For what it’s worth, I would find a #lang rust useful. I’m porting some Racket to Rust as part of a project soon, and having a reference implementation and interop would be a huge simplifier. If it were done well it would be possible to gradually port from Racket to Rust. Oddly, I don’t think it would be important for it to be fast. Thinking on it, I am tempted to give this a try if you do not. That or we could collaborate on it. Though everyone is busy of course.
I don’t think I would actually do it.
Even if you want to gradually port, why not just move bits to Rust and then call them from Racket using the FFI? I think that might be both easier and lower risk, since you wouldn’t have to start by attempting to reproduce an existing and very complex language.
I’m guessing #lang rust would actually give you a much tougher row to hoe on the FFI front, because then you’re likely to get stuck figuring out how #lang rust could import crates.
It’s funny you mention that because that was my biggest sticking point too. I still think it would be fun but it may be more challenging and less practical than I had guessed at first.
Fun syntax
Pretty much inherited the spirit of Lisp.