ENOTDIR: return io::FILE_NOT_DIR~;
+						case errno::ENOENT: return io::FILE_NOT_FOUND~;
+						case errno::ELOOP: return io::SYMLINK_FAILED~;
+						default: return io::GENERAL_ERROR~;
 					}
 				}
 			$case env::WIN32:
 				// TODO improve with better error handling.
 				if (win32::setCurrentDirectoryW(path.str_view().to_temp_utf16()!!)) return;
-				return io::GENERAL_ERROR?;
+				return io::GENERAL_ERROR~;
 			$default:
-				return io::UNSUPPORTED_OPERATION?;
+				return io::UNSUPPORTED_OPERATION~;
 		$endswitch
 	};
 }

lib/std/io/os/file_libc.c3

@@ -12,7 +12,7 @@ fn void*? native_fopen(String filename, String mode) @inline => @pool()
 	$else
 		void* file = libc::fopen(filename.zstr_tcopy(), mode.zstr_tcopy());
 	$endif
-		return file ?: file_open_errno()?;
+		return file ?: file_open_errno()~;
 }
 
 fn void? native_remove(String filename) => @pool()
@@ -27,10 +27,10 @@ fn void? native_remove(String filename) => @pool()
 		switch (libc::errno())
 		{
 			case errno::ENOENT:
-				return io::FILE_NOT_FOUND?;
+				return io::FILE_NOT_FOUND~;
 			case errno::EACCES:
 			default:
-				return io::FILE_CANNOT_DELETE?;
+				return io::FILE_CANNOT_DELETE~;
 		}
 	}
 }
@@ -46,19 +46,19 @@ fn void*? native_freopen(void* file, String filename, String mode) @inline => @p
 	$else
 		file = libc::freopen(filename.zstr_tcopy(), mode.zstr_tcopy(), file);
 	$endif
-		return file ?: file_open_errno()?;
+		return file ?: file_open_errno()~;
 }
 
 fn void? native_fseek(void* file, isz offset, Seek seek_mode) @inline
 {
-	if (libc::fseek(file, (SeekIndex)offset, seek_mode.ordinal)) return file_seek_errno()?;
+	if (libc::fseek(file, (SeekIndex)offset, seek_mode.ordinal)) return file_seek_errno()~;
 }
 
 
 fn usz? native_ftell(CFile file) @inline
 {
 	long index = libc::ftell(file);
-	return index >= 0 ? (usz)index : file_seek_errno()?;
+	return index >= 0 ? (usz)index : file_seek_errno()~;
 }
 
 fn usz? native_fwrite(CFile file, char[] buffer) @inline
@@ -68,7 +68,7 @@ fn usz? native_fwrite(CFile file, char[] buffer) @inline
 
 fn void? native_fputc(CInt c, CFile stream) @inline
 {
-	if (libc::fputc(c, stream) == libc::EOF) return io::EOF?;
+	if (libc::fputc(c, stream) == libc::EOF) return io::EOF~;
 }
 
 fn usz? native_fread(CFile file, char[] buffer) @inline

lib/std/io/os/file_nolibc.c3

@@ -28,7 +28,7 @@ FputcFn native_fputc_fn @weak @if(!$defined(native_fputc_fn));
 fn void*? native_fopen(String filename, String mode) @inline
 {
 	if (native_fopen_fn) return native_fopen_fn(filename, mode);
-	return io::UNSUPPORTED_OPERATION?;
+	return io::UNSUPPORTED_OPERATION~;
 }
 
 <*
@@ -39,7 +39,7 @@ fn void*? native_fopen(String filename, String mode) @inline
 fn void? native_remove(String filename) @inline
 {
 	if (native_remove_fn) return native_remove_fn(filename);
-	return io::UNSUPPORTED_OPERATION?;
+	return io::UNSUPPORTED_OPERATION~;
 }
 
 <*
@@ -49,35 +49,35 @@ fn void? native_remove(String filename) @inline
 fn void*? native_freopen(void* file, String filename, String mode) @inline
 {
 	if (native_freopen_fn) return native_freopen_fn(file, filename, mode);
-	return io::UNSUPPORTED_OPERATION?;
+	return io::UNSUPPORTED_OPERATION~;
 }
 
 fn void? native_fseek(void* file, isz offset, Seek seek_mode) @inline
 {
 	if (native_fseek_fn) return native_fseek_fn(file, offset, seek_mode);
-	return io::UNSUPPORTED_OPERATION?;
+	return io::UNSUPPORTED_OPERATION~;
 }
 
 fn usz? native_ftell(CFile file) @inline
 {
 	if (native_ftell_fn) return native_ftell_fn(file);
-	return io::UNSUPPORTED_OPERATION?;
+	return io::UNSUPPORTED_OPERATION~;
 }
 
 fn usz? native_fwrite(CFile file, char[] buffer) @inline
 {
 	if (native_fwrite_fn) return native_fwrite_fn(file, buffer);
-	return io::UNSUPPORTED_OPERATION?;
+	return io::UNSUPPORTED_OPERATION~;
 }
 
 fn usz? native_fread(CFile file, char[] buffer) @inline
 {
 	if (native_fread_fn) return native_fread_fn(file, buffer);
-	return io::UNSUPPORTED_OPERATION?;
+	return io::UNSUPPORTED_OPERATION~;
 }
 
 fn void? native_fputc(CInt c, CFile stream) @inline
 {
 	if (native_fputc_fn) return native_fputc_fn(c, stream);
-	return io::UNSUPPORTED_OPERATION?;
+	return io::UNSUPPORTED_OPERATION~;
 }

lib/std/io/os/fileinfo.c3

@@ -14,25 +14,25 @@ fn void? native_stat(Stat* stat, String path) @if(env::DARWIN || env::LINUX || e
 		switch (libc::errno())
 		{
 			case errno::EBADF:
-				return io::FILE_NOT_VALID?;
+				return io::FILE_NOT_VALID~;
 			case errno::EFAULT:
 				unreachable("Invalid stat");
 			case errno::EIO:
-				return io::GENERAL_ERROR?;
+				return io::GENERAL_ERROR~;
 			case errno::EACCES:
-				return io::NO_PERMISSION?;
+				return io::NO_PERMISSION~;
 			case errno::ELOOP:
-				return io::NO_PERMISSION?;
+				return io::NO_PERMISSION~;
 			case errno::ENAMETOOLONG:
-				return io::NAME_TOO_LONG?;
+				return io::NAME_TOO_LONG~;
 			case errno::ENOENT:
-				return io::FILE_NOT_FOUND?;
+				return io::FILE_NOT_FOUND~;
 			case errno::ENOTDIR:
-				return io::FILE_NOT_DIR?;
+				return io::FILE_NOT_DIR~;
 			case errno::EOVERFLOW:
-				return io::GENERAL_ERROR?;
+				return io::GENERAL_ERROR~;
 			default:
-				return io::UNKNOWN_ERROR?;
+				return io::UNKNOWN_ERROR~;
 		}
 	}
 }
@@ -47,14 +47,14 @@ fn usz? native_file_size(String path) @if(env::WIN32) => @pool()
 	return (usz)size.quadPart;
 }
 
-fn usz? native_file_size(String path) @if(!env::WIN32 && !env::DARWIN)
+fn usz? native_file_size(String path) @if(!env::WIN32 && !env::DARWIN && !env::LINUX && !env::ANDROID && !env::BSD_FAMILY)
 {
 	File f = file::open(path, "r")!;
 	defer (void)f.close();
 	return f.seek(0, Seek.END)!;
 }
 
-fn usz? native_file_size(String path) @if(env::DARWIN)
+fn usz? native_file_size(String path) @if(env::DARWIN || env::LINUX || env::ANDROID || env::BSD_FAMILY)
 {
 	Stat stat;
 	native_stat(&stat, path)!;

lib/std/io/os/getcwd.c3

@@ -12,7 +12,7 @@ macro String? getcwd(Allocator allocator)
 			defer if (free) libc::free(res);
 			if (!res)
 			{
-				if (libc::errno() != errno::ERANGE) return io::GENERAL_ERROR?;
+				if (libc::errno() != errno::ERANGE) return io::GENERAL_ERROR~;
 				res = win32::_wgetcwd(null, 0);
 				free = true;
 			}
@@ -27,7 +27,7 @@ macro String? getcwd(Allocator allocator)
 			if (!res)
 			{
 				// Improve error
-				if (libc::errno() != errno::ERANGE) return io::GENERAL_ERROR?;
+				if (libc::errno() != errno::ERANGE) return io::GENERAL_ERROR~;
 				res = posix::getcwd(null, 0);
 				free = true;
 			}
@@ -35,7 +35,7 @@ macro String? getcwd(Allocator allocator)
 			return res.copy(allocator);
 
 		$default:
-			return io::UNSUPPORTED_OPERATION?;
+			return io::UNSUPPORTED_OPERATION~;
 	$endswitch
 }
 

lib/std/io/os/ls.c3

@@ -11,7 +11,7 @@ fn PathList? native_ls(Path dir, bool no_dirs, bool no_symlinks, String mask, Al
 		directory = posix::opendir(dir.str_view() ? dir.str_view().zstr_tcopy() : (ZString)".");
 	};
 	defer if (directory) posix::closedir(directory);
-	if (!directory) return (path::is_dir(dir) ? io::CANNOT_READ_DIR : io::FILE_NOT_DIR)?;
+	if (!directory) return (path::is_dir(dir) ? io::CANNOT_READ_DIR : io::FILE_NOT_DIR)~;
 	Posix_dirent* entry;
 	while ((entry = posix::readdir(directory)))
 	{
@@ -38,7 +38,7 @@ fn PathList? native_ls(Path dir, bool no_dirs, bool no_symlinks, String mask, Al
 		WString result = dir.str_view().tconcat(`\*`).to_temp_wstring()!!;
 		Win32_WIN32_FIND_DATAW find_data;
 		Win32_HANDLE find = win32::findFirstFileW(result, &find_data);
-		if (find == win32::INVALID_HANDLE_VALUE) return io::CANNOT_READ_DIR?;
+		if (find == win32::INVALID_HANDLE_VALUE) return io::CANNOT_READ_DIR~;
 		defer win32::findClose(find);
 		do
 		{

lib/std/io/os/mkdir.c3

@@ -17,17 +17,17 @@ macro bool? native_mkdir(Path path, MkdirPermissions permissions)
 				case errno::EACCES:
 				case errno::EPERM:
 				case errno::EROFS:
-				case errno::EFAULT: return io::NO_PERMISSION?;
-				case errno::ENAMETOOLONG: return io::NAME_TOO_LONG?;
+				case errno::EFAULT: return io::NO_PERMISSION~;
+				case errno::ENAMETOOLONG: return io::NAME_TOO_LONG~;
 				case errno::EDQUOT:
-				case errno::ENOSPC: return io::OUT_OF_SPACE?;
+				case errno::ENOSPC: return io::OUT_OF_SPACE~;
 				case errno::EISDIR:
 				case errno::EEXIST: return false;
-				case errno::ELOOP: return io::SYMLINK_FAILED?;
-				case errno::ENOTDIR: return io::FILE_NOT_FOUND?;
-				case errno::ENOENT: return io::PARENT_DIR_MISSING?;
+				case errno::ELOOP: return io::SYMLINK_FAILED~;
+				case errno::ENOTDIR: return io::FILE_NOT_FOUND~;
+				case errno::ENOENT: return io::PARENT_DIR_MISSING~;
 				default:
-					return io::GENERAL_ERROR?;
+					return io::GENERAL_ERROR~;
 			}
 		$case env::WIN32:
 				// TODO security attributes
@@ -35,18 +35,18 @@ macro bool? native_mkdir(Path path, MkdirPermissions permissions)
 				switch (win32::getLastError())
 				{
 					case win32::ERROR_ACCESS_DENIED:
-						return io::NO_PERMISSION?;
+						return io::NO_PERMISSION~;
 					case win32::ERROR_DISK_FULL:
-						return io::OUT_OF_SPACE?;
+						return io::OUT_OF_SPACE~;
 					case win32::ERROR_ALREADY_EXISTS:
 						return false;
 					case win32::ERROR_PATH_NOT_FOUND:
-						return io::FILE_NOT_FOUND?;
+						return io::FILE_NOT_FOUND~;
 					default:
-						return io::GENERAL_ERROR?;
+						return io::GENERAL_ERROR~;
 				}
 		$default:
-			return io::UNSUPPORTED_OPERATION?;
+			return io::UNSUPPORTED_OPERATION~;
 	$endswitch
 	};
 }

lib/std/io/os/rmdir.c3

@@ -13,36 +13,36 @@ macro bool? native_rmdir(Path path)
 			if (!posix::rmdir(path.str_view().zstr_tcopy())) return true;
 			switch (libc::errno())
 			{
-				case errno::EBUSY: return io::BUSY?;
+				case errno::EBUSY: return io::BUSY~;
 				case errno::EACCES:
 				case errno::EPERM:
 				case errno::EROFS:
-				case errno::EFAULT: return io::NO_PERMISSION?;
-				case errno::ENAMETOOLONG: return io::NAME_TOO_LONG?;
+				case errno::EFAULT: return io::NO_PERMISSION~;
+				case errno::ENAMETOOLONG: return io::NAME_TOO_LONG~;
 				case errno::ENOTDIR:
 				case errno::ENOENT: return false;
-				case errno::ENOTEMPTY: return io::DIR_NOT_EMPTY?;
-				case errno::ELOOP: return io::SYMLINK_FAILED?;
-				default: return io::GENERAL_ERROR?;
+				case errno::ENOTEMPTY: return io::DIR_NOT_EMPTY~;
+				case errno::ELOOP: return io::SYMLINK_FAILED~;
+				default: return io::GENERAL_ERROR~;
 			}
 		$case env::WIN32:
 			if (win32::removeDirectoryW(path.str_view().to_temp_utf16()!!)) return true;
 			switch (win32::getLastError())
 			{
 				case win32::ERROR_ACCESS_DENIED:
-					return io::NO_PERMISSION?;
+					return io::NO_PERMISSION~;
 				case win32::ERROR_CURRENT_DIRECTORY:
-					return io::BUSY?;
+					return io::BUSY~;
 				case win32::ERROR_DIR_NOT_EMPTY:
-					return io::DIR_NOT_EMPTY?;
+					return io::DIR_NOT_EMPTY~;
 				case win32::ERROR_DIRECTORY:
 				case win32::ERROR_PATH_NOT_FOUND:
 					return false;
 				default:
-					return io::GENERAL_ERROR?;
+					return io::GENERAL_ERROR~;
 			}
 		$default:
-			return io::UNSUPPORTED_OPERATION?;
+			return io::UNSUPPORTED_OPERATION~;
 	$endswitch
 	};
 }

lib/std/io/os/rmtree.c3

@@ -10,7 +10,7 @@ fn void? native_rmtree(Path dir)
 	{
 		DIRPtr directory = posix::opendir(dir.str_view().zstr_tcopy());
 		defer if (directory) posix::closedir(directory);
-		if (!directory) return path::is_dir(dir) ? io::CANNOT_READ_DIR? : io::FILE_NOT_DIR?;
+		if (!directory) return path::is_dir(dir) ? io::CANNOT_READ_DIR~ : io::FILE_NOT_DIR~;
 		Posix_dirent* entry;
 		while ((entry = posix::readdir(directory)))
 		{
@@ -27,7 +27,7 @@ fn void? native_rmtree(Path dir)
 				if (libc::remove(new_path.str_view().zstr_tcopy()))
 				{
 					// TODO improve
-					return io::GENERAL_ERROR?;
+					return io::GENERAL_ERROR~;
 				}
 			};
 		}
@@ -44,7 +44,7 @@ fn void? native_rmtree(Path path)
 	String s = path.str_view().tconcat("\\*");
 	Win32_HANDLE find = win32::findFirstFileW(s.to_temp_utf16(), &find_data)!;
 
-	if (find == win32::INVALID_HANDLE_VALUE) return io::CANNOT_READ_DIR?;
+	if (find == win32::INVALID_HANDLE_VALUE) return io::CANNOT_READ_DIR~;
 	defer win32::findClose(find);
 	do
 	{

lib/std/io/os/temp_directory.c3

@@ -21,7 +21,7 @@ fn String? win32_get_known_folder_temp(Win32_REFKNOWNFOLDERID rfid) @private @if
 {
 	Win32_PWSTR path;
 	Win32_HRESULT res = win32::shGetKnownFolderPath(rfid, 0x00008000 /* KF_FLAG_CREATE */, null, &path);
-	if (res) return io::PATH_COULD_NOT_BE_FOUND?;
+	if (res) return io::PATH_COULD_NOT_BE_FOUND~;
 	return string::from_wstring(tmem, (WString)path);
 }
 
@@ -38,11 +38,11 @@ fn Path? native_home_directory(Allocator allocator) => @pool()
         $case NETBSD:
 		$case OPENBSD:
 		$case HAIKU:
-			return path::new(allocator, env::tget_var("HOME")) ?? io::PATH_COULD_NOT_BE_FOUND?;
+			return path::new(allocator, env::tget_var("HOME")) ?? io::PATH_COULD_NOT_BE_FOUND~;
 		$case WIN32:
 			return path::new(allocator, win32_get_known_folder_temp(&win32::FOLDERID_PROFILE));
 		$default:
-			return io::PATH_COULD_NOT_BE_FOUND?;
+			return io::PATH_COULD_NOT_BE_FOUND~;
 	$endswitch
 }
 
@@ -67,7 +67,7 @@ fn Path? native_user_directory(Allocator allocator, NativeSystemDir dir) => @poo
 				case TEMPLATES: return path::new(allocator, posix::xdg_user_dir_lookup(tmem, "TEMPLATES"));
 				case SAVED_GAMES:
 				case SCREENSHOTS: nextcase;
-				default: return io::PATH_COULD_NOT_BE_FOUND?;
+				default: return io::PATH_COULD_NOT_BE_FOUND~;
 			}
 		$case IOS:
 		$case MACOS:
@@ -85,7 +85,7 @@ fn Path? native_user_directory(Allocator allocator, NativeSystemDir dir) => @poo
 				case SAVED_GAMES:
 				case SCREENSHOTS:
 				case TEMPLATES: nextcase;
-				default: return io::PATH_COULD_NOT_BE_FOUND?;
+				default: return io::PATH_COULD_NOT_BE_FOUND~;
 			}
 		$case WIN32:
 			switch (dir)
@@ -100,10 +100,10 @@ fn Path? native_user_directory(Allocator allocator, NativeSystemDir dir) => @poo
 				case SCREENSHOTS: return path::new(allocator, win32_get_known_folder_temp(&win32::FOLDERID_SCREENSHOTS));
 				case TEMPLATES: return path::new(allocator, win32_get_known_folder_temp(&win32::FOLDERID_TEMPLATES));
 				case PUBLIC_SHARE: nextcase;
-				default: return io::PATH_COULD_NOT_BE_FOUND?;
+				default: return io::PATH_COULD_NOT_BE_FOUND~;
 			}
 		$default:
-			return io::PATH_COULD_NOT_BE_FOUND?;
+			return io::PATH_COULD_NOT_BE_FOUND~;
 	$endswitch
 }
 
@@ -121,15 +121,16 @@ fn Path? native_temp_directory(Allocator allocator) @if(!env::WIN32)
 fn Path? native_temp_directory(Allocator allocator) @if(env::WIN32) => @pool()
 {
 	Win32_DWORD len = win32::getTempPathW(0, null);
-	if (!len) return io::GENERAL_ERROR?;
-	Char16[] buff = mem::talloc_array(Char16, len + (usz)1);
-	if (!win32::getTempPathW(len, buff)) return io::GENERAL_ERROR?;
-	return path::new(allocator, string::tfrom_utf16(buff[:len]));
+	if (!len) return io::GENERAL_ERROR~;
+	Char16[] buff = mem::talloc_array(Char16, len);
+	Win32_DWORD res = win32::getTempPathW(len, buff);
+	if (!res) return io::GENERAL_ERROR~;
+	return path::new(allocator, string::tfrom_utf16(buff[:res]));
 }
 
 module std::io::os @if(env::NO_LIBC);
 import std::io::path;
 
-macro Path? native_home_directory(Allocator allocator) => io::PATH_COULD_NOT_BE_FOUND?;
-macro Path? native_temp_directory(Allocator allocator) => io::PATH_COULD_NOT_BE_FOUND?;
-fn Path? native_user_directory(Allocator allocator, NativeSystemDir dir) => io::PATH_COULD_NOT_BE_FOUND?;
+macro Path? native_home_directory(Allocator allocator) => io::PATH_COULD_NOT_BE_FOUND~;
+macro Path? native_temp_directory(Allocator allocator) => io::PATH_COULD_NOT_BE_FOUND~;
+fn Path? native_user_directory(Allocator allocator, NativeSystemDir dir) => io::PATH_COULD_NOT_BE_FOUND~;

lib/std/io/path.c3

@@ -86,7 +86,7 @@ fn PathList? ls(Allocator allocator, Path dir, bool no_dirs = false, bool no_sym
 	$if $defined(os::native_ls):
 		return os::native_ls(dir, no_dirs, no_symlinks, mask, allocator);
 	$else
-		return io::UNSUPPORTED_OPERATION?;
+		return io::UNSUPPORTED_OPERATION~;
 	$endif
 }
 
@@ -137,11 +137,11 @@ macro bool? rmdir(path)
 *>
 fn void? rmtree(Path path)
 {
-	if (!path.path_string.len) return INVALID_PATH?;
+	if (!path.path_string.len) return INVALID_PATH~;
 	$if $defined(os::native_rmtree):
 		return os::native_rmtree(path);
 	$else
-		return io::UNSUPPORTED_OPERATION?;
+		return io::UNSUPPORTED_OPERATION~;
 	$endif
 }
 
@@ -212,7 +212,7 @@ fn usz? start_of_base_name(String str, PathEnv path_env) @local
 		// Find the \ before "foo"
 		usz last_index = 2 + str[2..].index_of_char('\\')!;
 		// If they don't match, we're done
-		if (last_index > index) return INVALID_PATH?;
+		if (last_index > index) return INVALID_PATH~;
 		if (last_index != index) return index + 1;
 		// Otherwise just default to the volume length.
 	}
@@ -246,7 +246,7 @@ fn Path? String.to_absolute_path(self, Allocator allocator) => @pool()
 fn Path? Path.absolute(self, Allocator allocator)
 {
 	String path_str = self.str_view();
-	if (!path_str.len) return INVALID_PATH?;
+	if (!path_str.len) return INVALID_PATH~;
 	if (self.is_absolute()!) return new(allocator, path_str, self.env);
 	if (path_str == ".")
 	{
@@ -262,7 +262,7 @@ fn Path? Path.absolute(self, Allocator allocator)
 			const usz BUFFER_LEN = 4096;
 			WString buffer = (WString)mem::talloc_array(Char16, BUFFER_LEN);
 			buffer = win32::_wfullpath(buffer, path_str.to_temp_wstring()!, BUFFER_LEN);
-			if (!buffer) return INVALID_PATH?;
+			if (!buffer) return INVALID_PATH~;
 			return { string::from_wstring(allocator, buffer), WIN32, allocator };
 		};
 	$else
@@ -371,10 +371,10 @@ fn usz? volume_name_len(String path, PathEnv path_env) @local
 					base_found = i;
 					continue;
 				}
-				if (is_reserved_win32_path_char(c)) return INVALID_PATH?;
+				if (is_reserved_win32_path_char(c)) return INVALID_PATH~;
 			}
 			if (base_found > 0 && base_found + 1 < len) return len;
-			return INVALID_PATH?;
+			return INVALID_PATH~;
 		case 'A'..'Z':
 		case 'a'..'z':
 			return path[1] == ':' ? 2 : 0;
@@ -392,7 +392,7 @@ fn usz? volume_name_len(String path, PathEnv path_env) @local
 *>
 fn Path? Path.parent(self)
 {
-	if (self.path_string.len == 1 && is_separator(self.path_string[0], self.env)) return NO_PARENT?;
+	if (self.path_string.len == 1 && is_separator(self.path_string[0], self.env)) return NO_PARENT~;
 	foreach_r(i, c : self.path_string)
 	{
 		if (is_separator(c, self.env))
@@ -405,7 +405,7 @@ fn Path? Path.parent(self)
 					// Handle C:\foo
 					if (volume_len == i)
 					{
-						if (i + 1 == self.path_string.len) return NO_PARENT?;
+						if (i + 1 == self.path_string.len) return NO_PARENT~;
 						return { self.path_string[:i + 1], WIN32, null };
 					}
 				}
@@ -413,7 +413,7 @@ fn Path? Path.parent(self)
 			return { self.path_string[:i], self.env, null };
 		}
 	}
-	return NO_PARENT?;
+	return NO_PARENT~;
 }
 
 fn String? normalize(String path_str, PathEnv path_env = DEFAULT_ENV)
@@ -455,7 +455,7 @@ fn String? normalize(String path_str, PathEnv path_env = DEFAULT_ENV)
 
 		// The rest are names of the path elements, so check that the
 		// characters are valid.
-		if (is_reserved_path_char(c, path_env)) return INVALID_PATH?;
+		if (is_reserved_path_char(c, path_env)) return INVALID_PATH~;
 
 		// If we have '.' after a separator
 		if (c == '.' && previous_was_separator)
@@ -488,7 +488,7 @@ fn String? normalize(String path_str, PathEnv path_env = DEFAULT_ENV)
 					continue;
 				case 2:
 					// This is an error: /a/../..
-					if (len == path_start && has_root) return INVALID_PATH?;
+					if (len == path_start && has_root) return INVALID_PATH~;
 
 					// If this .. at the start, or after ../? If so, we just copy ..
 					if (len == path_start ||
@@ -657,21 +657,21 @@ macro bool is_reserved_path_char(char c, PathEnv path_env = DEFAULT_ENV)
 }
 fn bool? _mkdir(Path path, bool recursive = false, MkdirPermissions permissions = NORMAL) @private
 {
-	if (!path.path_string.len) return INVALID_PATH?;
+	if (!path.path_string.len) return INVALID_PATH~;
 	if (is_dir(path)) return false;
-	if (exists(path)) return io::FILE_NOT_DIR?;
+	if (exists(path)) return io::FILE_NOT_DIR~;
 
 	if (recursive)
 	{
 		if (try parent = path.parent()) mkdir(parent, true, permissions)!;
 	}
-	if (!is_dir(path.parent()) ?? false) return io::CANNOT_READ_DIR?;
+	if (!is_dir(path.parent()) ?? false) return io::CANNOT_READ_DIR~;
 
 	return os::native_mkdir(path, permissions);
 }
 
 fn bool? _rmdir(Path path) @private
 {
-	if (!path.path_string.len) return INVALID_PATH?;
+	if (!path.path_string.len) return INVALID_PATH~;
 	return os::native_rmdir(path);
 }

lib/std/io/stream.c3

@@ -1,7 +1,5 @@
 module std::io;
 import std::math;
-import std::core::env;
-
 
 interface InStream
 {
@@ -36,7 +34,7 @@ fn usz? available(InStream s)
 		s.seek(curr, Seek.SET)!;
 		return len - curr;
 	}
-	return 0;
+	return io::UNSUPPORTED_OPERATION~;
 }
 
 macro bool @is_instream(#expr) @const
@@ -85,26 +83,36 @@ macro usz? read_all(stream, char[] buffer)
 {
 	if (buffer.len == 0) return 0;
 	usz n = stream.read(buffer)!;
-	if (n != buffer.len) return UNEXPECTED_EOF?;
+	if (n != buffer.len) return UNEXPECTED_EOF~;
 	return n;
 }
 
 <*
+ This function will read to the end of the stream.
+
  @require @is_instream(stream)
 *>
 macro char[]? read_fully(Allocator allocator, stream)
 {
-	usz len = available(stream)!;
-	char* data = allocator::malloc_try(allocator, len)!;
-	defer catch allocator::free(allocator, data);
-	usz read = 0;
-	while (read < len)
+	// Efficient path if it is possible to pre-allocate
+	if (try len = available(stream))
 	{
-		read += stream.read(data[read:len - read])!;
+		char* data = allocator::malloc_try(allocator, len)!;
+		defer catch allocator::free(allocator, data);
+		usz read = 0;
+		while (read < len)
+		{
+			read += stream.read(data[read:len - read])!;
+		}
+		return data[:len];
 	}
-	return data[:len];
+	ByteWriter writer;
+	writer.init(allocator);
+	copy_to(stream, &writer)!;
+	return writer.array_view();
 }
 
+
 <*
  @require @is_outstream(stream)
 *>
@@ -112,7 +120,7 @@ macro usz? write_all(stream, char[] buffer)
 {
 	if (buffer.len == 0) return 0;
 	usz n = stream.write(buffer)!;
-	if (n != buffer.len) return INCOMPLETE_WRITE?;
+	if (n != buffer.len) return INCOMPLETE_WRITE~;
 	return n;
 }
 
@@ -128,7 +136,7 @@ macro usz? read_using_read_byte(s, char[] buffer)
 		if (catch err = c)
 		{
 			if (err == io::EOF) return len;
-			return err?;
+			return err~;
 		}
 		*cptr = c;
 		len++;
@@ -152,7 +160,7 @@ macro char? read_byte_using_read(s)
 {
 	char[1] buffer;
 	usz read = s.read(&buffer)!;
-	if (read != 1) return io::EOF?;
+	if (read != 1) return io::EOF~;
 	return buffer[0];
 }
 
@@ -197,12 +205,12 @@ macro usz? copy_through_buffer(InStream in, OutStream dst, char[] buffer) @local
 		if (catch err = len)
 		{
 			if (err == io::EOF) return total_copied;
-			return err?;
+			return err~;
 		}
 		if (!len) return total_copied;
 		usz written = dst.write(buffer[:len])!;
 		total_copied += len;
-		if (written != len) return INCOMPLETE_WRITE?;
+		if (written != len) return INCOMPLETE_WRITE~;
 	}
 }
 
@@ -224,8 +232,8 @@ macro usz? read_varint(stream, x_ptr)
 		char? c = stream.read_byte();
 		if (catch err = c)
 		{
-			if (err == io::EOF) return io::UNEXPECTED_EOF?;
-			return err?;
+			if (err == io::EOF) return io::UNEXPECTED_EOF~;
+			return err~;
 		}
 		n++;
 		if (c & 0x80 == 0)
@@ -241,7 +249,7 @@ macro usz? read_varint(stream, x_ptr)
 		x |= (c & 0x7F) << shift;
 		shift += 7;
 	}
-	return math::OVERFLOW?;
+	return math::OVERFLOW~;
 }
 <*
  @require @is_outstream(stream)

lib/std/io/stream/buffer.c3

@@ -49,7 +49,7 @@ fn usz? ReadBuffer.read(&self, char[] bytes) @dynamic
 fn char? ReadBuffer.read_byte(&self) @dynamic
 {
 	if (self.read_idx == self.write_idx) self.refill()!;
-	if (self.read_idx == self.write_idx) return io::EOF?;
+	if (self.read_idx == self.write_idx) return io::EOF~;
 	char c = self.bytes[self.read_idx];
 	self.read_idx++;
 	return c;
@@ -132,5 +132,5 @@ fn void? WriteBuffer.write_byte(&self, char c) @dynamic
 fn void? WriteBuffer.write_pending(&self) @local
 {
 	self.index -= self.wrapped_stream.write(self.bytes[:self.index])!;
-	if (self.index != 0) return INCOMPLETE_WRITE?;
+	if (self.index != 0) return INCOMPLETE_WRITE~;
 }