Thanks to a colossal amount of work by Jacob Young, the x86_64 backend of Zig is now passing 101% of the behavior tests compared to the LLVM backend.

Or perhaps put another way, the LLVM backend is passing 99% of the behavior tests compared to the x86_64 backend 😉

Still a few more issues to tackle before it can be made the default, however.

Seems good for now but note this is going to disrupt my ultimate plan of creating a viable binary including when there are compilation errors. Including when there are syntax errors.

Пока всё выглядит хорошо, но учти, что это нарушит мой главный план — создавать работоспособный бинарник даже при наличии ошибок компиляции. В том числе при наличии синтаксических ошибок.

Sure. In general, we're finally nearing the end of language stabilization phase of the roadmap (although there are still a few breaking language changes planned). The next phase will be standard library stabilization. The first step of that will be minimization of the standard library, aggressively deleting what is deemed unnecessary. Then it will be auditing, then it will completeness additions, in collaboration with Zig communities and ecosystem.

Before, when the language was younger, any additions to the std lib were welcomed as they helped test pilot the language. Now, the std lib is big enough and the ecosystem is big enough that there are other better ways to test pilot the language.

This is accepted mainly motivated by the SPIR-V backend, but the types will naturally be supported on all targets, and is expected to also fulfill the ergonomics use case for graphics programming in general.