Report unconstrained region in hidden types lazily by adwinwhite · Pull Request #152218 · rust-lang/rust

adwinwhite · 2026-02-06T12:37:16Z

Fixes rust-lang/trait-system-refactor-initiative#264

I didn't copy the mechanism of HIR typeck as I found that we just need to be lax in the unconstrained region case.
It already ignores non-defining uses and invalid params.

About tests, I'm having trouble coming up with more complex ones. 🙁

This fixes ukanren and codecrafters-redis-rust but not #151322 and #151323.
I believe they are a different problem.

r? @lcnr

adwinwhite · 2026-02-06T12:41:24Z

tests/ui/traits/next-solver/opaques/eventually-constrained-region.rs

+
+type MultiUse2 = impl Sized;
+#[define_opaque(MultiUse2)]
+fn capture_different_universals_not_on_bounds<'a, 'b, 'c>(a: &'a i32, b: &'b i32, c: &'c i32)  {


Just make sure that we report error for each invalid capture.
In my other attempt, only one unconstrained error per opaque definition is reported.

hmm, I personally don't actually think this is too important and we do only report one error in HIR typeck 🤔

please move the check-pass regression test into a separate file

lcnr

I am slightly unhappy about having both a hidden_types and unconstrained_hidden_types field but think it's a fairly clean solution.

The alternative would be to change hidden_types to an enum with a variant for unconstrained_hidden_types (either a single one or a list), but I don't think that's necessarily better 🤔

View changes since this review

lcnr · 2026-02-09T14:13:23Z

tests/ui/traits/next-solver/opaques/eventually-constrained-region.rs

+
+type MultiUse2 = impl Sized;
+#[define_opaque(MultiUse2)]
+fn capture_different_universals_not_on_bounds<'a, 'b, 'c>(a: &'a i32, b: &'b i32, c: &'c i32)  {


hmm, I personally don't actually think this is too important and we do only report one error in HIR typeck 🤔

lcnr · 2026-02-09T14:13:52Z

tests/ui/traits/next-solver/opaques/eventually-constrained-region.rs

+
+type MultiUse2 = impl Sized;
+#[define_opaque(MultiUse2)]
+fn capture_different_universals_not_on_bounds<'a, 'b, 'c>(a: &'a i32, b: &'b i32, c: &'c i32)  {


please move the check-pass regression test into a separate file

lcnr · 2026-02-09T14:16:15Z

compiler/rustc_borrowck/src/root_cx.rs

+                unreachable!("the body should depend on opaques type if it has opaque use");
+            };
+            result.deferred_opaque_type_errors.push(error);
+        }


is there a reason you're return a list of errors instead of eagerly adding them t deferred_opaque_type_errors in detect_unconstrained_hidden_types (i guess then called handle_unconstrained_hidden_types)

also, why the def_id check here? 🤔 it's not immediately clear to me whether this assert is important. We could avoid some code if we don't check htis as the def_id is otherwise unused

Ah, you're right. Eagerly adding the errors in detect_unconstrained_hidden_types is cleaner.

The def_id is needed because we need to insert error to the body where it originates from. The region var only makes sense in its own infcx. Yeah, I should have comments for this.

lcnr · 2026-02-09T14:17:56Z

compiler/rustc_borrowck/src/root_cx.rs

    pub tcx: TyCtxt<'tcx>,
    root_def_id: LocalDefId,
    hidden_types: FxIndexMap<LocalDefId, ty::DefinitionSiteHiddenType<'tcx>>,
+    unconstrained_hidden_types: Vec<UnexpectedHiddenRegion<'tcx>>,


comment please

rustbot · 2026-02-10T04:23:10Z

This PR was rebased onto a different main commit. Here's a range-diff highlighting what actually changed.

Rebasing is a normal part of keeping PRs up to date, so no action is needed—this note is just to help reviewers.

lcnr · 2026-02-10T16:29:41Z

tests/ui/traits/next-solver/opaques/eventually-constrained-region.rs

+
+// Some defining uses of opaque types can't constrain captured regions to universals.
+// Previouly, we eagerly report error in this case.
+// Now we report error only if there's no fully defining use from all bodies of the typeck root.


link the issue this is a regression test for

lcnr · 2026-02-12T16:36:06Z

tests/ui/traits/next-solver/opaques/eventually-constrained-region.rs

+
+#![feature(type_alias_impl_trait)]
+
+struct Invar<'a>(*mut &'a ());


Suggested change

struct Invar<'a>(*mut &'a ());

struct Inv<'a>(*mut &'a ());

sorry, this is the convention we generally use and Invar is confusing to me. Please also change this test to use RPIT instead of tait

lcnr · 2026-02-12T16:42:23Z

compiler/rustc_borrowck/src/region_infer/opaque_types/mod.rs

+/// We handle `UnexpectedHiddenRegion` error lazily in the next solver.
+/// Hope that it can be resolved by another fully defining use, which may not exist.
+/// We detect eventually unconstrained region via this function after collecting defining uses from all
+/// bodies.


Suggested change

/// We handle `UnexpectedHiddenRegion` error lazily in the next solver.

/// Hope that it can be resolved by another fully defining use, which may not exist.

/// We detect eventually unconstrained region via this function after collecting defining uses from all

/// bodies.

/// We handle `UnexpectedHiddenRegion` error lazily in the next solver as

/// there may be a fully defining use in another body.

///

/// In case such a defining use does not exist, we register an error here.

lcnr

nits, then r=me

View changes since this review

adwinwhite · 2026-02-13T05:27:45Z

@bors r=lcnr rollup

rust-bors · 2026-02-13T05:27:48Z

📌 Commit 20c46d6 has been approved by lcnr

It is now in the queue for this repository.

@lcnr

…-handling-keep-all-errors, r=lcnr Report unconstrained region in hidden types lazily Fixes rust-lang/trait-system-refactor-initiative#264 I didn't copy the mechanism of HIR typeck as I found that we just need to be lax in the unconstrained region case. It already ignores non-defining uses and invalid params. About tests, I'm having trouble coming up with more complex ones. 🙁 This fixes `ukanren` and `codecrafters-redis-rust` but not rust-lang#151322 and rust-lang#151323. I believe they are a [different problem](rust-lang#151322 (comment)). r? @lcnr

@lcnr

…-handling-keep-all-errors, r=lcnr Report unconstrained region in hidden types lazily Fixes rust-lang/trait-system-refactor-initiative#264 I didn't copy the mechanism of HIR typeck as I found that we just need to be lax in the unconstrained region case. It already ignores non-defining uses and invalid params. About tests, I'm having trouble coming up with more complex ones. 🙁 This fixes `ukanren` and `codecrafters-redis-rust` but not rust-lang#151322 and rust-lang#151323. I believe they are a [different problem](rust-lang#151322 (comment)). r? @lcnr

…uwer Rollup of 14 pull requests Successful merges: - #152323 (Fix ICE in borrowck when recovering `fn_sig` for `-> _`) - #152469 (Remove unused features) - #152515 (Extract `DepKindVTable` constructors to their own module) - #152555 (Port `#[rustc_diagnostic_item]` to the new attribute parsers) - #152218 (Report unconstrained region in hidden types lazily) - #152356 (Improve the `inline_fluent!` macro) - #152392 (Fix ICE in supertrait_vtable_slot when supertrait has missing generics) - #152407 (Add regression test for type_const with unit struct ctor under mGCA) - #152440 (Fix typos and grammar in compiler and build documentation) - #152536 (bootstrap: add explicit UTF-8 encoding to text-mode open() calls) - #152554 (Remove `deprecated_safe` and its corresponding feature gate) - #152556 (doc: move riscv64a23-unknown-linux-gnu to tier 2) - #152563 (Replace "bug" with "issue" in triagebot ping messages) - #152565 (fix missleading error for tuple ctor) Failed merges: - #152512 (core: Implement feature `float_exact_integer_constants`) - #152296 (Port `rust_nonnull_optimization_guaranteed` and `rustc_do_not_const_check` to the new attribute parser)

@lcnr

Rollup merge of #152218 - adwinwhite:fix-mir-borrowck-opaque-handling-keep-all-errors, r=lcnr Report unconstrained region in hidden types lazily Fixes rust-lang/trait-system-refactor-initiative#264 I didn't copy the mechanism of HIR typeck as I found that we just need to be lax in the unconstrained region case. It already ignores non-defining uses and invalid params. About tests, I'm having trouble coming up with more complex ones. 🙁 This fixes `ukanren` and `codecrafters-redis-rust` but not #151322 and #151323. I believe they are a [different problem](#151322 (comment)). r? @lcnr

rustbot assigned lcnr Feb 6, 2026

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. WG-trait-system-refactor The Rustc Trait System Refactor Initiative (-Znext-solver) labels Feb 6, 2026

adwinwhite commented Feb 6, 2026

View reviewed changes

adwinwhite marked this pull request as draft February 6, 2026 13:28

rustbot added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Feb 6, 2026

adwinwhite marked this pull request as ready for review February 7, 2026 15:07

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. and removed S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. labels Feb 7, 2026

lcnr reviewed Feb 9, 2026

View reviewed changes

adwinwhite force-pushed the fix-mir-borrowck-opaque-handling-keep-all-errors branch from a509e96 to 03e0020 Compare February 10, 2026 04:23

This comment has been minimized.

Sign in to view

adwinwhite force-pushed the fix-mir-borrowck-opaque-handling-keep-all-errors branch from 03e0020 to 7b1cc81 Compare February 10, 2026 05:29

lcnr reviewed Feb 10, 2026

View reviewed changes

adwinwhite force-pushed the fix-mir-borrowck-opaque-handling-keep-all-errors branch from 7b1cc81 to 10301d5 Compare February 11, 2026 04:54

lcnr reviewed Feb 12, 2026

View reviewed changes

adwinwhite force-pushed the fix-mir-borrowck-opaque-handling-keep-all-errors branch from 10301d5 to ae57df8 Compare February 13, 2026 03:33

report unconstrained region in hidden types lazily

20c46d6

adwinwhite force-pushed the fix-mir-borrowck-opaque-handling-keep-all-errors branch from ae57df8 to 20c46d6 Compare February 13, 2026 03:39

rust-bors bot added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Feb 13, 2026

JonathanBrouwer mentioned this pull request Feb 13, 2026

Rollup of 14 pull requests #152571

Closed

This was referenced Feb 13, 2026

Rollup of 14 pull requests #152572

Closed

Rollup of 14 pull requests #152574

Merged

rust-bors bot merged commit 98e8f99 into rust-lang:main Feb 14, 2026
11 checks passed

rustbot added this to the 1.95.0 milestone Feb 14, 2026


		#![feature(type_alias_impl_trait)]

		struct Invar<'a>(*mut &'a ());

Uh oh!

Conversation

adwinwhite commented Feb 6, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lcnr left a comment • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lcnr Feb 9, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

rustbot commented Feb 10, 2026

Uh oh!

This comment has been minimized.

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lcnr left a comment • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

adwinwhite commented Feb 13, 2026

Uh oh!

rust-bors bot commented Feb 13, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

adwinwhite commented Feb 6, 2026 •

edited

Loading

lcnr left a comment •

edited by rustbot

Loading

lcnr Feb 9, 2026 •

edited

Loading

lcnr left a comment •

edited by rustbot

Loading