rustc_codegen_gcc VS gccrs

It's true that if this project succeeds, that would be an outcome, but it's probably worth noting that you'd really only have to add support for the GCC backend to do that, and not reimplement the frontend as well (parsing, type checking, lifetime checking, etc.). There's an unrelated project working to do that: https://github.com/antoyo/rustc_codegen_gcc that would likely yield those same benefits for less effort.

platform support (may improve soon)

Here's the gcc backend for rustc here https://github.com/antoyo/rustc_codegen_gcc Since this just swaps the optimizer/codegen module and reuses all the other stuff from rustc it needs less work and can already compile many valid rust programs.

I'd like to use the rustc_codegen_gcc backend to compile some Rust programs, but I'm confused about how to install and use the patched libgccjit dependency. I've downloaded the fork provided but don't know where to go from there.

I was being a little sloppy and mixing together "actual embedded hardware" with "extensions for applications written in other languages" in my head. I think your point about LLVM is still accurate, though I hear peeps about different projects working on GCC support from time to time.

This doesn't have anything to do with using GCC to compile rust, but instead using rustc to compile Rust using the GCC backend. You can do that today by using the linked project, rustc_codegen_gcc, which is intended to be integrated into rustc at some point in the (hopefully near) future.

Rather than mrustc you should probably look at rustc_codegen_gcc which uses gcc as a backend to rustc (WIP) or GCC Rust which tries to implement a rust frontend for gcc (also WIP). I think rustc_codegen_gcc looks the most promising at the moment

Libgccjit AOT Codegen for Rustc\ (2 comments)

There is a Rust front-end for GCC that is under active development [1]. If the chip vendors are not willing to develop and upstream a LLVM back-end then they can feel free to start contributing to it.

[1] https://rust-gcc.github.io/

That's why gccrs doesn't even consider lifetime checking a part of the language (they plan to use Polonius, too).

There's ongoing work on a Rust front-end for GCC (https://github.com/Rust-GCC/gccrs). Bit barebones right now -- ie, even core doesn't compile -- but there's funding, demand, and regular progress, so it'll only get better from there. Once gccrs can compile core, it should be ready to compile most of Rust, and thus if you've taught the calling conventions for C to GCC, you're golden.

I recommend checking out the GCC Rust frontend project.

That is what theyre refering to, yes. The GitHub is named https://github.com/Rust-GCC/gccrs

- But this misses so much extra context information

3. Macro invocations there are really subtle rules on how you treat macro invocations such as this which is not documented at all https://github.com/Rust-GCC/gccrs/blob/master/gcc/rust/expan...

Some day I personally want to write a blog post about how complicated and under spec'd Rust is, then write one about the stuff i do like it such as iterators being part of libcore so i don't need reactive extensions.

(Speaking of which, Rust-GCC (or gcc-rs or gccrs or whichever other of their names they decide is the primary one) isn't even going to be a complete C++ implementation. Their plan is to implement enough to compile Polonius (the NLL 2.0 borrow checker being developed in Rust for rustc) and then share that since borrow-checking isn't necessary for codegen... only to identify and reject invalid programs... making the C++ portion of it not that different in scope from mrustc.)

That bridge will be crossed with gccrs (compiling Rust with gcc directly, coming next month with GCC 13) and rust_codegen_gcc (rustc frontend, GCC backend, works now but just doesn’t yet have an “easy” setup)