Repository navigation
Clear fmt/clippy lints and add CI with both checks enforcing - #2
gregoryStarr wants to merge 3 commits into
Conversation
Mechanical rustfmt pass over src/. No logic changes — kept as a separate commit so the formatting churn doesn't obscure the clippy fixes that follow. Verified with `cargo check --package rekordbox-analyzer --all-targets` and `cargo test --package rekordbox-analyzer` (4 tests pass).
- analyze.rs: use `n.is_multiple_of(emit_every)` instead of `n % x == 0` - keys.rs: use `a.abs_diff(b)` instead of the manual if/else - features.rs: drive the frame counter with `.enumerate()` rather than a manually incremented `frames` local. The counter was only read by the `frames > 4` guard and bumped at the end of the body, so enumerate's index is identical at every use. - main.rs: targeted `#[allow(clippy::too_many_arguments)]` on cmd_top and cmd_compat. Their parameters mirror the corresponding clap `Command` variants 1:1, so a wrapper struct would only duplicate those fields. Scoped to the two functions — no crate-level allow. `cargo clippy --package rekordbox-analyzer --all-targets -- -D warnings` and `cargo fmt --package rekordbox-analyzer -- --check` both exit 0; 4 tests pass.
The repo had no .github/ directory, so there were no advisory steps to promote — this adds the workflow directly in the enforcing end state: - Format: `cargo fmt --package rekordbox-analyzer -- --check` - Clippy: `cargo clippy --package rekordbox-analyzer --all-targets -- -D warnings` - Test: `cargo test --package rekordbox-analyzer --all-targets` Neither check is advisory: no continue-on-error anywhere, and clippy runs with -D warnings. The tree passes all three as of the preceding two commits. Scoped to the rekordbox-analyzer package throughout; the selecta-tauri workspace member is skipped because it needs GTK/WebKit system libraries that aren't present on the Linux runner.
|
| Run | Branch | Workflow file | Rust job |
|---|---|---|---|
| 1 | assistant/brave-bardeen-ebiwjn (#1) |
#1's ci.yml |
run-level failure in 3s |
| 2 (attempt 2) | assistant/brave-bardeen-ebiwjn (#1) |
#1's ci.yml |
2s, no steps |
| 3 | assistant/brave-bardeen-ebiwjn (#1) |
#1's ci.yml |
run-level failure in 3s |
| 4 attempt 1 | this branch | this PR's ci.yml |
3s, no steps |
| 4 attempt 2 | this branch | this PR's ci.yml |
2s, no steps |
All 4 runs this workflow has ever had, on two different branches with two entirely different ci.yml files, failed the same way. There is no green baseline anywhere in the repo's Actions history. Since the job dies before checkout, nothing in the Rust source or in the step commands can be the cause.
Runners are not simply unavailable — in run 2 the Frontend job ran for 38s while Rust and Site both died in 2s.
The three commands the workflow runs all pass locally
On this branch's head, on ubuntu, rustc 1.94.1:
What I can't do from here
I have no fix to port — there's no passing run to compare against and the logs aren't retrievable, so I can't root-cause past "the job never started." This looks like a repository- or account-level Actions problem (runner provisioning, an allowed-actions policy, or Actions billing/minutes). That needs someone with access to the repo's Actions settings and the account's billing page; it isn't fixable from a commit.
My one permitted re-run is spent. I'll keep watching this PR and will re-check rather than re-run again.
Generated by Claude Code
Clears every rustfmt and clippy warning in the crate, then adds a CI workflow where both checks are blocking.
Commits
1.
style: apply cargo fmt across the crate— mechanical rustfmt pass oversrc/, no logic changes. Kept separate so the formatting churn doesn't obscure the fixes that follow.2.
fix(lints): clear all clippy warningsmanual_is_multiple_ofsrc/analyze.rs:174n.is_multiple_of(emit_every)manual_abs_diffsrc/keys.rs:48a.abs_diff(b)explicit_counter_loopsrc/features.rs:63.enumerate()too_many_arguments×2src/main.rs(cmd_top,cmd_compat)#[allow]The first two came from
cargo clippy --fix; both rewrites are semantically identical.For
explicit_counter_loop, theframeslocal was read only by theframes > 4guard and incremented at the end of the loop body, soenumerate()'s index is identical at every use.For
too_many_arguments, I went with a scoped#[allow]plus a one-line comment rather than a parameter struct. Both functions mirror theirCommand::Top/Command::Compatclap variants 1:1 — the enum variant already is the parameter struct, so a wrapper would duplicate those fields and add a field-by-field copy in the dispatch arm without removing anything. The allows sit on the two functions only; there is no crate-level allow.3.
ci: add workflow with enforcing fmt and clippy checks— see the note below.Note on scope
The task this branch came from described editing
.github/workflows/ci.ymlto removecontinue-on-error: truefrom existing "Format (advisory)" / "Clippy (advisory)" steps. This repo has no.github/directory — there was no workflow and no advisory steps to promote. The workflow is therefore added directly in the enforcing end state, which matches the stated goal:Format—cargo fmt --package rekordbox-analyzer -- --checkClippy—cargo clippy --package rekordbox-analyzer --all-targets -- -D warningsTest—cargo test --package rekordbox-analyzer --all-targetsNo
continue-on-erroranywhere; clippy runs with-D warnings.The task also referred to the package as
track2mix(libtrack2mix_core, plus atrack2mix-tauricrate). The actual names arerekordbox-analyzer(librekordbox_analyzer, binrekordbox-analyzer) andselecta-tauri; the rename hasn't landed onmain. I used the real names. If the rename is meant to happen, it's a separate change.Every step is scoped to the
rekordbox-analyzerpackage —selecta-tauriundersrc-tauri/is skipped because it links against GTK/WebKit system libraries not present on the Linux runner.Verification
All run locally on the final tree:
cargo fmt --package rekordbox-analyzer -- --checkcargo clippy --package rekordbox-analyzer --all-targets -- -D warningscargo test --package rekordbox-analyzer --all-targetsOne correction to the original task notes: the crate does not have zero tests.
src/keys.rsalready has a#[cfg(test)] mod testswith 4 passing Camelot tests.src/features.rsdoes still have none, so that remains the obvious place to add coverage — not done here, as it wasn't part of this task.Generated by Claude Code