The seemingly contradictory `#[repr(Rust, C)]` is allowed

[fmease]

#[repr(Rust)] is already on beta and will soon hit stable.

I just found out that #[repr(C, Rust)] successfully compiles while I expected it to error due to contradictory representations.

Further, I wonder if #[repr(Rust, transparent)] should be allowed? Currently it errors with E0692, transparent struct cannot have other repr hints. (see err's response)

[compiler-errors]

Seems pretty obvious we should have exactly one repr, and only allow repr modifiers like packed afterwards. (Rust, transparent) seems like it's asking for two mutually exclusive things.

[marmeladema]

Should a C-struct be able to contain a Rust-struct ?
Currently the compiler is happy about this kind of construct too: https://play.rust-lang.org/?version=beta&mode=debug&edition=2021&gist=0588a57ac361ed85664914d86cb14060

If not, it might deserve it's own issue?

[workingjubilee]

That seems kosher to me? It might not be if

#[repr(C)]
pub struct ReprC(ReprRust, ReprRust);

But that's a different matter and has more to do with defining tuples in C being unclear.