layout_of unexpected rigid alias delayed bug

[lcnr]

fixes#152545. The trait solver can keep aliases as rigid even if they are not well-formed

rust/compiler/rustc_next_trait_solver/src/solve/normalizes_to/mod.rs

Lines 315 to 345 in d7daac0

	if !cx.has_item_definition(target_item_def_id){
	// If the impl is missing an item, it's either because the user forgot to
	// provide it, or the user is not *obligated* to provide it (because it
	// has a trivially false `Sized` predicate). If it's the latter, we cannot
	// delay a bug because we can have trivially false where clauses, so we
	// treat it as rigid.
	if cx.impl_self_is_guaranteed_unsized(impl_def_id){
	match ecx.typing_mode(){
	// Trying to normalize such associated items is always ambiguous
	// during coherence to avoid cyclic reasoning. See the example in
	// tests/ui/traits/trivial-unsized-projection-in-coherence.rs.
	//
	// As this ambiguity is unavoidable we emit a nested ambiguous
	// goal instead of using `Certainty::AMBIGUOUS`. This allows us to
	// return the nested goals to the parent `AliasRelate` goal. This
	// would be relevant if any of the nested goals refer to the `term`.
	// This is not the case here and we only prefer adding an ambiguous
	// nested goal for consistency.
	ty::TypingMode::Coherence => {
	ecx.add_goal(GoalSource::Misc, goal.with(cx,PredicateKind::Ambiguous));
	returnthen(ecx,Certainty::Yes);
	}
	ty::TypingMode::Analysis{ .. }
	| ty::TypingMode::Borrowck{ .. }
	| ty::TypingMode::PostBorrowckAnalysis{ .. }
	| ty::TypingMode::PostAnalysis => {
	ecx.structurally_instantiate_normalizes_to_term(
	goal,
	goal.predicate.alias,
	);
	returnthen(ecx,Certainty::Yes);

r? types

[jackh726]

@bors r+ rollup

[rust-bors]

📌 Commit b3d9fbc has been approved by jackh726

It is now in the queue for this repository.

[apiraino]

Beta backport accepted as per compiler team on Zulip. A backport PR will be authored by the release team at the end of the current development cycle. Backport labels handled by them.

@rustbot label +beta-accepted

[lcnr]

fixes#152545. The trait solver can keep aliases as rigid even if they are not well-formed

rust/compiler/rustc_next_trait_solver/src/solve/normalizes_to/mod.rs

Lines 315 to 345 in d7daac0

	if !cx.has_item_definition(target_item_def_id){
	// If the impl is missing an item, it's either because the user forgot to
	// provide it, or the user is not *obligated* to provide it (because it
	// has a trivially false `Sized` predicate). If it's the latter, we cannot
	// delay a bug because we can have trivially false where clauses, so we
	// treat it as rigid.
	if cx.impl_self_is_guaranteed_unsized(impl_def_id){
	match ecx.typing_mode(){
	// Trying to normalize such associated items is always ambiguous
	// during coherence to avoid cyclic reasoning. See the example in
	// tests/ui/traits/trivial-unsized-projection-in-coherence.rs.
	//
	// As this ambiguity is unavoidable we emit a nested ambiguous
	// goal instead of using `Certainty::AMBIGUOUS`. This allows us to
	// return the nested goals to the parent `AliasRelate` goal. This
	// would be relevant if any of the nested goals refer to the `term`.
	// This is not the case here and we only prefer adding an ambiguous
	// nested goal for consistency.
	ty::TypingMode::Coherence => {
	ecx.add_goal(GoalSource::Misc, goal.with(cx,PredicateKind::Ambiguous));
	returnthen(ecx,Certainty::Yes);
	}
	ty::TypingMode::Analysis{ .. }
	| ty::TypingMode::Borrowck{ .. }
	| ty::TypingMode::PostBorrowckAnalysis{ .. }
	| ty::TypingMode::PostAnalysis => {
	ecx.structurally_instantiate_normalizes_to_term(
	goal,
	goal.predicate.alias,
	);
	returnthen(ecx,Certainty::Yes);

r? types

[jackh726]

@bors r+ rollup

[rust-bors]

📌 Commit b3d9fbc has been approved by jackh726

It is now in the queue for this repository.

[apiraino]

Beta backport accepted as per compiler team on Zulip. A backport PR will be authored by the release team at the end of the current development cycle. Backport labels handled by them.

@rustbot label +beta-accepted

[lcnr]

fixes#152545. The trait solver can keep aliases as rigid even if they are not well-formed

rust/compiler/rustc_next_trait_solver/src/solve/normalizes_to/mod.rs

Lines 315 to 345 in d7daac0

	if !cx.has_item_definition(target_item_def_id){
	// If the impl is missing an item, it's either because the user forgot to
	// provide it, or the user is not *obligated* to provide it (because it
	// has a trivially false `Sized` predicate). If it's the latter, we cannot
	// delay a bug because we can have trivially false where clauses, so we
	// treat it as rigid.
	if cx.impl_self_is_guaranteed_unsized(impl_def_id){
	match ecx.typing_mode(){
	// Trying to normalize such associated items is always ambiguous
	// during coherence to avoid cyclic reasoning. See the example in
	// tests/ui/traits/trivial-unsized-projection-in-coherence.rs.
	//
	// As this ambiguity is unavoidable we emit a nested ambiguous
	// goal instead of using `Certainty::AMBIGUOUS`. This allows us to
	// return the nested goals to the parent `AliasRelate` goal. This
	// would be relevant if any of the nested goals refer to the `term`.
	// This is not the case here and we only prefer adding an ambiguous
	// nested goal for consistency.
	ty::TypingMode::Coherence => {
	ecx.add_goal(GoalSource::Misc, goal.with(cx,PredicateKind::Ambiguous));
	returnthen(ecx,Certainty::Yes);
	}
	ty::TypingMode::Analysis{ .. }
	| ty::TypingMode::Borrowck{ .. }
	| ty::TypingMode::PostBorrowckAnalysis{ .. }
	| ty::TypingMode::PostAnalysis => {
	ecx.structurally_instantiate_normalizes_to_term(
	goal,
	goal.predicate.alias,
	);
	returnthen(ecx,Certainty::Yes);

r? types

[jackh726]

@bors r+ rollup

[rust-bors]

📌 Commit b3d9fbc has been approved by jackh726

It is now in the queue for this repository.

[apiraino]

Beta backport accepted as per compiler team on Zulip. A backport PR will be authored by the release team at the end of the current development cycle. Backport labels handled by them.

@rustbot label +beta-accepted

[lcnr]

fixes#152545. The trait solver can keep aliases as rigid even if they are not well-formed

rust/compiler/rustc_next_trait_solver/src/solve/normalizes_to/mod.rs

Lines 315 to 345 in d7daac0

	if !cx.has_item_definition(target_item_def_id){
	// If the impl is missing an item, it's either because the user forgot to
	// provide it, or the user is not *obligated* to provide it (because it
	// has a trivially false `Sized` predicate). If it's the latter, we cannot
	// delay a bug because we can have trivially false where clauses, so we
	// treat it as rigid.
	if cx.impl_self_is_guaranteed_unsized(impl_def_id){
	match ecx.typing_mode(){
	// Trying to normalize such associated items is always ambiguous
	// during coherence to avoid cyclic reasoning. See the example in
	// tests/ui/traits/trivial-unsized-projection-in-coherence.rs.
	//
	// As this ambiguity is unavoidable we emit a nested ambiguous
	// goal instead of using `Certainty::AMBIGUOUS`. This allows us to
	// return the nested goals to the parent `AliasRelate` goal. This
	// would be relevant if any of the nested goals refer to the `term`.
	// This is not the case here and we only prefer adding an ambiguous
	// nested goal for consistency.
	ty::TypingMode::Coherence => {
	ecx.add_goal(GoalSource::Misc, goal.with(cx,PredicateKind::Ambiguous));
	returnthen(ecx,Certainty::Yes);
	}
	ty::TypingMode::Analysis{ .. }
	| ty::TypingMode::Borrowck{ .. }
	| ty::TypingMode::PostBorrowckAnalysis{ .. }
	| ty::TypingMode::PostAnalysis => {
	ecx.structurally_instantiate_normalizes_to_term(
	goal,
	goal.predicate.alias,
	);
	returnthen(ecx,Certainty::Yes);

r? types

[jackh726]

@bors r+ rollup

[rust-bors]

📌 Commit b3d9fbc has been approved by jackh726

It is now in the queue for this repository.

[apiraino]

Beta backport accepted as per compiler team on Zulip. A backport PR will be authored by the release team at the end of the current development cycle. Backport labels handled by them.

@rustbot label +beta-accepted

[lcnr]

fixes#152545. The trait solver can keep aliases as rigid even if they are not well-formed

rust/compiler/rustc_next_trait_solver/src/solve/normalizes_to/mod.rs

Lines 315 to 345 in d7daac0

	if !cx.has_item_definition(target_item_def_id){
	// If the impl is missing an item, it's either because the user forgot to
	// provide it, or the user is not *obligated* to provide it (because it
	// has a trivially false `Sized` predicate). If it's the latter, we cannot
	// delay a bug because we can have trivially false where clauses, so we
	// treat it as rigid.
	if cx.impl_self_is_guaranteed_unsized(impl_def_id){
	match ecx.typing_mode(){
	// Trying to normalize such associated items is always ambiguous
	// during coherence to avoid cyclic reasoning. See the example in
	// tests/ui/traits/trivial-unsized-projection-in-coherence.rs.
	//
	// As this ambiguity is unavoidable we emit a nested ambiguous
	// goal instead of using `Certainty::AMBIGUOUS`. This allows us to
	// return the nested goals to the parent `AliasRelate` goal. This
	// would be relevant if any of the nested goals refer to the `term`.
	// This is not the case here and we only prefer adding an ambiguous
	// nested goal for consistency.
	ty::TypingMode::Coherence => {
	ecx.add_goal(GoalSource::Misc, goal.with(cx,PredicateKind::Ambiguous));
	returnthen(ecx,Certainty::Yes);
	}
	ty::TypingMode::Analysis{ .. }
	| ty::TypingMode::Borrowck{ .. }
	| ty::TypingMode::PostBorrowckAnalysis{ .. }
	| ty::TypingMode::PostAnalysis => {
	ecx.structurally_instantiate_normalizes_to_term(
	goal,
	goal.predicate.alias,
	);
	returnthen(ecx,Certainty::Yes);

r? types

[jackh726]

@bors r+ rollup

[rust-bors]

📌 Commit b3d9fbc has been approved by jackh726

It is now in the queue for this repository.

[apiraino]

Beta backport accepted as per compiler team on Zulip. A backport PR will be authored by the release team at the end of the current development cycle. Backport labels handled by them.

@rustbot label +beta-accepted

[lcnr]

fixes#152545. The trait solver can keep aliases as rigid even if they are not well-formed

rust/compiler/rustc_next_trait_solver/src/solve/normalizes_to/mod.rs

Lines 315 to 345 in d7daac0

	if !cx.has_item_definition(target_item_def_id){
	// If the impl is missing an item, it's either because the user forgot to
	// provide it, or the user is not *obligated* to provide it (because it
	// has a trivially false `Sized` predicate). If it's the latter, we cannot
	// delay a bug because we can have trivially false where clauses, so we
	// treat it as rigid.
	if cx.impl_self_is_guaranteed_unsized(impl_def_id){
	match ecx.typing_mode(){
	// Trying to normalize such associated items is always ambiguous
	// during coherence to avoid cyclic reasoning. See the example in
	// tests/ui/traits/trivial-unsized-projection-in-coherence.rs.
	//
	// As this ambiguity is unavoidable we emit a nested ambiguous
	// goal instead of using `Certainty::AMBIGUOUS`. This allows us to
	// return the nested goals to the parent `AliasRelate` goal. This
	// would be relevant if any of the nested goals refer to the `term`.
	// This is not the case here and we only prefer adding an ambiguous
	// nested goal for consistency.
	ty::TypingMode::Coherence => {
	ecx.add_goal(GoalSource::Misc, goal.with(cx,PredicateKind::Ambiguous));
	returnthen(ecx,Certainty::Yes);
	}
	ty::TypingMode::Analysis{ .. }
	| ty::TypingMode::Borrowck{ .. }
	| ty::TypingMode::PostBorrowckAnalysis{ .. }
	| ty::TypingMode::PostAnalysis => {
	ecx.structurally_instantiate_normalizes_to_term(
	goal,
	goal.predicate.alias,
	);
	returnthen(ecx,Certainty::Yes);

r? types

[jackh726]

@bors r+ rollup

[rust-bors]

📌 Commit b3d9fbc has been approved by jackh726

It is now in the queue for this repository.

[apiraino]

Beta backport accepted as per compiler team on Zulip. A backport PR will be authored by the release team at the end of the current development cycle. Backport labels handled by them.

@rustbot label +beta-accepted

[lcnr]

fixes#152545. The trait solver can keep aliases as rigid even if they are not well-formed

rust/compiler/rustc_next_trait_solver/src/solve/normalizes_to/mod.rs

Lines 315 to 345 in d7daac0

	if !cx.has_item_definition(target_item_def_id){
	// If the impl is missing an item, it's either because the user forgot to
	// provide it, or the user is not *obligated* to provide it (because it
	// has a trivially false `Sized` predicate). If it's the latter, we cannot
	// delay a bug because we can have trivially false where clauses, so we
	// treat it as rigid.
	if cx.impl_self_is_guaranteed_unsized(impl_def_id){
	match ecx.typing_mode(){
	// Trying to normalize such associated items is always ambiguous
	// during coherence to avoid cyclic reasoning. See the example in
	// tests/ui/traits/trivial-unsized-projection-in-coherence.rs.
	//
	// As this ambiguity is unavoidable we emit a nested ambiguous
	// goal instead of using `Certainty::AMBIGUOUS`. This allows us to
	// return the nested goals to the parent `AliasRelate` goal. This
	// would be relevant if any of the nested goals refer to the `term`.
	// This is not the case here and we only prefer adding an ambiguous
	// nested goal for consistency.
	ty::TypingMode::Coherence => {
	ecx.add_goal(GoalSource::Misc, goal.with(cx,PredicateKind::Ambiguous));
	returnthen(ecx,Certainty::Yes);
	}
	ty::TypingMode::Analysis{ .. }
	| ty::TypingMode::Borrowck{ .. }
	| ty::TypingMode::PostBorrowckAnalysis{ .. }
	| ty::TypingMode::PostAnalysis => {
	ecx.structurally_instantiate_normalizes_to_term(
	goal,
	goal.predicate.alias,
	);
	returnthen(ecx,Certainty::Yes);

r? types

[jackh726]

@bors r+ rollup

[rust-bors]

📌 Commit b3d9fbc has been approved by jackh726

It is now in the queue for this repository.

[apiraino]

Beta backport accepted as per compiler team on Zulip. A backport PR will be authored by the release team at the end of the current development cycle. Backport labels handled by them.

@rustbot label +beta-accepted

[lcnr]

fixes#152545. The trait solver can keep aliases as rigid even if they are not well-formed

rust/compiler/rustc_next_trait_solver/src/solve/normalizes_to/mod.rs

Lines 315 to 345 in d7daac0

	if !cx.has_item_definition(target_item_def_id){
	// If the impl is missing an item, it's either because the user forgot to
	// provide it, or the user is not *obligated* to provide it (because it
	// has a trivially false `Sized` predicate). If it's the latter, we cannot
	// delay a bug because we can have trivially false where clauses, so we
	// treat it as rigid.
	if cx.impl_self_is_guaranteed_unsized(impl_def_id){
	match ecx.typing_mode(){
	// Trying to normalize such associated items is always ambiguous
	// during coherence to avoid cyclic reasoning. See the example in
	// tests/ui/traits/trivial-unsized-projection-in-coherence.rs.
	//
	// As this ambiguity is unavoidable we emit a nested ambiguous
	// goal instead of using `Certainty::AMBIGUOUS`. This allows us to
	// return the nested goals to the parent `AliasRelate` goal. This
	// would be relevant if any of the nested goals refer to the `term`.
	// This is not the case here and we only prefer adding an ambiguous
	// nested goal for consistency.
	ty::TypingMode::Coherence => {
	ecx.add_goal(GoalSource::Misc, goal.with(cx,PredicateKind::Ambiguous));
	returnthen(ecx,Certainty::Yes);
	}
	ty::TypingMode::Analysis{ .. }
	| ty::TypingMode::Borrowck{ .. }
	| ty::TypingMode::PostBorrowckAnalysis{ .. }
	| ty::TypingMode::PostAnalysis => {
	ecx.structurally_instantiate_normalizes_to_term(
	goal,
	goal.predicate.alias,
	);
	returnthen(ecx,Certainty::Yes);

r? types

[jackh726]

@bors r+ rollup

[rust-bors]

📌 Commit b3d9fbc has been approved by jackh726

It is now in the queue for this repository.

[apiraino]

Beta backport accepted as per compiler team on Zulip. A backport PR will be authored by the release team at the end of the current development cycle. Backport labels handled by them.

@rustbot label +beta-accepted