- Generating typeid from function gives incorrect typeid #2816
- Recursive definitions not discovered when initializer is access on other const #2817
- Slice overrun detected late hit codegen assert #2822
* add deprecations to macro slot builtins
* refactor all stdlib uses of now-deprecated EMPTY_MACRO_SLOT; release notes
* update incorrect releasenotes ref to this pr
* remove leftover comments from refactoring
* remove unnecessary `EmptySlot`-like type in countingsort; use private macro directly
- Comparing an uint and int[<4>] was incorrectly assumed to be uint compared to int, causing a crash instead of an error.
- When an `int[*][6]` was given too few values, the compiler would assert instead of giving an error.
- Comparing an uint and int[<4>] was incorrectly assumed to be uint compared to int, causing a crash instead of an error.
- When an `int[*][6]` was given too few values, the compiler would assert instead of giving an error.
- Using named parameters with builtins would cause a crash.
- In some cases, using missing identifiers with builtins would cause a crash.
- Using `$defined` with function call missing arguments would cause a crash.
- Adding @nostrip to a test function would crash.
- Mixing struct splat, non-named params and named params would crash rather than to print an error.
- Creating a char vector from bytes would crash.
- Using $$wstr16 with an illegal argument would crash instead of printing an error.
- Passing different types to arg 1 and 2 for $$matrix_transpose would trigger an assert.
- Zero init of optional compile time variable would crash the compiler.
- Using multiple declaration for generics in generic module would fail.
- Defining an extern const without a type would crash rather than print an error.
- Typedef followed by brace would trigger an assert.
- Union with too big member would trigger an assert.
* Merge AsciiCharset CT/non-CT Functions
* release notes
* incorporate helpful review feedback
* re-separate 'create_set' and 'contains' but keep 'combine_sets'; update tests
* tabs (annoying IDE)
* Restored old code verbatim for smaller diff. Split combine_sets into easier to macro/function for runtime / macro version, this also allows for more easy type checks.
---------
Co-authored-by: Christoffer Lerno <christoffer@aegik.com>
* std/io/os/temp_directory.c3: fix INVALID_PATH in Windows native_temp_directory
- Use the actual length from GetTempPathW for Windows temp path slice
- We can remove the workaround in the test_suite_runner for WIN32 and
create all directories in %temp% now
* Updated releasenotes.
---------
Co-authored-by: Christoffer Lerno <christoffer@aegik.com>
* Require parenthesized assignment expressions in condition of 'if' statements #2716
* Move analysis to semantic checker and also check while. And update tests and release notes.
---------
Co-authored-by: Christoffer Lerno <christoffer@aegik.com>
* Fixed macro hash_vec(vec): it was checking only 8 bytes
* use $sizeof() directly as it is correct now
* Added unit test for vector types
Tests that hashes match for same values and that they don't match
when changing a single bit on any of the vector elements
* Changed hardcoded value for the compiler option
* Release note
---------
Co-authored-by: Christoffer Lerno <christoffer@aegik.com>
For-loop in LinkedList.to_format goes brrrrrr and prints the value of
the first node an infinte number of times because it does not properly
iterate the linked list.
Fix the list iterations and add the missing string format specifier.
Bug can be reproduced with:
```
import std;
fn void main() => @pool()
{
io::printfn("%s", linkedlist::@tnew{usz}({1,2,3}));
}
```
