A group holds one active and one pending run, so `ci-${{ github.ref }}` let a
later push to main replace a queued run and skip its `cargo xtask ci`. Pull
request pushes keep the shared group (that supersession is the point); every
other run — a push to main, a manual one — now gets `ci-<run_id>`, a group of
its own, and always runs.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
- ci.yml: cancel-in-progress only for pull requests, so a push to main never cancels the run of
the push before it.
- ci --quick: `git diff --name-status -z -M` and `git ls-files -z`, parsed as NUL-separated
fields (a rename or copy takes two paths), so a non-ASCII path, which git otherwise quotes and
escapes, still maps to its package.
- ci --quick: files outside a package that it compiles in (include_str!/include_bytes!, a build
script's rerun-if-changed: fonts, camera profiles and the app icon under assets/,
docs/whats-new.md, docs/raw/coverage-sweep.tsv, docs/raf-corpus.sha256, contributors/) count
as a change to that package. A test scans the workspace and fails when one is not mapped.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- .github/workflows/ci.yml: the full `cargo xtask ci` on every pull request and push to main
(Linux x64, craft-fonts, rust-cache for dependencies), so it becomes the merge gate.
- `cargo xtask ci --quick [--base REF]`: fmt, clippy, the changed packages' tests, layers and
assets, plus heif / parity / wasm when the changes touch them, measured against the merge base
with origin/main (xtask/src/ci_quick.rs). Workspace-wide files test everything.
- A machine-wide lock (xtask/src/ci_lock.rs): a second `ci`, from another worktree, waits and
names the run it waits for. LIGHTCRAFT_CI_SLOTS=N, LIGHTCRAFT_CI_LOCK=off|<dir>.
- README, CLAUDE.md quality gate, gaps.md gap 9, target-app-parity.md stability note.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
