Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Remove a HACK by instead inferring opaque types during expected/formal type checking #123864

Merged
merged 2 commits into from
Apr 15, 2024
Merged

Remove a HACK by instead inferring opaque types during expected/formal type checking #123864

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
0c60779
Avoid an intermediate Option and just create the resulting enum directly
oli-obk Apr 12, 2024
24653a5
Remove a HACK by instead inferring opaque types during expected/forma…
oli-obk Apr 12, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
26 changes: 0 additions & 26 deletions compiler/rustc_hir_typeck/src/fn_ctxt/_impl.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -711,32 +711,6 @@ impl<'a, 'tcx> FnCtxt<'a, 'tcx> {
let formal_ret = self.resolve_vars_with_obligations(formal_ret);
let ret_ty = expected_ret.only_has_type(self)?;

// HACK(oli-obk): This is a hack to keep RPIT and TAIT in sync wrt their behaviour.
// Without it, the inference
// variable will get instantiated with the opaque type. The inference variable often
// has various helpful obligations registered for it that help closures figure out their
// signature. If we infer the inference var to the opaque type, the closure won't be able
// to find those obligations anymore, and it can't necessarily find them from the opaque
// type itself. We could be more powerful with inference if we *combined* the obligations
// so that we got both the obligations from the opaque type and the ones from the inference
// variable. That will accept more code than we do right now, so we need to carefully consider
// the implications.
// Note: this check is pessimistic, as the inference type could be matched with something other
// than the opaque type, but then we need a new `TypeRelation` just for this specific case and
// can't re-use `sup` below.
// See tests/ui/impl-trait/hidden-type-is-opaque.rs and
// tests/ui/impl-trait/hidden-type-is-opaque-2.rs for examples that hit this path.
if formal_ret.has_infer_types() {
for ty in ret_ty.walk() {
if let ty::GenericArgKind::Type(ty) = ty.unpack()
&& let ty::Alias(ty::Opaque, ty::AliasTy { def_id, .. }) = *ty.kind()
&& self.can_define_opaque_ty(def_id)
{
return None;
}
}
}

let expect_args = self
.fudge_inference_if_ok(|| {
let ocx = ObligationCtxt::new(self);
Expand Down
16 changes: 6 additions & 10 deletions compiler/rustc_hir_typeck/src/fn_ctxt/checks.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -297,22 +297,18 @@ impl<'a, 'tcx> FnCtxt<'a, 'tcx> {
// 3. Check if the formal type is a supertype of the checked one
// and register any such obligations for future type checks
let supertype_error = self.at(&self.misc(provided_arg.span), self.param_env).sup(
DefineOpaqueTypes::No,
DefineOpaqueTypes::Yes,
formal_input_ty,
coerced_ty,
);
let subtyping_error = match supertype_error {

// If neither check failed, the types are compatible
match supertype_error {
Ok(InferOk { obligations, value: () }) => {
self.register_predicates(obligations);
None
Compatibility::Compatible
}
Err(err) => Some(err),
};

// If neither check failed, the types are compatible
match subtyping_error {
None => Compatibility::Compatible,
Some(_) => Compatibility::Incompatible(subtyping_error),
Err(err) => Compatibility::Incompatible(Some(err)),
}
};

Expand Down
Loading