Skip to content

Latest commit

 

History

History
114 lines (78 loc) · 6.24 KB

README.md

File metadata and controls

114 lines (78 loc) · 6.24 KB

rustc_codegen_clr

Warning

This project is still early in its developement. Bugs, crashes and miscompilations are expected. DO NOT USE IT FOR ANYTHING SERIOUS.

rustc_codegen_clr is an experimental Rust compiler backend(plugin), which allows you to transpile Rust into .NET assebmlies, or C source files.

The end goal of the project is allowing Rust to be used in places where it could not be used before.

.NET Interop layer

The project aims to provide a way to easily use Rust libraries in .NET. It comes with a Rust/.NET interop layer, which allows you to easily interact with .NET code from Rust:

use mychorizza::*;
fn main(){
    // Alocate a new GC-managed string builder
    let stringBuilder = StringBuilder::empty();
    // You can easily operate on GC-managed types
    mstring.AppendChar('H');
    mstring.AppendChar('i');
    mstring.AppendChar('.');
}

This should allow you to integrate Rust code with exisitng .NET codebases, and should allow you to use .NET-specific libraries or APIs from Rust.

The project will also include support for defining .NET classes from Rust, allowing .NET code to easily call Rust. This is currently heavily WIP, and any feedback is appreciated.

// Early WIP syntax, subject to change.
dotnet_typedef! {
  class MyClass inherits [Some::External::Assebmly]SomeNamespace::SomeClass{
    virtual fn ToString(_this:MyClass)->MString{
      "I am a class defined in Rust!".into_managed()
    },
  }
}

With this approach, the classes and APIs exposed to .NET can be easily used from other .NET languages, like F# or C#. The safety of this glue layer can be checked by the Rust compiler, which should make interop issues much less likely.

C support

While .NET is the main foccus of my work, this project can also be used to compile Rust to C, by setting the C_MODE enviroment flag to 1.

This may seem like a strange and unrelated feature, but the project was written in such a way that this is not only possible, but relatively easy.

My representation of .NETs IR maps nicely to C, which means that I was able to add support for compiling Rust to C in 2-3K LOC. Almost all of the codebase is reused, with the C and .NET specific code only present in the very last stage of compilation.

This means that, instead of having to maintain 2 separate projects, I can maintian one project. Bug fixes to the .NET side of things also fix C bugs. Because of that, the support for C in the project is almost as good as support for .NET

Current state of the project

The project currently supports most Rust features (except proc macros), but it is not bug-free. It can compile a mostly working version of Rust std, but there are many minor bugs make such std not 100% functional.

Most components of std are about 95% working in .NET, and 80% working in C.

Currently, the GCC and clang C compilers are supported, with plans to add support for tcc, and maybe even sdcc.

So, you can compile a lot of existing Rust code, but it may not necessarily work.

core, std, and alloc uint tests.

.NET

Name Pass Faliure Crash \ Timeout OK precentage
Core tests 1662 39 12 97.02%
Alloc tests 616 8 40 92.77%
Alloc benches 464 0 0 100.00%
Test Harness tests 57 0 100.00%
std tests 931 43 64 89.69%
Core benches 491 1 98.99%

C

Name Pass Faliure OK precentage
Core tests 1419 294 82.83%

FAQ

Q: What is it?

A: This is a compiler backend for rustc, which targets the .NET platform and runtime; this would enable you to use some Rust libraries from C#/F#, with little effort.

Q: Is Rust's memory management useless in .NET?

A: Rust code typically uses the stack more than the heap, which can speed up code running within the CLR runtime. Heap-allocated objects are allocated from unmanaged (non-GC) memory and are allocated and freed in the same way as in Rust.

Q: Is this useless since I can already load shared libraries from C#?

A: The Rust APIs this codegen exposes to C#/F# code are only slightly easier to use than those exposed by a .so or .dll Rust library. Interop still requires some effort, but the Rust code is bundled with everything else. Types used from C# are guaranteed to be the same as those in C#, preventing mismatch issues. All types can be safely sent between Rust and C#, with exactly the same layout. Additionally, since all Rust code compiled with this codegen can be bundled with C#/F# code, you no longer need to ship different versions of the library for different architectures. Any architecture supported by CLR works out of the box, using the exact same assembly.

You can also avoid the cost of switching between code running within and outside the runtime. This cost is not unbearable, but it is not easily eliminated, and reducing it can have safety penalties. In this case, all code runs within the runtime, meaning there is no transition between code running inside and outside the runtime.

Compiling Rust to CLR can potentially improve JIT optimization. Since the CLR's JIT now sees all the code, it can make better decisions about optimization, resulting in faster code.

Q: Compatibility?

A: rustc_codegen_clr is only tested on Linux x86_64, with the CoreCLR runtime (more commonly known as simply the .NET runtime), on .NET 8. It should work on other platforms, but it is not guaranteed.

Q: Whata about Mono?

A The support for the Mono runtime is not as good as it could be. Due to not supported features and differences, 128-bit integers and checked 64-bit integer arithmetic are not supported on Mono. Aligned allocators(__rust_alloc) and certain intrinsics are also not supported. I plan to expand support for Mono, but my resources are limited.

Q: Are there any issues?

A: While the backend is extensively tested, it is still far from perfect, and there are still many edge cases that may break this backend. A: Currently, there are no .NET-specific versions of std or .NET specific target triples. This means that you will need separate .NET assemblies for each OS.

Licensing

rustc_codegen_clr is dual licensed under MIT license or Apache License, Version 2.0.