Zig: Additional backend to generate C code

[uael]

I would like to known the position of the zig team on a C backend in addition to LLVM.

I known that not a lot of people will need such a backend, it's a very specific need which mostly target embedded system using custom architecture and C compiler. It's still the only missing step to stuck out from C, which would be very cool 👍

Thank you and keep the good work

[andrewrk]

I see it as one of many desirable targets. It's not provided by LLVM so this would be something provided by an alternate backend.

[uael]

There is one by the Julia team (resurrected llvm C backend) but not very usable on a work basis, and it's tied to llvm, which I don't see as a good solution.
Since it's desirable I will start thinking about it, I'm open to any advice for where to start, I'm still very unfamiliar with the Zig compiler

[andrewrk]

I don't think this project is startable yet; there are too many unfinished prerequisites.

• self-hosted compiler (#89)
• support for alternative backends besides LLVM in the self-hosted compiler

[uael]

Ok I will be patient then, thank you for enlightenment and quick answers :ok_hand:

[andrewchambers]

The closer the output is to human readable C, the better. For me it's like an insurance policy for the bus factor problem. The llvm to C translators aren't the best from that perspective.

[uael]

Duplicate of #5107, closing

[andrewrk]

Thanks - I had forgotten about this one. Let's keep the earlier one.

[andrewrk]

One note here is that once we are shipping self-hosted as the main zig compiler, we can remove LLVM from the bootstrapping process by having the stage1 zig compiler emit C code rather than LLVM IR. That is a separate issue however, since this issue is for self-hosted.

[andrewrk]

This is now part of stage2, and further improvements can be their own issues.