$ the-wire · showcase
PartialOrd chaining methods removed, rust-analyzer lockfile fix
By RepoJournal · Filed · About Rust · Composed from the cited sources · methodology
A quiet night across the Rust repos: internal PartialOrd trait plumbing got simpler, rust-analyzer fixed a Nix-specific metadata bug, and the rest is subtree pulls and crate index churn.
Remove redundant `PartialOrd` chaining methods rust-lang/rust
Max Dexheimer removed __chaining_le/__chaining_lt and __chaining_ge/__chaining_gt from the PartialOrd chaining internals, where the code showed the pairs were implemented identically in every impl. Anyone who has been hand-optimizing against those methods will need to re-check their assumptions; the rollup carrying it also landed higher-ranked predicate handling in imperfect derives diagnostics.
Search manifest ancestors for the lockfile when fetching cargo metadata rust-lang/rust-analyzer
Under a Nix toolchain, macro-generated TyCtxt query getters such as tcx.mir_keys(()) resolved to {unknown}; onlycs traces that to three independent bugs, one fixed in 7015591, one in #23297, and the last here. The remaining defect was the rustc-src lockfile never being found, since the component ships compiler sources at [rustc-src]/rust/compiler/rustc/Cargo.toml with no workspace root manifest.
fix: compute ref-match on undotted self rust-lang/rust-analyzer
A4-Tacks fixed reference-match computation on undotted self and method call name-matching, so hovering an argument now surfaces fd &mut self.info instead of the previous sparse set of candidates. The practical effect is more accurate type-and-name resolution inside impl blocks during inline hints and completion.
Rollup merge of #164688 - RalfJung:qnx-test-remove-dir, r=Darksonn rust-lang/rust
test_dir_remove_dir is re-enabled on QNX after the earlier disabling was reverted; the justification for skipping it cited remove_dir_all under concurrent mutation, which is not what the test exercises. The underlying problem was a buggy fallback implementation of remove_dir, so QNX filesystem behavior now gets test coverage again.
Merge pull request #23531 from A4-Tacks/fn-mis-name-recover rust-lang/rust-analyzer
The parser recovery for a function missing its name improved: an opening parenthesis after fn now produces a PARAM_LIST rather than a pair of ERROR nodes, collapsing what used to be three diagnostics into one. Also merged: a rustls-webpki bump to 0.103.16 with webpki-roots 1.0.10, Miri's NetBSD test filters, and a nightly-2026-10-22 subtree pull into rustfmt.
