Function rustc_hir_analysis::check::compare_impl_item::collect_return_position_impl_trait_in_trait_tys

pub(super) fn collect_return_position_impl_trait_in_trait_tys<'tcx>(
    tcx: TyCtxt<'tcx>,
    impl_m_def_id: LocalDefId
) -> Result<&'tcx DefIdMap<EarlyBinder<Ty<'tcx>>>, ErrorGuaranteed>

Given a method def-id in an impl, compare the method signature of the impl against the trait that it’s implementing. In doing so, infer the hidden types that this method’s signature provides to satisfy each return-position impl Trait in the trait signature.

The method is also responsible for making sure that the hidden types for each RPITIT actually satisfy the bounds of the impl Trait, i.e. that if we infer impl Trait = Foo, that Foo: Trait holds.

For example, given the sample code:

use std::ops::Deref;

trait Foo {
    fn bar() -> impl Deref<Target = impl Sized>;
             // ^- RPITIT #1        ^- RPITIT #2
}

impl Foo for () {
    fn bar() -> Box<String> { Box::new(String::new()) }
}

The hidden types for the RPITITs in bar would be inferred to: * impl Deref (RPITIT #1) = Box<String> * impl Sized (RPITIT #2) = String

The relationship between these two types is straightforward in this case, but may be more tenuously connected via other impls and normalization rules for cases of more complicated nested RPITITs.