Take into account negative impls in "trait item not found" suggestions

[LeSeulArtichaut]

This removes the suggestion to implement a trait for a type when that type already has a negative implementation for the trait, and replaces it with a note to point out that the trait is explicitely unimplemented, as suggested by @scottmcm.

Helps with #79683.

r? @scottmcm do you want to review this?

[LeSeulArtichaut]

This is actually sort of a shortcut, consider the following code:

#![feature(negative_impls)]

trait Foo {
    fn foo(&self);
}

trait Bar {}

impl<T: Foo> !Bar for T {}

fn foo<T: Bar>(x: T) {
    x.foo()
}

Which gives the following error:

error[E0599]: no method named `foo` found for type parameter `T` in the current scope
  --> main.rs:12:7
   |
12 |     x.foo()
   |       ^^^ method not found in `T`
   |
   = help: items from traits can only be used if the type parameter is bounded by the trait
help: the following trait defines an item `foo`, perhaps you need to restrict type parameter `T` with it:
   |
11 | fn foo<T: Foo + Bar>(x: T) {
   |        ^^^^^^^^

error: aborting due to previous error

So I think this could be improved but I think it's reasonable to do it in a follow-up PR since this PR does not change the current diagnostic.

[LeSeulArtichaut]

My PR also doesn't work in:

#![feature(negative_impls)]

trait Bar {
    fn bar(&self);
}

impl<T: Copy> !Bar for T {}

fn foo(x: ()) {
    x.bar()
}

fn main() {}

error[E0599]: no method named `bar` found for unit type `()` in the current scope
  --> main.rs:10:7
   |
10 |     x.bar()
   |       ^^^ method not found in `()`
   |
   = help: items from traits can only be used if the trait is implemented and in scope
note: `Bar` defines an item `bar`, perhaps you need to implement it
  --> test.rs:3:1
   |
3  | trait Bar {

[LeSeulArtichaut]

r? @lcnr

[lcnr]

lgtm

one additional small nit would be to use

match explicitly_negative/potential_candidates {
    [] => {},
    [trait_info] => ...
    slice => ...
}

instead of the current if not_empty { if let ... { ... } }

[lcnr]

r=me

@bors delegate+

[bors]

✌️ @LeSeulArtichaut can now approve this pull request

[LeSeulArtichaut]

Thanks for your review @lcnr!
@bors r=lcnr

[bors]

📌 Commit cfc38d2 has been approved by lcnr

[bors]

⌛ Testing commit cfc38d2 with merge 0d7c7ea307a01cf5e08d93bca34e168d5e122e9e...

[Dylan-DPC-zz]

failed in rollup log

@bors r-

[LeSeulArtichaut]

@Dylan-DPC How is this PR causing that "could not parse code block as Rust code" error?

[bors]

☀️ Try build successful - checks-actions
Build commit: 0d7c7ea307a01cf5e08d93bca34e168d5e122e9e (0d7c7ea307a01cf5e08d93bca34e168d5e122e9e)

[LeSeulArtichaut]

I don't know where this try build comes from, but based on this I will take the liberty of accepting this again.
@bors r=lcnr

[bors]

📌 Commit cfc38d2 has been approved by lcnr

[Dylan-DPC-zz]

I don't know but the error was tough to debug and this looked like t he closest PR :D

[LeSeulArtichaut]

I'd guess the failure comes from #79816 which IIUC creates this new lint?

[Dylan-DPC-zz]

yeah the test failure now on that PR, confirms it. sorry about the mess :P

[bors]

⌛ Testing commit cfc38d2 with merge a6491be...

[bors]

☀️ Test successful - checks-actions
Approved by: lcnr
Pushing a6491be to master...