Andrew Kelley

Thanks! Great patch.

This has timed out twice now on the Windows CI machines.

2 months old, never was passing CI

My currently favored solution to this problem is to delete the non-array hash map implementation from the standard library. Automatically rehashing after every R insert/deletion operations leads to non-uniform performance...

`items` is [part of the API](https://ziglang.org/documentation/0.6.0/std/#std;ArrayList). It is not an internal implementation detail.

Huh? this doesn't make sense to me at all. First of all distros will be using --system so there will be no downloads at all. Secondly, I don't see how...

I think this is mainly due to `has_side_effects` being no longer an early return case like it was before #19826. If it has side effects it shouldn't be calling cache.hit()....

current status is: ``` CMake Error at /nix/store/mk7637n6h4d437gl5cplxvd3k2l66awd-cmake-3.29.2/share/cmake-3.29/Modules/CMakeTestCCompiler.cmake:67 (message): The C compiler "/home/andy/dev/zig-bootstrap/out/host/bin/zig" is not able to compile a simple test program. It fails with the following output: Change Dir:...

With LLVM 19.1.0rc4 I get this: ``` CMake Error at /nix/store/28s979fhz8d6h1rny3n5zh1ybmby6kjz-cmake-3.29.2/share/cmake-3.29/Modules/CMakeTestCCompiler.cmake:67 (message): The C compiler "/home/andy/dev/zig-bootstrap/out/host/bin/zig" is not able to compile a simple test program. It fails with the following...

Those error indicate missing loongarch-specific logic in the zig standard library.