Repository navigation
Missing impls for tuples? #16154
Description
rust-analyzer version: current master 0ed815fa
rustc version:
rustc 1.76.0-nightly (6a6287132 2023-12-17)
binary: rustc
commit-hash: 6a62871320e262661bb1a0ea7f8aec9d3abeddf2
commit-date: 2023-12-17
host: aarch64-apple-darwin
release: 1.76.0-nightly
LLVM version: 17.0.6
trait Boba {
type Foo;
}
impl<T: Eq + std::hash::Hash> Boba for T {
type Foo = String;
}
fn test() {
let a: <(bool, bool) as Boba>::Foo;
// ^ currently: <(isize, isize) as Boba>::Foo. This should've been `String`.
let a: <isize as Boba>::Foo;
// ^ currently: String
}I could've sworn that this worked before but I might be wrong.
Activity
Veykril commentedon Dec 18, 2023
Might've been #15970? Only thing I can think of that looks relevant and that has been merged recently (assuming this used to work)
cynecx commentedon Dec 19, 2023
Okay, I might have imagined things because I can reproduce this with a build from early November 🤷♂️.
Veykril commentedon Dec 19, 2023
Actually, this isn't a rust-analyzer regression. It's a core library change that trips over r-a rust-lang/rust#117050.
The impl macro makes use of nightly features (surprise surprise ...), and the behavior of those features has changed, which now breaks tuple impls for r-a because the expansion is wrong.
This also explains why r-a is giving me errors on our database traits now that I updated my toolchain yesterday
Auto merge of #16165 - Veykril:meta-vars, r=Veykril