Rollup merge of #118262 - spastorino:relate-iats-using-eq, r=compiler-errors
Relate Inherent Associated Types using eq
We should call `eq` instead of `sup` as we're relating `Ty` directly and not `Binder<TraitRef>`.
This is part of #118118 but unrelated to that PR.
r? `@compiler-errors` `@lcnr`
diff --git a/compiler/rustc_hir_analysis/src/astconv/mod.rs b/compiler/rustc_hir_analysis/src/astconv/mod.rs
index 102c837..ff92d4c 100644
--- a/compiler/rustc_hir_analysis/src/astconv/mod.rs
+++ b/compiler/rustc_hir_analysis/src/astconv/mod.rs
@@ -1674,10 +1674,7 @@ fn fold_const(
let impl_ty = ocx.normalize(&cause, param_env, impl_ty);
// Check that the self types can be related.
- // FIXME(inherent_associated_types): Should we use `eq` here? Method probing uses
- // `sup` for this situtation, too. What for? To constrain inference variables?
- if ocx.sup(&ObligationCause::dummy(), param_env, impl_ty, self_ty).is_err()
- {
+ if ocx.eq(&ObligationCause::dummy(), param_env, impl_ty, self_ty).is_err() {
return false;
}
