Folding/`ReErased` cleanups #150861

nnethercote · 2026-01-09T06:28:01Z

Various cleanups I found while reading this code closely, mostly involving folding and the use of ReErased.

r? @lcnr

rustbot · 2026-01-09T06:28:04Z

Some changes occurred in src/tools/clippy

cc @rust-lang/clippy

Some changes occurred to the core trait solver

cc @rust-lang/initiative-trait-system-refactor

Some changes occurred to the CTFE / Miri interpreter

cc @rust-lang/miri

Some changes occurred to the CTFE machinery

cc @RalfJung, @oli-obk, @lcnr

rust-analyzer is developed in its own repository. If possible, consider making this change to rust-lang/rust-analyzer instead.

cc @rust-lang/rust-analyzer

lnicola · 2026-01-09T06:52:43Z

src/tools/rust-analyzer/crates/hir-ty/src/next_solver/solver.rs

        &self,
-        _dst: Ty<'db>,
        _src: Ty<'db>,
+        _dst: Ty<'db>,


Nb.: we're currently using an older version of rustc_next_trait_solver (from crates.io), but this is fine because it's a no-op.

lcnr · 2026-01-11T11:00:19Z

compiler/rustc_infer/src/infer/outlives/test_type_match.rs

    assert!(!outlives_predicate.has_escaping_bound_vars());
-    let erased_outlives_predicate = tcx.erase_and_anonymize_regions(outlives_predicate);
-    let outlives_ty = erased_outlives_predicate.skip_binder().0;
+    let outlives_ty = tcx.erase_and_anonymize_regions(outlives_predicate.skip_binder().0);


that impacts behavior. for<'a> Ty<'a, 'param> currently gets changed to Ty<'bound0, 'erased> but now stays as Ty<'a, 'erased> where bound0 and 'a are escaping bound region.

This doesn't matter as we need to special case the bound regions of the outlives predicate anyways and use a hashmap for them in MatchAgainstHigherRankedOutlives.

hmm. I am not too happy about this. I like having all regions in there anonymized, even if it doesn't matter. If we anonymize, we could use a SmallVec instead of a hashmap here 🤷 idk if this is in any way perf relevant, i'd guess not.

Please either revert this commit, or add a comment that we don't anonymize the escaping bound regions, but that this is fine as MatchAgainstHigherRankedOutlives handles them specially anyways

I'll just remove the change. I wasn't intending to change behaviour.

lcnr · 2026-01-11T11:02:31Z

compiler/rustc_hir_analysis/src/check/wfcheck.rs

                    let ty = tcx.type_of(variant.tail().did).instantiate_identity();
-                    let ty = tcx.erase_and_anonymize_regions(ty);
-                    assert!(!ty.has_infer());
+                    assert!(!ty.has_type_flags(TypeFlags::HAS_TY_INFER | TypeFlags::HAS_CT_INFER));


!ty.has_non_region_infer()? though 🤷 that assert is unnecessary, ty is the output of a query. It never has infer vars

Ok, I'll remove the assertion

lcnr · 2026-01-11T11:05:10Z

compiler/rustc_trait_selection/src/solve/delegate.rs

    ) -> Result<Certainty, NoSolution> {
-        // Erase regions because we compute layouts in `rustc_transmute`,
-        // which will ICE for region vars.
-        let (dst, src) = self.tcx.erase_and_anonymize_regions((dst, src));


it's not about looking at regions, we do call layout_of for types here though... surprised that removing this doesn't cause test failures.,

Do you encounter region vars here? if so, why does fn is_transmutable not ICE when calling Tree::from_ty

I'll just remove this commit.

lcnr · 2026-01-11T11:06:22Z

some thoughts, otherwise r=me

Because these folders only change regions. Note: `BottomUpFolder` folds all regions, while `fold_regions` skips some bound regions. But that's ok because these two folders only modify `ReVar`s.

The exact same call appears earlier in this function.

The only thing we do with the result is consult the `.def_id` field, which is unaffected by `erase_and_anonymize_regions`.

- Remove the vacuous `Types`, which provides extremely little value. - Make sure `src` comes before `dst` in all transmute-related functions. (Currently it's a mix: sometimes `src` is first, sometimes it is second`.)

It duplicates `has_erased_regions` from the compiler.

rustbot · 2026-01-11T22:23:07Z

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.

nnethercote · 2026-01-11T23:53:01Z

I removed the commits mentioned in the reviews, and a couple more involving needs_drops because there is an assert within that function that checks for erased regions. With those changes, this meets the r=me conditions above. Thanks for the review!

@bors r=lcnr

rust-bors · 2026-01-11T23:53:04Z

📌 Commit e7036b1 has been approved by lcnr

It is now in the queue for this repository.

@lcnr

Folding/`ReErased` cleanups Various cleanups I found while reading this code closely, mostly involving folding and the use of `ReErased`. r? @lcnr

@ghost

Rollup of 5 pull requests Successful merges: - #150861 (Folding/`ReErased` cleanups) - #150941 (rustc_parse_format: improve diagnostics for unsupported python numeric grouping) - #150972 (Rename EII attributes slightly (being consistent in naming things foreign items, not extern items)) - #150980 (Use updated indexes to build reverse map for delegation generics) - #150986 (std: Fix size returned by UEFI tcp4 read operations) r? @ghost

@ghost

Rollup of 8 pull requests Successful merges: - #150861 (Folding/`ReErased` cleanups) - #150869 (Emit error instead of delayed bug when meeting mismatch type for const tuple) - #150920 (Use a hook to decouple `rustc_mir_transform` from `rustc_mir_build`) - #150941 (rustc_parse_format: improve diagnostics for unsupported python numeric grouping) - #150972 (Rename EII attributes slightly (being consistent in naming things foreign items, not extern items)) - #150980 (Use updated indexes to build reverse map for delegation generics) - #150986 (std: Fix size returned by UEFI tcp4 read operations) - #150996 (Remove `S-waiting-on-bors` after a PR is merged) r? @ghost

@ghost

Rollup of 8 pull requests Successful merges: - #150861 (Folding/`ReErased` cleanups) - #150869 (Emit error instead of delayed bug when meeting mismatch type for const tuple) - #150920 (Use a hook to decouple `rustc_mir_transform` from `rustc_mir_build`) - #150941 (rustc_parse_format: improve diagnostics for unsupported python numeric grouping) - #150972 (Rename EII attributes slightly (being consistent in naming things foreign items, not extern items)) - #150980 (Use updated indexes to build reverse map for delegation generics) - #150986 (std: Fix size returned by UEFI tcp4 read operations) - #150996 (Remove `S-waiting-on-bors` after a PR is merged) r? @ghost

@lcnr

Rollup merge of #150861 - folding-cleanups, r=lcnr Folding/`ReErased` cleanups Various cleanups I found while reading this code closely, mostly involving folding and the use of `ReErased`. r? @lcnr

@ghost

Rollup of 8 pull requests Successful merges: - rust-lang/rust#150861 (Folding/`ReErased` cleanups) - rust-lang/rust#150869 (Emit error instead of delayed bug when meeting mismatch type for const tuple) - rust-lang/rust#150920 (Use a hook to decouple `rustc_mir_transform` from `rustc_mir_build`) - rust-lang/rust#150941 (rustc_parse_format: improve diagnostics for unsupported python numeric grouping) - rust-lang/rust#150972 (Rename EII attributes slightly (being consistent in naming things foreign items, not extern items)) - rust-lang/rust#150980 (Use updated indexes to build reverse map for delegation generics) - rust-lang/rust#150986 (std: Fix size returned by UEFI tcp4 read operations) - rust-lang/rust#150996 (Remove `S-waiting-on-bors` after a PR is merged) r? @ghost

rustbot assigned lcnr Jan 9, 2026

This comment has been minimized.

Sign in to view

lnicola reviewed Jan 9, 2026

View reviewed changes

lcnr reviewed Jan 11, 2026

View reviewed changes

lcnr closed this Jan 11, 2026

rustbot removed the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Jan 11, 2026

lcnr reopened this Jan 11, 2026

rustbot added the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Jan 11, 2026

This comment has been minimized.

Sign in to view

nnethercote force-pushed the folding-cleanups branch from b72108d to c3ceba7 Compare January 11, 2026 21:28

This comment has been minimized.

Sign in to view

nnethercote added 6 commits January 12, 2026 08:51

Replace two BottomUpFolders with fold_regions.

2ae2987

Because these folders only change regions. Note: `BottomUpFolder` folds all regions, while `fold_regions` skips some bound regions. But that's ok because these two folders only modify `ReVar`s.

Whitespace fixes.

71d1b2c

Remove redundant call to erase_and_anonymize_regions.

02e0879

The exact same call appears earlier in this function.

Remove useless call to erase_and_anonymize_regions.

5e51092

The only thing we do with the result is consult the `.def_id` field, which is unaffected by `erase_and_anonymize_regions`.

Clean up src/dst transmute mess.

46d8c2b

- Remove the vacuous `Types`, which provides extremely little value. - Make sure `src` comes before `dst` in all transmute-related functions. (Currently it's a mix: sometimes `src` is first, sometimes it is second`.)

clippy: remove ty_has_erased_regions.

e7036b1

It duplicates `has_erased_regions` from the compiler.

nnethercote force-pushed the folding-cleanups branch from c3ceba7 to e7036b1 Compare January 11, 2026 22:23

rust-bors bot added the S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. label Jan 11, 2026

rust-bors bot removed the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Jan 11, 2026

matthiaskrgr mentioned this pull request Jan 12, 2026

Rollup of 5 pull requests #150994

Closed

matthiaskrgr mentioned this pull request Jan 12, 2026

Rollup of 8 pull requests #151003

Merged

rust-bors bot merged commit 79445c3 into rust-lang:main Jan 12, 2026
11 checks passed

rustbot added this to the 1.94.0 milestone Jan 12, 2026

nnethercote deleted the folding-cleanups branch January 12, 2026 21:40

Uh oh!

Folding/ReErased cleanups #150861

Folding/ReErased cleanups #150861

Uh oh!

Conversation

nnethercote commented Jan 9, 2026

Uh oh!

rustbot commented Jan 9, 2026

Uh oh!

This comment has been minimized.

lnicola Jan 9, 2026

Choose a reason for hiding this comment

Uh oh!

lcnr Jan 11, 2026

Choose a reason for hiding this comment

Uh oh!

nnethercote Jan 11, 2026

Choose a reason for hiding this comment

Uh oh!

lcnr Jan 11, 2026

Choose a reason for hiding this comment

Uh oh!

nnethercote Jan 11, 2026

Choose a reason for hiding this comment

Uh oh!

lcnr Jan 11, 2026

Choose a reason for hiding this comment

Uh oh!

nnethercote Jan 11, 2026

Choose a reason for hiding this comment

Uh oh!

lcnr commented Jan 11, 2026

Uh oh!

This comment has been minimized.

This comment has been minimized.

rustbot commented Jan 11, 2026

Uh oh!

nnethercote commented Jan 11, 2026

Uh oh!

rust-bors bot commented Jan 11, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

Folding/`ReErased` cleanups #150861

Folding/`ReErased` cleanups #150861