Support assoc type defaults for assoc type constraints of trait objects

[ShoyuVanilla]

As stated in https://rust-lang.github.io/rfcs/2532-associated-type-defaults.html#trait-objects

cc: #29661, the fourth checkbox; "Implement the changes to object types specified in the RFC"

[rustbot]

HIR ty lowering was modified

cc @fmease

[rustbot]

r? @jackh726

rustbot has assigned @jackh726.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

Why was this reviewer chosen?

The reviewer was selected based on:

• Owners of files modified in this PR: compiler
• compiler expanded to 68 candidates
• Random selection from 15 candidates

[ShoyuVanilla]

The cycle here is unfortunate but I think this might be inevitable 🤔

[fmease]

Could you please also update the large comment above. I'm referring to the part where it says The user has to constrain all associated types & consts via bindings unless the corresponding associated item has a where Self: Sized clause.. Could you update it to say sth. along the lines of […] unless […] has a […] clause or it has a default value which we would then pick.

Thanks!

[fmease]

Could you please also add a corresponding test for type-level associated constants (feature min_generic_const_args (mGCA))? Please put it into tests/ui/const-generics/associated-const-bindings/ and name it something like dyn-compat-assoc-const-defaults.rs following the established naming convention.

It should look something like:

// Type associated consts needn't be specified in the trait object type
// if they have a default in which case we use said default.
//@ check-pass

#![feature(min_generic_const_args, associated_type_defaults)]
#![expect(incomplete_features)]

trait Trait {
    type const N: usize = 1;
}

fn main() {
    // `Trait::N` defaults to `1`.
    let _: dyn Trait;
    let _: [(); <dyn Trait as Trait>::N] = [(); 1];
}

(The the phrasing of the description is admittedly rather awkward, feel free to improve it)

(I've intentionally proactively added associated_type_defaults since that will become necessary for such type-level assoc consts once PR #152385 is merged.)

[fmease]

I'm not entirely sure if your current approach can handle the following code that was suggested in similar form in the RFC:

#![feature(associated_type_defaults)]

trait Trait {
    type T = Self::U;
    type U;
}

fn main() {
    let _: dyn Trait<U = ()>; // `Trait::T` should default to `()`
}

I have a hunch that this might not work right now but I might be mistaken. I think this will get rejected with the value of associated type T must be specified similar to:

trait Trait: SuperTrait<T = Self::U> { type U; }
trait SuperTrait { type T; }

fn main() {
    let _: dyn Trait<U = ()>;
    //~^ ERROR the value of the associated type `T` in `SuperTrait` must be specified
}

In any case, encapsulating this as a UI test would be nice I think.