-Ztrait-solver=next-coherence allows overlapping impls

[cynecx]

trait Foo<'a, K: ?Sized> {
    fn foo(self);
}

// K=str => &'a [&'a str]: Foo<'a, str>
impl<'a, K: ?Sized> Foo<'a, K> for &'a [&'a K] {
    fn foo(self) {
        println!("a");
    }
}

// K=&'a str => &'a [&'a str]: Foo<'a, str>
impl<'a, K: Owned<'a>> Foo<'a, K::Ref> for &'a [K] {
    fn foo(self) {
        println!("b");
    }
}

trait Owned<'a> {
    type Ref: ?Sized;
}

impl<'a> Owned<'a> for &'a str {
    type Ref = str;
}

fn main() {
    let a = &[""] as &[_];
    Foo::<'static, str>::foo(a);
}

https://godbolt.org/z/9Khvh38oK

With -Ztrait-solver=next-coherence:

error[E0283]: type annotations needed: cannot satisfy `&[&str]: Foo<'static, str>`
  --> <source>:29:5
   |
29 |     Foo::<'static, str>::foo(a);
   |     ^^^^^^^^^^^^^^^^^^^^^^^^
   |
note: multiple `impl`s satisfying `&[&str]: Foo<'static, str>` found
  --> <source>:6:1
   |
6  | impl<'a, K: ?Sized> Foo<'a, K> for &'a [&'a K] {
   | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
...
13 | impl<'a, K: Owned<'a>> Foo<'a, K::Ref> for &'a [K] {
   | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

error: aborting due to previous error

For more information about this error, try `rustc --explain E0283`.
Compiler returned: 1

With -Ztrait-solver=next: Compiles.

With classic solver (As expected):

error[E0119]: conflicting implementations of trait `Foo<'_, _>` for type `&[&_]`
  --> <source>:13:1
   |
6  | impl<'a, K: ?Sized> Foo<'a, K> for &'a [&'a K] {
   | ---------------------------------------------- first implementation here
...
13 | impl<'a, K: Owned<'a>> Foo<'a, K::Ref> for &'a [K] {
   | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ conflicting implementation for `&[&_]`

error: aborting due to previous error

rustc 1.76.0-nightly (2f8d81f9d 2023-11-21)
binary: rustc
commit-hash: 2f8d81f9dbac6b8df982199f69da04a4c8357227
commit-date: 2023-11-21
host: x86_64-unknown-linux-gnu
release: 1.76.0-nightly
LLVM version: 17.0.5

[compiler-errors]

I'm almost certain this will be fixed by #117088, cc @lcnr

edit: Yes, I've confirmed this manually.

[cynecx]

This has been fixed since #117088 has been merged.