Nonsensical diagnostic when returning closure

[danielkeller]

Given the following code:

fn bar<Input>() -> impl Fn(Input) {
    |i| todo!()
}
    
fn foo() -> impl Fn(&str) {
    bar()
}

The current output is:

error[E0308]: mismatched types
 --> src/lib.rs:5:13
  |
1 | fn bar<Input>() -> impl Fn(Input) {
  |                    --------------
  |                    |
  |                    the expected opaque type
  |                    the found opaque type
...
5 | fn foo() -> impl Fn(&str) {
  |             ^^^^^^^^^^^^^ one type is more general than the other
  |
  = note: expected associated type `<impl Fn<(&str,)> as FnOnce<(&str,)>>::Output`
             found associated type `<impl Fn<(&str,)> as FnOnce<(&str,)>>::Output`

For more information about this error, try `rustc --explain E0308`.
error: could not compile `playground` due to previous error

It seems like a bug that the "expected" and "found" types are the same.

[lqd]

Note that these impl Fns, and the others in https://internals.rust-lang.org/t/nonsensical-error-message-about-mismatched-types/15422, will unfortunately trigger #57374 with NLLs turned on.

[bjorn3]

With the nightly only -Zverbose the output is:

error[E0308]: mismatched types
 --> <anon>:5:13
  |
1 | fn bar<Input>() -> impl Fn(Input) {
  |                    --------------
  |                    |
  |                    the expected opaque type
  |                    the found opaque type
...
5 | fn foo() -> impl Fn(&str) {
  |             ^^^^^^^^^^^^^ one type is more general than the other
  |
  = note: expected associated type `<Opaque(DefId(0:5 ~ rust_out[8787]::bar::{opaque#0}), [&'_#0r str]) as FnOnce<(&'_#0r str,)>>::Output`
             found associated type `<Opaque(DefId(0:5 ~ rust_out[8787]::bar::{opaque#0}), [&'_#0r str]) as FnOnce<(&RePlaceholder(Placeholder { universe: U3, name: BrAnon(0) }) str,)>>::Output`

error: aborting due to previous error

For more information about this error, try `rustc --explain E0308`.

[estebank]

CC @oli-obk, who has a couple of PRs in flight that will likely affect the output of this code.

[oli-obk]

there may be an easy fix in old regionck, as we should be able to normalize the placeholder lifetime back to the higher kinded lifetime that it came from (even if just for diagnostics).

But our efforts are probably better spent in fixing nll error once I get through all this.