Confusing "the caller chooses a type which can be different" [E308]

[jendrikw]

Code

fn f<T>(t: &T) -> T {
    t
}

Current output

error[E0308]: mismatched types
 --> src/lib.rs:2:5
  |
1 | fn f<T>(t: &T) -> T {
  |      -            - expected `T` because of return type
  |      |
  |      expected this type parameter
2 |     t
  |     ^ expected type parameter `T`, found `&T`
  |
  = note: expected type parameter `_`
                  found reference `&_`
  = note: the caller chooses a type for `T` which can be different from `&T`

Desired output

error[E0308]: mismatched types
 --> src/lib.rs:2:5
  |
1 | fn f<T>(t: &T) -> T {
  |      -            - expected `T` because of return type
  |      |
  |      expected this type parameter
2 |     t
  |     ^ expected type parameter `T`, found `&T`
  |
  = note: expected type parameter `_`
                  found reference `&_`

Rationale and extra context

The statement "the caller chooses a type for T which can be different from &T" is confusing because T and &T are always different. I think it would be best to supress that note when the types only differ in reference-ness. The message was introduced in #122195.

Other cases

No response

Rust Version

rustc 1.81.0-nightly (3cf924b93 2024-06-15)
binary: rustc
commit-hash: 3cf924b934322fd7b514600a7dc84fc517515346
commit-date: 2024-06-15
host: x86_64-unknown-linux-gnu
release: 1.81.0-nightly
LLVM version: 18.1.7

Anything else?

No response

[jendrikw]

@rustbot label +D-confusing

[fmease]

Thanks, that note definitely doesn't make sense.

CC @jieyouxu (#122195).

I suggested to move note_caller_chooses_ty_for_ty_param out of try_suggest_return_impl_trait here in order to catch more cases but that was a bit too aggressive as can be witnessed here — that's on me.

We should replicate some parts of try_suggest_return_impl_trait, namely checking that the type param in question does not occur in the return expression type. That did cross my mind back then but I didn't really follow up on that thought for some reason.

[jieyouxu]

Thanks for the cc @fmease, I'll take a look ~today/tomorrow

[jieyouxu]

Opened #126558 to avoid making the note if found return type contains the expected return type (type parameter).