Exact

Exact Union RegExp

Repository files (latest commit first)

Filename	Latest commit message	Latest commit date
Transaction Science 2465cf30c3 clippy: the simplifications a machine can make without changing the code's shape ... Of 5250 warnings across the seventeen workspaces, these are the ones whose fix is mechanical and local: a redundant `.into()`, `clone` on a `Copy`, a derivable `Default`, `useless_vec`, `default_constructed_unit_structs`, `manual_is_multiple_of`, `uninlined_format_args`. 34 files, 24 lines shorter. Deliberately NOT in here, because a tidy-up should not decide them: The ~2000 cast lints (cast_possible_truncation 914, cast_precision_loss 664, cast_sign_loss 236, cast_possible_wrap 186) and ~837 doc-style ones (doc_markdown 554, overindented lists 192, lazy continuation 91). In numeric code the casts are the work; "fixing" them means 2000 allows or try_into().unwrap() churn. With too_many_arguments, type_complexity, unused_self and many_single_char_names that is ~3170 of the 5250 -- a question about whether these crates carry clippy::pedantic, answered once at crate level, not 3170 times at call sites. unused_imports (88) and unused_unsafe (81): cfg-dependent here. jouleclaw-loader-gguf's `std::ffi::c_void` is behind the `numa` feature and its `std::arch::aarch64::*` sits inside an aarch64-gated fn -- dead in this configuration, live in another. collapsible_if (168): its fix is correct but keeps the old body indentation, and this tree is not rustfmt-clean so it cannot be normalised. An earlier pass fixed 19 files that way, inspected them and reverted; the lint is now excluded up front rather than fixed and undone. chunks_exact_to_as_chunks (65), should_implement_trait (34), result_large_err (52), needless_range_loop (191): API changes and loop rewrites, each wanting its own pass with the tests watched. Regenerated on current main rather than rebased: six aac-web files were touched by both this pass and the 71 commits that landed meanwhile, and re-running a machine pass on the new base is correct by construction where resolving its conflicts by hand is not. aac needed nothing the second time -- those simplifications had already landed. Verified on the build host, every check asserting it did the work rather than trusting an exit status: all ten touched workspaces' tests pass, `ran == 10` asserted, and no let-chain appears in the diff. ⚠️ eoc-cascade's `hit_calibration_process_energy` fails intermittently and is NOT from this change: on clean origin/main with no local modification it went pass, fail, fail in three consecutive runs. The check above admits it only when it is the sole failing target, so a second failure in eoc would still fail. Co-Authored-By: Claude Opus 5 <[email protected]>		2026-10-18 15:30:10 -04:00
..
crates	clippy: the simplifications a machine can make without changing the code's shape	2026-10-18 15:30:10 -04:00
Cargo.lock	joule-moq-relay: a MOQT draft-22 relay over native QUIC, WebTransport and a WebSocket fallback	2026-10-17 04:33:00 -04:00
Cargo.toml	joule-moq-relay: a MOQT draft-22 relay over native QUIC, WebTransport and a WebSocket fallback	2026-10-17 04:33:00 -04:00