You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Commit 4065d4d
Browse filesBrowse the repository at this point in the historyBrowse files
cleanup: no clippy warnings, no compiler warnings, and cargo fmt over the workspace
`cargo clippy --workspace --all-targets` is now silent (was 159 warnings) and `cargo fmt --all -- --check` is empty.
Also removes the six planning documents under `docs/` that the release no longer ships — `architecture.md`,
`cpp-format.md`, `format-strings.md`, `lsp-status.md`, `option-matrix.md`, `remaining.md`; `docs/configuration.md` stays.
**What clippy could fix itself** (34 sites, applied by `clippy --fix`): `needless_borrow`, `useless_conversion`,
`redundant_closure`, `manual_contains`, `collapsible_if`, `single_element_loop`, `while_let_loop`, `question_mark`,
`match_like_matches_macro`, `sort_by_key`, `needless_update`, `print_literal`, `useless_format`, `iter_nth_zero`,
`double_parens`, `chunks_exact_to_as_chunks`, `bind_instead_of_map`, `doc_lazy_continuation`, `obfuscated_if_else`,
`needless_range_loop` in the formatter, `items_after_test_module` (the moved function is `init_depths`).
**What it could not, and what each answer was:**
* `too_many_arguments` / `type_complexity` (26 sites): answered with a **crate-level allow and a rationale** in
`cpp_code_analysis` and `cpp_formatter` — a question in this layer is asked with `(index, scopes, root, path)` and
then with what it is about, which is not a data clump to bundle, and the memo tables are spelled out where the
reader checks them against the code that inserts into them.
* `large_enum_variant` (1): allowed on the enum, with the reason — boxing the found case of `Known` would allocate on
the hot path of every member lookup to save bytes in a value that lives for one question.
* `len_without_is_empty` (1): `is_empty` added to `Surfaces`.
* `wrong_self_convention` (1): `as_a_template` takes its receiver by value, so it is now `into_a_template`.
* `zombie_processes` (1): the progress test kills its server and now **waits** on it.
* `assertions_on_constants` (2): the ranking invariants in `matching.rs` are `const _: () = assert!(…)`, so a
reordering fails the build instead of a test.
* `ptr_arg` (1): `&PathBuf` → `&Path`.
* `duplicated_attribute` (1): the inner `#![allow(dead_code)]` in `walk_cache.rs` duplicated the one on its `mod`
declaration.
* `empty_line_after_doc_comments` (14): each was a **floating doc block** with no item — the text is kept, as `//`
comments, because pretending it documented the next item would have been a new lie. The one exception is the doc
block that belonged to `fn read_a_template_id_here`, which is now where it belongs.
* `items_after_test_module` (8): handled by the fix where it could move the item; `layout.rs`''s `init_depths` is the
one that moved.
* dead test helpers (2): `#[allow(dead_code)]` on the shared `assert_all_parse` (only some test binaries call it),
and the unused `inside_the_call` in `c_project.rs` deleted.
**One real correctness fix came out of it**: `features.rs` asserted `!if … { true } else { false } && !…`, which is a
negation of a literal and read as "typeof is supported" — the test now says what it meant,
`assert!(!strict_cpp.support(Feature::Typeof))`.
Suites: `cpp_parser` 33, `cpp_code_analysis` 50, `cpp_ls --release` 18, `cpp_formatter` — all green; `cargo build
--release --workspace` and `cargo clippy --workspace --all-targets` both warn-free.
0 commit comments