next solver: prefer to select impl candidates over global where-clause candidates - #162655

dianne · 2026-09-11T17:26:10Z

Ideally we'd use a more appropriate typing environment when doing const-eval for const-to-pat, which would also fix that (since the problem clauses wouldn't be present to begin with). Being able to do that seems kind of far off, though, so here's a quick fix that (mostly) matches what the old solver does.

dianne · 2026-09-11T17:28:31Z

-    // Don't winnow until `Certainty::Yes` -- we don't need to winnow until
-    // codegen, and only on the good path.
+    // Don't winnow until `Certainty::Yes` -- we don't need to winnow until constant evaluation or
+    // codegen.


not totally sure what to do with this comment, but seeing as it was wrong before, I'd like to update it in some way

View changes since the review

seems good enough.

select is slightly weird in general rn and will clean this up post next-solver stabilization https://github.com/rust-lang/trait-system-refactor-initiative/blob/main/notes/stabilization-report/proof-tree-visitors.md#select

dianne · 2026-09-11T17:42:14Z

+        (
+            CandidateSource::ParamEnv(ParamEnvSource::Global),
+            CandidateSource::Impl(_) | CandidateSource::BuiltinImpl(_),
+        ) => true,


maybe this could be more aggressive? I'm not sure what else would be needed, so I'm treating this as a targeted fix. so far, I haven't been able to find any examples that would still fail without this if const-to-pat resolved and evaluated consts in a clean environment (e.g. as it does when full gca is enabled)

View changes since the review

dianne · 2026-09-11T17:47:20Z

+#![feature(const_trait_impl)]
+#![feature(const_clone)]


this is for testing the preference for builtin impls over global where-clause candidates. I haven't found a way to do that without unstable features

View changes since the review

want to split this out into a separate test given that the other one doesn't rely on unstable features?

sounds good to me, yeah

dianne · 2026-09-11T17:48:25Z

+//@ revisions: current next
+//@ ignore-compare-mode-next-solver (explicit revisions)
+//@[next] compile-flags: -Znext-solver


could maybe get rid of the current-solver revision but I figured I'd include it as a sanity check

View changes since the review

dianne · 2026-09-11T18:54:06Z

r? @lcnr or reassign

rustbot · 2026-09-11T18:54:14Z

Some changes occurred to the core trait solver

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

lcnr

single nit, then r=me

thanks for working on this 😊

View changes since this review

dianne · 2026-09-28T15:55:17Z

nit addressed! I needed to come up with new test names and tweak the comments on the split-out test a bit to make up for missing context, but I'm going to assume that's minor enough to @bors r=lcnr

rust-bors · 2026-09-28T15:55:21Z

📌 Commit b6d0fb1 has been tentatively approved by lcnr

It will be put into the queue for this repository once PR CI succeeds.

…=lcnr next solver: prefer to select impl candidates over global where-clause candidates Fixes rust-lang#162331 Ideally we'd use a more appropriate typing environment when doing const-eval for const-to-pat, which would also fix that (since the problem clauses wouldn't be present to begin with). Being able to do that seems kind of far off, though, so here's a quick fix that (mostly) matches what the old solver does.

…uwer Rollup of 15 pull requests Successful merges: - #158936 (Add `std::fs::{Home|Media}Dirs`) - #129036 (Additional NonZero conversions) - #158997 (Avoid recording unnameable `extern crate` aliases in diagnostic metadata) - #161015 (Stabilize `funnel_shifts` (including `const`)) - #161712 (Stabilize `Result::into_{ok,err}`) - #162493 (Add support for -Zsanitizer-cfi-minimal-runtime) - #162655 (next solver: prefer to select impl candidates over global where-clause candidates) - #162862 (Fix intra doc link resolution when a doc comment is composed of both inner and outer doc comment) - #163200 (make `RustaceansAreAwesome` satisfy trait bounds) - #163331 (Move `Arc` and `Rc` into `rcs` mod) - #163427 (implement #![feature(gca_adts)]) - #163428 (do not complain about unstable target features on nightly) - #163444 (Add `stable_rustc` helper in `run-make-support`) - #163447 (Allow using different index types when reading and writing to tables) - #163450 (Force the correct type variable to never for method resolution on an adjusted never type)

…=lcnr next solver: prefer to select impl candidates over global where-clause candidates Fixes rust-lang#162331 Ideally we'd use a more appropriate typing environment when doing const-eval for const-to-pat, which would also fix that (since the problem clauses wouldn't be present to begin with). Being able to do that seems kind of far off, though, so here's a quick fix that (mostly) matches what the old solver does.

…uwer Rollup of 9 pull requests Successful merges: - #162655 (next solver: prefer to select impl candidates over global where-clause candidates) - #162832 (add `Div` and `Mul` for `Complex<{float}>`) - #162862 (Fix intra doc link resolution when a doc comment is composed of both inner and outer doc comment) - #163024 (Add `Dir` equivalents of `fs::metadata` & `fs::symlink_metadata`) - #163200 (make `RustaceansAreAwesome` satisfy trait bounds) - #163210 (std: split stack overflow module) - #163331 (Move `Arc` and `Rc` into `rcs` mod) - #163183 (Add .seek_read_buf_exact() to std::os::windows::fs::FileExt) - #163450 (Force the correct type variable to never for method resolution on an adjusted never type)

…uwer Rollup of 9 pull requests Successful merges: - #158997 (Avoid recording unnameable `extern crate` aliases in diagnostic metadata) - #162655 (next solver: prefer to select impl candidates over global where-clause candidates) - #162832 (add `Div` and `Mul` for `Complex<{float}>`) - #162862 (Fix intra doc link resolution when a doc comment is composed of both inner and outer doc comment) - #163200 (make `RustaceansAreAwesome` satisfy trait bounds) - #163210 (std: split stack overflow module) - #163331 (Move `Arc` and `Rc` into `rcs` mod) - #163183 (Add .seek_read_buf_exact() to std::os::windows::fs::FileExt) - #163450 (Force the correct type variable to never for method resolution on an adjusted never type)

Rollup merge of #162655 - dianne:drop-paramenv-candidates, r=lcnr next solver: prefer to select impl candidates over global where-clause candidates Fixes #162331 Ideally we'd use a more appropriate typing environment when doing const-eval for const-to-pat, which would also fix that (since the problem clauses wouldn't be present to begin with). Being able to do that seems kind of far off, though, so here's a quick fix that (mostly) matches what the old solver does.

…uwer Rollup of 9 pull requests Successful merges: - rust-lang/rust#158997 (Avoid recording unnameable `extern crate` aliases in diagnostic metadata) - rust-lang/rust#162655 (next solver: prefer to select impl candidates over global where-clause candidates) - rust-lang/rust#162832 (add `Div` and `Mul` for `Complex<{float}>`) - rust-lang/rust#162862 (Fix intra doc link resolution when a doc comment is composed of both inner and outer doc comment) - rust-lang/rust#163200 (make `RustaceansAreAwesome` satisfy trait bounds) - rust-lang/rust#163210 (std: split stack overflow module) - rust-lang/rust#163331 (Move `Arc` and `Rc` into `rcs` mod) - rust-lang/rust#163183 (Add .seek_read_buf_exact() to std::os::windows::fs::FileExt) - rust-lang/rust#163450 (Force the correct type variable to never for method resolution on an adjusted never type)

dianne commented Sep 11, 2026

View reviewed changes

rustbot assigned lcnr Sep 11, 2026

dianne marked this pull request as ready for review September 11, 2026 18:54

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 Sep 11, 2026

lcnr reviewed Sep 28, 2026

View reviewed changes

dianne force-pushed the drop-paramenv-candidates branch from 79010d2 to e10ae86 Compare September 28, 2026 13:26

select: prefer impl candidates over where clauses

b6d0fb1

dianne force-pushed the drop-paramenv-candidates branch from e10ae86 to b6d0fb1 Compare September 28, 2026 15:36

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 Sep 28, 2026

JonathanBrouwer mentioned this pull request Sep 28, 2026

Rollup of 12 pull requests #163456

Closed

JonathanBrouwer mentioned this pull request Sep 28, 2026

Rollup of 15 pull requests #163460

Closed

JonathanBrouwer mentioned this pull request Sep 29, 2026

Rollup of 9 pull requests #163469

Closed

JonathanBrouwer mentioned this pull request Sep 29, 2026

Rollup of 9 pull requests #163472

Merged

rust-bors Bot merged commit 31ee28f into rust-lang:main Sep 29, 2026
13 checks passed

rustbot added this to the 1.101.0 milestone Sep 29, 2026

dianne deleted the drop-paramenv-candidates branch September 29, 2026 12:32

maxdexh mentioned this pull request Sep 30, 2026

Nightly associated types regression E0284 #163550

Open

github-actions Bot mentioned this pull request Oct 3, 2026

Toolchain upgrade to nightly-2026-10-05 flux-rs/flux#1811

Draft

Navigation Menu

Uh oh!

next solver: prefer to select impl candidates over global where-clause candidates - #162655