fatal error if both `-Znext-solver` and GCE are enabled by Dnreikronos · Pull Request #158738 · rust-lang/rust

Dnreikronos · 2026-07-03T14:59:16Z

the next solver already rejects generic_const_exprs with -Znext-solver=globally, but we can still walk into ConstKind::Expr while compiling after that error and crash anyway.

imo this should stay as a recovery fix, not a gce implementation. ambiguity still lgtm to me here: idk that we want to say the expr is solved when the feature combo is already invalid, we just need to not ice after the real error.

long-term, proper ConstKind::Expr support would be a much bigger thing. the next solver would need to type-check, normalize, compare, and evaluate or prove these exprs structurally, including nested binops, unops, casts, calls, and the result type for ConstArgHasType. that's out of scope for this pr.

this teaches both const evaluatable and const arg has type paths to recover on ConstKind::Expr, then adds the union repro as a ui test.

rustbot · 2026-07-03T14:59:21Z

r? @mejrs

rustbot has assigned @mejrs.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

Why was this reviewer chosen?

The reviewer was selected based on:

Owners of files modified in this PR: compiler
compiler expanded to 75 candidates
Random selection from 21 candidates

mejrs · 2026-07-03T18:51:48Z

r? @BoxyUwU

rustbot · 2026-07-03T18:51:53Z

BoxyUwU is currently at their maximum review capacity.
They may take a while to respond.

BoxyUwU · 2026-07-08T12:06:11Z

I don't think we want to complicate the new solver with logic for supporting GCE stuff. In the short term generic_const_args and min_generic_const_args should obviate this feature 🤔 can you instead make the error about new solver and GCE being mutually incompatible be fatal so that compilation just ends rather than trucking on

BoxyUwU · 2026-07-08T12:06:21Z

@rustbot author

rustbot · 2026-07-08T12:06:25Z

Reminder, once the PR becomes ready for a review, use @rustbot ready.

rustbot · 2026-07-08T19:57:22Z

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.

Dnreikronos · 2026-07-08T19:59:23Z

@BoxyUwU updated this based on your feedback.

the -znext-solver=globally + generic_const_exprs incompatibility is fatal now, so we stop during feature gating before gce can lower into constkind::expr and reach the new solver. i also removed the solver-side ambiguity recovery. imo that is cleaner, since that path was basically handling a state we already know is unsupported.

i renamed the regression test from recovery to incompatibility, and changed the const-trait known-bug test to use -znext-solver=coherence so it keeps testing the const-trait diagnostics instead of getting swallowed by this fatal gate.

also rebased on latest origin/main and resolved the conflict in tests/ui/traits/const-traits/unsatisfied-const-trait-bound.stderr.

ltm if you want anything adjusted. idk if there is a better place for the test, but this felt like the least noisy split to me.

@rustbot ready

BoxyUwU · 2026-07-22T13:32:29Z

@bors r+ rollup

rust-bors · 2026-07-22T13:32:33Z

📌 Commit a5a918c has been approved by BoxyUwU

It is now in the queue for this repository.

…expr_recovery, r=BoxyUwU next_trait_solver: Recover from GCE const exprs fixes rust-lang#158428 the next solver already rejects `generic_const_exprs` with `-Znext-solver=globally`, but we can still walk into `ConstKind::Expr` while compiling after that error and crash anyway. imo this should stay as a recovery fix, not a gce implementation. ambiguity still lgtm to me here: idk that we want to say the expr is solved when the feature combo is already invalid, we just need to not ice after the real error. long-term, proper `ConstKind::Expr` support would be a much bigger thing. the next solver would need to type-check, normalize, compare, and evaluate or prove these exprs structurally, including nested binops, unops, casts, calls, and the result type for `ConstArgHasType`. that's out of scope for this pr. this teaches both const evaluatable and const arg has type paths to recover on `ConstKind::Expr`, then adds the union repro as a ui test.

…uwer Rollup of 12 pull requests Successful merges: - #159712 (Subtree sync for rustc_codegen_cranelift) - #155697 (Stabilize c-variadic function definitions) - #159285 (Simplify `apply_effects_in_range`) - #159596 (unify the AST repr of type const and const RHS) - #159607 (test: update riscv32e-registers.rs for LLVM 24 MC diagnostic changes) - #159659 (Move `Limit` out of `rustc_hir`) - #159707 (fix error when a dangling ref in a ManuallyDrop is used in a pattern) - #158479 (Reject static item as direct const generic arg) - #158738 (next_trait_solver: Recover from GCE const exprs) - #159451 (Remove config cloning in compiletest) - #159646 (Increase depth for float infer var fallback hack) - #159705 (bootstrap: Prefer `cfg!(not(test))` when skipping code paths during unit tests)

…uwer Rollup of 11 pull requests Successful merges: - #159712 (Subtree sync for rustc_codegen_cranelift) - #155697 (Stabilize c-variadic function definitions) - #159285 (Simplify `apply_effects_in_range`) - #159596 (unify the AST repr of type const and const RHS) - #159607 (test: update riscv32e-registers.rs for LLVM 24 MC diagnostic changes) - #159659 (Move `Limit` out of `rustc_hir`) - #159707 (fix error when a dangling ref in a ManuallyDrop is used in a pattern) - #158738 (next_trait_solver: Recover from GCE const exprs) - #159451 (Remove config cloning in compiletest) - #159646 (Increase depth for float infer var fallback hack) - #159705 (bootstrap: Prefer `cfg!(not(test))` when skipping code paths during unit tests)

Rollup merge of #158738 - Dnreikronos:next_trait_solver/gce_expr_recovery, r=BoxyUwU next_trait_solver: Recover from GCE const exprs fixes #158428 the next solver already rejects `generic_const_exprs` with `-Znext-solver=globally`, but we can still walk into `ConstKind::Expr` while compiling after that error and crash anyway. imo this should stay as a recovery fix, not a gce implementation. ambiguity still lgtm to me here: idk that we want to say the expr is solved when the feature combo is already invalid, we just need to not ice after the real error. long-term, proper `ConstKind::Expr` support would be a much bigger thing. the next solver would need to type-check, normalize, compare, and evaluate or prove these exprs structurally, including nested binops, unops, casts, calls, and the result type for `ConstArgHasType`. that's out of scope for this pr. this teaches both const evaluatable and const arg has type paths to recover on `ConstKind::Expr`, then adds the union repro as a ui test.

…uwer Rollup of 11 pull requests Successful merges: - rust-lang/rust#159712 (Subtree sync for rustc_codegen_cranelift) - rust-lang/rust#155697 (Stabilize c-variadic function definitions) - rust-lang/rust#159285 (Simplify `apply_effects_in_range`) - rust-lang/rust#159596 (unify the AST repr of type const and const RHS) - rust-lang/rust#159607 (test: update riscv32e-registers.rs for LLVM 24 MC diagnostic changes) - rust-lang/rust#159659 (Move `Limit` out of `rustc_hir`) - rust-lang/rust#159707 (fix error when a dangling ref in a ManuallyDrop is used in a pattern) - rust-lang/rust#158738 (next_trait_solver: Recover from GCE const exprs) - rust-lang/rust#159451 (Remove config cloning in compiletest) - rust-lang/rust#159646 (Increase depth for float infer var fallback hack) - rust-lang/rust#159705 (bootstrap: Prefer `cfg!(not(test))` when skipping code paths during unit tests)

rustbot assigned mejrs Jul 3, 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 Jul 3, 2026

rustbot assigned BoxyUwU and unassigned mejrs Jul 3, 2026

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 Jul 8, 2026

Recover from GCE const exprs in next solver

70bfa0a

Dnreikronos force-pushed the next_trait_solver/gce_expr_recovery branch from 027341c to 7645fac Compare July 8, 2026 19:57

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 Jul 8, 2026

This comment has been minimized.

Sign in to view

Reject GCE with the global next solver

6e45dd2

Dnreikronos force-pushed the next_trait_solver/gce_expr_recovery branch from 779ef3e to 6e45dd2 Compare July 9, 2026 18:11

BoxyUwU reviewed Jul 14, 2026

View reviewed changes

Comment thread tests/ui/traits/const-traits/unsatisfied-const-trait-bound.rs Outdated

Restore next solver flag in const trait test

a5a918c

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 Jul 22, 2026

JonathanBrouwer mentioned this pull request Jul 22, 2026

Rollup of 12 pull requests #159723

Closed

JonathanBrouwer mentioned this pull request Jul 22, 2026

Rollup of 11 pull requests #159728

Merged

rust-bors Bot merged commit ac24945 into rust-lang:main Jul 22, 2026
13 checks passed

rustbot added this to the 1.99.0 milestone Jul 22, 2026

lcnr changed the title ~~next_trait_solver: Recover from GCE const exprs~~ next_trait_solver: fatal error when encountering GCE const exprs Jul 24, 2026

lcnr changed the title ~~next_trait_solver: fatal error when encountering GCE const exprs~~ fatal error if both -Znext-solver and GCE are enabled Jul 24, 2026

BoxyUwU mentioned this pull request Jul 24, 2026

Full Const Generics rust-lang/rust-project-goals#100

Open

7 tasks

Uh oh!

Uh oh!

Conversation

Dnreikronos commented Jul 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rustbot commented Jul 3, 2026

Uh oh!

mejrs commented Jul 3, 2026

Uh oh!

rustbot commented Jul 3, 2026

Uh oh!

BoxyUwU commented Jul 8, 2026

Uh oh!

BoxyUwU commented Jul 8, 2026

Uh oh!

rustbot commented Jul 8, 2026

Uh oh!

rustbot commented Jul 8, 2026

Uh oh!

Dnreikronos commented Jul 8, 2026

Uh oh!

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

Uh oh!

BoxyUwU commented Jul 22, 2026

Uh oh!

rust-bors Bot commented Jul 22, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

Dnreikronos commented Jul 3, 2026 •

edited

Loading