Tracking issue for dual-proc-macros

[ehuss]

Implemented in #58013
Cargo side is rust-lang/cargo#6547

This feature causes Cargo to build a proc-macro crate for both the target and the host, and for rustc to load both.

I'm not sure if this is a perma-unstable feature for the compiler-only, or if there is any need to use it elsewhere. But a tracking issue may be helpful for anyone looking for more information.

[crlf0710]

This might has something to do with #62558, according to the patches in termux/termux-packages#4324 (comment)

[mati865]

See also #62558 (comment)

[petrochenkov]

Some more detailed explanation of why this feature is necessary and how it works - https://rust-lang.zulipchat.com/#narrow/stream/131828-t-compiler/topic/-Z.20dual-proc-macros.

[pnkfelix]

Discussed in T-compiler backlog bonanza

Our understanding is that this is a fragile bit of code in here to support cross-compiling of rustc itself. We don't want to stabilize it as is. Its possible we want to try to get rid of it, depending on how one interprets the dialogue at #t-compiler > -Z dual-proc-macros

@rustbot label: S-tracking-design-concerns S-tracking-perma-unstable