feat: add test for async-trait type mismatch

[Aditya-PS-05]

closes #19957 (remaining part)

Made a test to verify that issue #19957 is resolved. The test passes successfully, confirming that the async-trait type mismatch issue has been fixed by the migration to the new trait solver (commit d1288f6).

About Test

As we can't directly use #[async_trait] in hir-ty tests (it's not available in minicore), the test simulates what the #[async_trait] macro expands to.

The test verifies that no type mismatch occurs between the Pin<Box<dyn Future>> signature and the Pin<Box<impl Future>> returned by the async block.

[rustbot]

⚠️ Warning ⚠️

• There are issue links (such as #123) in the commit messages of the following commits.
  Please move them to the PR description, to avoid spamming the issues with references to the commit, and so this bot can automatically canonicalize them to avoid issues with subtree.
  • 5954ede

[ShoyuVanilla]

This won't gonna work since we have no Box in our minicore. So, it won't cause mismatch even if the issue exists in the real world, as we don't emit type mismatches for unknown types. And it would be good if you could check that the same test code fails on the past version.

[Aditya-PS-05]

When I tested on three commits,

Test file;

#[test]
// #[should_panic(expected = "Unexpected type mismatches")]
fn regression_19957() {
    // This test documents issue #19957: async-trait patterns incorrectly produce
    // type mismatches between Pin<Box<dyn Future>> and Pin<Box<impl Future>>.
    //
    // The test currently FAILS because the bug is not yet fixed.
    // When the bug is fixed, remove the #[should_panic] attribute.
    check_no_mismatches(
        r#"
//- minicore: future, pin, result, error, send, coerce_unsized, dispatch_from_dyn
use core::{future::Future, pin::Pin};

#[lang = "owned_box"]
pub struct Box<T: ?Sized> {
    inner: *mut T,
}

impl<T> Box<T> {
    fn pin(value: T) -> Pin<Box<T>> {
        // Implementation details don't matter here for type checking
        loop {}
    }
}

impl<T: ?Sized + core::marker::Unsize<U>, U: ?Sized> core::ops::CoerceUnsized<Box<U>> for Box<T> {}

impl<T: ?Sized + core::ops::DispatchFromDyn<U>, U: ?Sized> core::ops::DispatchFromDyn<Box<U>> for Box<T> {}

pub struct ExampleData {
    pub id: i32,
}

// Simulates what #[async_trait] expands to
pub trait SimpleModel {
    fn save<'life0, 'async_trait>(
        &'life0 self,
    ) -> Pin<Box<dyn Future<Output = i32> + Send + 'async_trait>>
    where
        'life0: 'async_trait,
        Self: 'async_trait;
}

impl SimpleModel for ExampleData {
    fn save<'life0, 'async_trait>(
        &'life0 self,
    ) -> Pin<Box<dyn Future<Output = i32> + Send + 'async_trait>>
    where
        'life0: 'async_trait,
        Self: 'async_trait,
    {
        // Body creates Pin<Box<impl Future>>, which should coerce to Pin<Box<dyn Future>>
        Box::pin(async move { self.id })
    }
}
"#,
    );
}

1.) On 2b2d9d8 (it's a commit before d1288f6)

Result:

   Compiling hir-ty v0.0.0 (/tmp/test-old/crates/hir-ty)
    Finished `test` profile [unoptimized + debuginfo] target(s) in 46.78s
     Running unittests src/lib.rs (target/debug/deps/hir_ty-b1d8e0b478de4a85)

running 1 test
test tests::regression::new_solver::regression_19957 ... FAILED

failures:

---- tests::regression::new_solver::regression_19957 stdout ----

thread 'tests::regression::new_solver::regression_19957' (124130) panicked at crates/hir-ty/src/tests/regression/new_solver.rs:564:5:
Unexpected type mismatches:
1109..1141: expected Pin<Box<dyn Future<Output = i32> + Send + 'async_trait>>, got Pin<Box<impl Future<Output = i32>>>

note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace


failures:
    tests::regression::new_solver::regression_19957

test result: FAILED. 0 passed; 1 failed; 0 ignored; 0 measured; 917 filtered out; finished in 0.07s

error: test failed, to rerun pass `-p hir-ty --lib`

2.) On d1288f6 commit (this is the commit where the new solver trait was implemented)

Result:

   Compiling hir-ty v0.0.0 (/tmp/test-old/crates/hir-ty)
    Finished `test` profile [unoptimized + debuginfo] target(s) in 48.29s
     Running unittests src/lib.rs (target/debug/deps/hir_ty-b1d8e0b478de4a85)

running 1 test
test tests::regression::new_solver::regression_19957 ... FAILED

failures:

---- tests::regression::new_solver::regression_19957 stdout ----

thread 'tests::regression::new_solver::regression_19957' (139607) panicked at crates/hir-ty/src/tests.rs:246:9:
Unexpected type mismatches:
1109..1141: expected Pin<Box<dyn Future<Output = i32> + Send + 'async_trait>>, got Pin<Box<impl Future<Output = i32>>>

note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace


failures:
    tests::regression::new_solver::regression_19957

test result: FAILED. 0 passed; 1 failed; 0 ignored; 0 measured; 916 filtered out; finished in 0.07s

error: test failed, to rerun pass `-p hir-ty --lib`

3. On the latest commit

Result:

   Compiling hir-ty v0.0.0 (/tmp/test-old/crates/hir-ty)
    Finished `test` profile [unoptimized + debuginfo] target(s) in 46.58s
     Running unittests src/lib.rs (target/debug/deps/hir_ty-da16ccfda9040153)

running 1 test
test tests::regression::new_solver::regression_19957 ... FAILED

failures:

---- tests::regression::new_solver::regression_19957 stdout ----

thread 'tests::regression::new_solver::regression_19957' (150186) panicked at crates/hir-ty/src/tests.rs:246:9:
Unexpected type mismatches:
1109..1141: expected Pin<Box<dyn Future<Output = i32> + Send + 'async_trait>>, got Pin<Box<impl Future<Output = i32>>>

note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace


failures:
    tests::regression::new_solver::regression_19957

test result: FAILED. 0 passed; 1 failed; 0 ignored; 0 measured; 920 filtered out; finished in 0.05s

error: test failed, to rerun pass `-p hir-ty --lib`

The results are consistent showing the original issue #19957 is not actually solved yet contrasting what said in #19957 (comment)

Additional test which I did on #19957 (comment) code snippet and got the similar result.

[ShoyuVanilla]

Looks great! Thanks! I left a nit which fixes the CI failure

[ShoyuVanilla]

This should be removed, I guess? 😄

[Aditya-PS-05]

As the test will fail, this will still make the CI to fail.

[ShoyuVanilla]

Oh, I missed your comment above and I only looked into the tidy test failure. Yeah, this seems not fixed yet

Edit) Well, I've tested a bit more and this seems fixed. I guess we need to adjust some lang item imports and implementations in the test

[ShoyuVanilla]

Yeah, adding the following impl to minicore fixes the failure 😄

impl<Ptr, U> CoerceUnsized<Pin<U>> for Pin<Ptr>
where
    Ptr: CoerceUnsized<U>,
{
}

[Aditya-PS-05]

@ShoyuVanilla done.