[next-trait] expected &dyn T found &Y

[jhgg]

cc #20443 - this does not repro in stable release.

rust-analyzer version: 0.4.2586-standalone (a905e3b 2025-08-18)

rustc version: rustc 1.89.0-nightly (99e7c15 2025-06-01)

editor or extension: VSCode

relevant settings: N/A

code snippet to reproduce:

trait Foo {
    fn bar(&self) -> u32 {
        0xCAFE
    }
}

fn debug(_: &dyn Foo) {}

impl Foo for i32 {}

fn main() {
    debug(&1);
     //   ^^ expected &dyn Foo, found &i32
}

Similar to #20489, the error does not occur if the impl of Foo has no methods, e.g.

trait Foo { }

[flodiebold]

Coercion still uses Chalk, so this is an old bug.

[jhgg]

Coercion still uses Chalk, so this is an old bug.

This does not repro in stable though.

[jhgg]

I minimized the repro to not involve serde.

[ShoyuVanilla]

Those trait object coercion issues (this and #20489) might be related to the first item in #20422 . I'll give this a try.

[Austaras]

But testing on master with

    check_no_mismatches(
        r#"
//- minicore: coerce_unsized, dispatch_from_dyn
trait Foo {
    fn bar(&self) -> u32 {
        0xCAFE
    }
}

fn debug(_: &dyn Foo) {}

impl Foo for i32 {}

fn main() {
    debug(&1);
}"#,
    );

reports no error.

[ChayimFriedman2]

You're right, it's fixed on master. Not sure what fixed it, though. I'll add a test and close this.

[ShoyuVanilla]

Funny that I've started next-solver migration of inference table to fix this issue but this has been fixed somehow 😂 Of course inference table migration should be done for many reasons.

[jhgg]

I do not think this is actually fixed. I am trying latest nightly (r0.4.2604-standalone (fa19df2 2025-09-04))

Error is still very much present. #20489 is also not fixed as well.

[ChayimFriedman2]

I get no error:

rust-analyzer diagnostics does not report an error either. Are you sure you are using the latest r-a?