Suggest defining variable as mutable on &mut _ type mismatch in pats

[WaffleLapkin]

Suggest writing mut a where &mut a was written but a non-ref type provided.

Since we still don't have "apply either one of the suggestions but not both" kind of thing, the interaction with the suggestion of removing &[mut] or moving it to the type is weird, and idk how to make it better..

r? @compiler-errors

[compiler-errors]

I prefer to declare a mutable {thing} use: -- the mut variable_name part is redundant with the suggestion that follows direectly after

[compiler-errors]

Looks fine after nits

[compiler-errors]

sorry, my suggestion was not complete! you should probably remove this .map since it's no longer Option. I should have been more clear.

[compiler-errors]

Cool, thanks.

@bors r+ rollup

[bors]

📌 Commit 1dfb53b has been approved by compiler-errors

[camelid]

Since we still don't have "apply either one of the suggestions but not both" kind of thing, the interaction with the suggestion of removing &[mut] or moving it to the type is weird, and idk how to make it better..

I think that actually has been implemented: span_suggestions()

[compiler-errors]

Does span_suggestions() work correctly with Applicability::MachineApplicable?

[camelid]

Does span_suggestions() work correctly with Applicability::MachineApplicable?

In what regard? I think rustfix won't autoapply the suggestions if you use span_suggestions (since it doesn't know which one you want).

[WaffleLapkin]

@camelid rustfix tries to apply all suggestions:

; cargo +stage1 fix --broken-code
Checking t v0.1.0 (/home/waffle/projects/repos/t)
warning: error applying suggestions to `src/main.rs`
The full error message was:
> Cannot replace slice of data that was already replaced
This likely indicates a bug in either rustc or cargo itself,
and we would appreciate a bug report! You're likely to see
a number of compiler warnings after this message which cargo
attempted to fix but failed. If you could open an issue at
https://github.com/rust-lang/rust/issues
quoting the full output of this command we'd be very appreciative!
Note that you may be able to make some more progress in the near-term
fixing code with the `--broken-code` flag
Fixed src/main.rs (0 fixes)
Finished dev [unoptimized + debuginfo] target(s) in 0.15s

Also the output of the diagnostic is not perfect either, since suggestions don't have labels and it's hard to distinguish them:

error[E0308]: mismatched types
--> src/main.rs:1:7
|
1 | fn _f(&mut _a: i32) {}
| ^^^^^^^ --- expected due to this
| |
| expected `i32`, found `&mut _`
|
= note: expected type `i32`
found mutable reference `&mut _`
help: consider removing `&mut` from the pattern or using `mut` to declare variable as mutable
|
1 - fn _f(&mut _a: i32) {}
1 + fn _f(_a: i32) {}
|
1 | fn _f(mut _a: i32) {}
| ~~~

I think to fix this problem we need a different format for suggestions, that would allow for more flexibility. For example marking placeholders as such (so the tooling could allow users to fill them) or having "either one of these suggestions, but not all" kind of thing. I was meaning to try implementing something like this for a while, but haven't yet.

[camelid]

or having "either one of these suggestions, but not all" kind of thing

I believe that's what span_suggestions is supposed to be, but the implementation is non-optimal. In this case, I think you should be able to fix the crash by making the suggestion non-MachineApplicable (see rust-lang/cargo#13030). When @compiler-errors asked in #98431 (comment) about MachineApplicable, I'd forgotten about this issue with rustfix 😅