Suggesting adding lifetime to return type

[kckeiks]

When there is a (elided) lifetime mismatch, we output a hint to introduce a lifetime but the hint does not suggest adding a lifetime to the return type.

Given the following code:

trait T {
    fn foo(&self, x: &i32) -> Option<&i32> {
        Some(x)
    }
}

The current output is:

error[E0623]: lifetime mismatch
  --> src/main.rs:11:9
   |
10 |     fn foo(&self, x: &i32) -> Option<&i32> {
   |                      ----     ------------
   |                      |
   |                      this parameter and the return type are declared with different lifetimes...
11 |         Some(x)
   |         ^^^^^^^ ...but data from `x` is returned here
   |
   = note: each elided lifetime in input position becomes a distinct lifetime
help: consider introducing a named lifetime parameter
   |
10 |     fn foo<'a>(&'a self, x: &'a i32) -> Option<&i32> {
   |           ++++  ++           ++

error: aborting due to previous error

For more information about this error, try `rustc --explain E0623`.

Ideally the output should also suggest adding 'a to the return type:

error[E0623]: lifetime mismatch
  --> src/main.rs:11:9
   |
10 |     fn foo(&self, x: &i32) -> Option<&i32> {
   |                      ----     ------------
   |                      |
   |                      this parameter and the return type are declared with different lifetimes...
11 |         Some(x)
   |         ^^^^^^^ ...but data from `x` is returned here
   |
   = note: each elided lifetime in input position becomes a distinct lifetime
help: consider introducing a named lifetime parameter
   |
10 |     fn foo<'a>(&'a self, x: &'a i32) -> Option<&'a i32> {
   |           ++++  ++           ++                 ++

error: aborting due to previous error

For more information about this error, try `rustc --explain E0623`.

[kckeiks]

The diagnosis error shown in the issue description is created by the type NiceRegionError.
There is another type, LifetimeContext, that displays a better hint for similar code:

fn foo(x: &i32, y: &i32) -> Option<&i32> {
    Some(y)
}

Output:

error[E0106]: missing lifetime specifier
 --> src/main.rs:1:36
  |
1 | fn foo(x: &i32, y: &i32) -> Option<&i32> {
  |           ----     ----            ^ expected named lifetime parameter
  |
  = help: this function's return type contains a borrowed value, but the signature does not say whether it is borrowed from `x` or `y`
help: consider introducing a named lifetime parameter
  |
1 | fn foo<'a>(x: &'a i32, y: &'a i32) -> Option<&'a i32> {
  |       ++++     ++          ++                 ++

error: aborting due to previous error

For more information about this error, try `rustc --explain E0106`.

These types have different information about the code, for example, LifetimeContext has span info for all parameters in function so it can produce a nice suggestion but NiceRegionError has "region" information which I think limits the quality of the hint it can produce. It would be great if hints produced could be more consistent regardless of where in the analysis step they're being generated from.

[kckeiks]

@rustbot label P-low

[rustbot]

Error: Label P-low can only be set by Rust team members

Please let @rust-lang/release know if you're having trouble with this bot.

[Dylan-DPC]

This is now resolved and shows the lifetime in the return type in the suggestion.