`mismatched types, one type is more general than the other` error shows types with the same name

[micouy]

Example code:

fn main() {
    let mut v: Vec<i32> = vec![];
    let f = |_| true;
    v.retain(f);
}

Error message:

error[E0308]: mismatched types
 --> src/main.rs:4:7
  |
4 |     v.retain(f);
  |       ^^^^^^ one type is more general than the other
  |
  = note: expected type `FnOnce<(&i32,)>`
             found type `FnOnce<(&i32,)>`

rustc version:

rustc 1.48.0-nightly (dbb73f8f7 2020-09-12)
binary: rustc
commit-hash: dbb73f8f79ab176a897d5a95e696adb71b957cbe
commit-date: 2020-09-12
host: x86_64-apple-darwin
release: 1.48.0-nightly
LLVM version: 11.0

[jyn514]

Possible duplicate of #76353

[camelid]

:)

rust/compiler/rustc_infer/src/infer/error_reporting/note.rs

Line 288 in d92d28e

// I can't think how to do better than this right now. -nikomatsakis

[micouy]

For those having the same problem - adding type annotations in the closure definition makes the error go away.

fn main() {
    let mut v: Vec<i32> = vec![];
    let f = |_: &i32| true; // now it compiles
    v.retain(f);
}

[jyn514]

let f = |_: &_| true; will also work.

[camelid]

searched nightlies: from nightly-2019-01-01 to nightly-2020-10-28
regressed nightly: nightly-2020-06-24
searched commits: from 6bb3dbf to ff5b446
regressed commit: 1557fb0

bisected with cargo-bisect-rustc v0.6.0

Host triple: x86_64-apple-darwin
Reproduce with:

cargo bisect-rustc --preserve --prompt --start=2019-01-01 

[camelid]

This is an error -> very confusing error issue.

I actually don't understand the errors that used to be emitted. Can anyone else make sense of it?

error[E0631]: type mismatch in closure arguments
 --> src/main.rs:4:14
  |
3 |     let f = |_| true;
  |             -------- found signature of `fn(_) -> _`
4 |     v.retain(f);
  |              ^ expected signature of `for<'r> fn(&'r i32) -> _`

error[E0271]: type mismatch resolving `for<'r> <[closure@src/main.rs:3:13: 3:21] as std::ops::FnOnce<(&'r i32,)>>::Output == bool`
 --> src/main.rs:4:7
  |
4 |     v.retain(f);
  |       ^^^^^^ expected bound lifetime parameter, found concrete lifetime

error: aborting due to 2 previous errors

Unfortunately, the regression was in a rollup :/

[camelid]

The regression was in a rollup, but I think it's pretty likely that #73496 is the cause, so cc @estebank