* fix(#2983): classifier exit-code discipline, base-tag staging, drop vestigial merge-back Three issues surfaced by CodeRabbit's post-merge review of #2981 plus a production failure on the v1.39.1 release run. (1) Overloaded classifier exit code scripts/diff-touches-shipped-paths.cjs reused exit 1 for both the legitimate "no shipped paths" result and Node's default exit on uncaught throw, so any classifier failure (corrupt package.json, EPERM, etc.) was indistinguishable from a normal skip — the workflow's `if ! ... ; then skip` idiom would silently drop the commit. Distinct exit codes now: 0 shipped — at least one path is in the npm `files` whitelist 1 not shipped — CI / test / docs / planning only 2 classifier error — workflow MUST fail-fast uncaughtException + unhandledRejection + try/catch around fs/JSON parsing all route to exit 2 with stderr context. (2) Classifier missing at the base tag (CRITICAL) `Prepare hotfix branch` runs `git checkout -b "$BRANCH" "$BASE_TAG"` BEFORE the cherry-pick loop, replacing the working tree with the base tag's contents. Base tags predating #2980 (notably v1.39.0, the most likely next hotfix base) don't have scripts/diff-touches-shipped-paths.cjs at all — `node <missing>` exits non-zero — `if !` skips every commit — empty hotfix branch published. Strictly worse than the original #2980 push-rejection, which at least failed loudly. Stage the classifier from the dispatched ref's working tree into $RUNNER_TEMP at the top of the run script (before any working-tree- mutating git command). The cherry-pick loop now references $CLASSIFIER (staged) instead of the in-tree path. Sanity guards: refuse to start if scripts/diff-touches-shipped-paths.cjs is missing in the dispatched ref, refuse to proceed if cp didn't materialize $CLASSIFIER. The cherry-pick loop captures node's exit via ${PIPESTATUS[1]} and dispatches via explicit case: 0 proceed with cherry-pick 1 skip into NON_SHIPPED_SKIPPED * emit ::error:: + exit "$CLASSIFIER_RC" (3) Drop the merge-back PR step Auto-cherry-pick only picks commits already on main (`git cherry HEAD origin/main` outputs the unmerged ones; we filter fix:/chore: from main). By construction every code commit on the hotfix branch is already on main. The only hotfix-branch-only commit is `chore: bump version to X.Y.Z for hotfix`, which either no-ops against main or rewinds main's in-progress version. The merge-back PR was vestigial. It also failed in production on run 25232968975 with `GitHub Actions is not permitted to create or approve pull requests (createPullRequest)` — org policy blocks PR creation from the workflow's GH_TOKEN. Even without that block, the PR would have nothing useful to merge. Step removed. The `pull-requests: write` permission granted solely for the merge-back step has been dropped from the release job (least-privilege). Regression coverage tests/bug-2983-classifier-exit-codes-and-base-tag-staging.test.cjs adds 12 assertions across two describe blocks: - 5 classifier behavioral: exit 0/1 preserved, exit 2 on missing package.json, exit 2 on malformed JSON, exit-code constants exported. - 7 workflow contract: classifier staged before checkout, target is $RUNNER_TEMP, missing-source guard, missing-staged guard, PIPESTATUS-based dispatch, error branch fails workflow, loop uses staged path (not in-tree). tests/bug-2980-hotfix-only-picks-shipping-changes.test.cjs updated where it asserted the pre-#2983 `if ! ... ; then` shape: now accepts the post-#2983 case-dispatch form. The test still proves the classifier participates; bug-2983 enforces the specific shape. Run summary references for the curious reviewer: - Run 25232010071 — original #2980 trigger (workflow-file push rejection) - Run 25232968975 — failed merge-back step that prompted the "is this even useful?" question that drove the removal Closes #2983 * fix(#2983): address CodeRabbit findings on PR #2984 Two findings, both real, both fixed. (1) [Critical] PIPESTATUS capture clobbered by `|| true` Pre-fix shape: git diff-tree ... | node "$CLASSIFIER" || true CLASSIFIER_RC="${PIPESTATUS[1]}" When the classifier exits 1 ("not shipped" — common case) or 2 (error), `|| true` triggers the right-hand side. `true` is a one-command "pipeline" that overwrites PIPESTATUS to (0). ${PIPESTATUS[1]} on the next line is therefore unset (or stale under set -u). The case dispatch then matched the empty string — falling into `*)` and failing the workflow on every non-shipped commit, OR matching `0)` after some shells default-init unset to 0 and silently picking commits that don't ship. Local repro confirms the issue: $ bash -c 'set -euo pipefail; false | sh -c "exit 7" || true; \ echo "PIPESTATUS: ${PIPESTATUS[*]}"; \ echo "[1]: ${PIPESTATUS[1]:-<unset>}"' PIPESTATUS: 0 [1]: <unset> Fix: bracket the pipeline in `set +e`/`set -e`, snapshot PIPESTATUS into a local array on the very next line, then dispatch on the snapshot: set +e git diff-tree ... | node "$CLASSIFIER" PIPE_RC=("${PIPESTATUS[@]}") set -e DIFFTREE_RC="${PIPE_RC[0]}" CLASSIFIER_RC="${PIPE_RC[1]}" The snapshot must happen on the first line after the pipeline; any intervening simple command resets PIPESTATUS. The array form is invariant against that. Bonus from the new shape: $DIFFTREE_RC is now also captured. git diff-tree is unlikely to fail on a known-good $SHA, but if it does, we no longer feed partial/empty input to the classifier and call it "not shipped." A non-zero DIFFTREE_RC emits ::error::git diff-tree failed and exits. (2) [Minor] Stale "Merge-back PR opened against main" summary line The hotfix run summary still printed: echo "- Merge-back PR opened against main" But the merge-back step itself was removed in the previous commit on this branch. Operators reading the summary would expect a PR that doesn't exist. Replaced with explicit non-action text: echo "- No merge-back PR (auto-picked commits are already on main)" Test coverage bug-2983 test file gains 3 assertions: - PIPE_RC array-snapshot pattern is required (regex matches the exact `PIPE_RC=("${PIPESTATUS[@]}")` form). - The `pipeline || true; ${PIPESTATUS[1]}` antipattern is explicitly forbidden via assert.doesNotMatch. - DIFFTREE_RC is captured from PIPE_RC[0] and a non-zero value triggers ::error::git diff-tree failed. - Run summary forbids `Merge-back PR opened against main` and requires the new non-action sentence. bug-2964 test's loop-anchor window bumped 6 KB → 8 KB to accommodate the additional pre-pick scaffolding (the test's own comment had already anticipated this kind of growth, citing prior precedents from #2970 and #2980). Mark CodeRabbit comments resolved post-commit. Refs CR finding ids 3175253571, 3175253578 on PR #2984.
This commit is contained in:
Tom Boucher
@@ -8,6 +8,7 @@ Format follows [Keep a Changelog](https://keepachangelog.com/en/1.1.0/).
|
||||
|
||||
### Fixed
|
||||
|
||||
- **`release-sdk` hotfix flow hardened against silent classifier failures, missing-classifier-at-base-tag, and a vestigial merge-back PR step** — three issues surfaced by CodeRabbit's post-merge review of #2981 plus a production failure on the v1.39.1 release run. **(1)** `scripts/diff-touches-shipped-paths.cjs` reused exit code `1` for both the legitimate "no shipped paths" classifier result and Node's default uncaught-throw exit, so any tooling failure was indistinguishable from a normal skip. The script now uses `0` (shipped), `1` (not shipped), `2` (classifier error) with `try`/`catch` + `uncaughtException`/`unhandledRejection` handlers routing all failure paths to exit `2`. **(2)** The workflow's `git checkout -b "$BRANCH" "$BASE_TAG"` overwrote the working tree with the base tag's contents *before* the cherry-pick loop ran the classifier — but base tags predating the classifier's introduction (notably v1.39.0) don't have the file in their tree, so `node scripts/diff-touches-shipped-paths.cjs` would exit non-zero and silently drop every commit, producing an empty hotfix release. The classifier is now staged into `$RUNNER_TEMP` at the top of `Prepare hotfix branch` (before any working-tree-mutating git command), and the loop references that staged copy. The cherry-pick loop snapshots `$PIPESTATUS` into a local array (`PIPE_RC=("${PIPESTATUS[@]}")`) immediately after the classifier pipeline — under bracketed `set +e`/`set -e` — and dispatches via explicit `case`: `0` proceeds, `1` skips into `NON_SHIPPED_SKIPPED`, anything else emits `::error::shipped-paths classifier failed for $SHA (exit N)` and fails the workflow. CodeRabbit on PR #2984 caught a subtler bug in the first iteration: `pipeline \|\| true; RC=${PIPESTATUS[1]}` is broken because `\|\| true` runs `true` as its own one-command pipeline on the failure paths, overwriting `PIPESTATUS` to `(0)` and leaving `${PIPESTATUS[1]}` unset. The array-snapshot form is invariant against this. The same hardening also surfaces `git diff-tree`'s exit code (via `PIPE_RC[0]`); a non-zero diff-tree result now also fails the workflow rather than feeding partial input to the classifier. **(3)** Removed the `Open merge-back PR (hotfix only)` step. The auto-cherry-pick hotfix flow only picks commits already on main (`git cherry HEAD origin/main` outputs the unmerged ones), so by construction every code commit on the hotfix branch is already on main. The only hotfix-branch-only commit is the version-bump chore, which would either no-op against main or rewind main's in-progress version. The step also failed in production with `GitHub Actions is not permitted to create or approve pull requests (createPullRequest)` (org policy) on run [25232968975](https://github.com/gsd-build/get-shit-done/actions/runs/25232968975). The `pull-requests: write` permission previously granted to the release job has been dropped in line with least-privilege. The run-summary line that previously echoed `Merge-back PR opened against main` has been replaced with `No merge-back PR (auto-picked commits are already on main)` so operators reading the summary see an accurate non-action statement (CodeRabbit on PR #2984). Regression covered by `tests/bug-2983-classifier-exit-codes-and-base-tag-staging.test.cjs` (15 assertions across exit-code semantics, classifier staging, error dispatch, PIPESTATUS-snapshot hardening, diff-tree fail-fast, merge-back removal, and run-summary accuracy). (#2983)
|
||||
- **`release-sdk` hotfix only cherry-picks commits that change what actually ships** — the `fix:`/`chore:` filter in `Prepare hotfix branch` was too broad: it picked any commit with that conventional-commit type regardless of whether the diff could affect the published npm package. CI-only fixes (release-sdk.yml itself, hotfix tooling, test-only commits) were getting cherry-picked into hotfix branches even though they cannot change the tarball — and the subset touching `.github/workflows/*` then caused the prepare job's `git push` to be rejected by GitHub because the default `GITHUB_TOKEN` lacks the `workflow` scope, aborting the run. v1.39.1 hit this on PR #2977 (run [25232010071](https://github.com/gsd-build/get-shit-done/actions/runs/25232010071)). The loop now pre-skips any candidate commit whose `git diff-tree` output doesn't intersect the npm tarball's shipped paths (entries in `package.json` `files`, plus `package.json` itself, which `npm pack` always includes). Skipped commits land in a new `NON_SHIPPED_SKIPPED` summary bucket framed as informational — non-shipping commits cannot affect the package, so the skip needs no operator action. The shipped-paths classifier lives in `scripts/diff-touches-shipped-paths.cjs` so its rules (file-OR-directory prefix matching `npm pack` semantics, the always-shipped rule for `package.json`, the lockfile-not-shipped rule) are unit-testable. Regression covered by `tests/bug-2980-hotfix-only-picks-shipping-changes.test.cjs`. (#2980)
|
||||
- **`release-sdk` hotfix workflow fails on real run with `npm error Version not changed`** — the `release` job's `Bump in-tree version (not committed)` step ran `npm version "$VERSION"` without `--allow-same-version`, so it errored on real (non-dry-run) hotfix runs because `prepare` had already committed the bump on the hotfix branch. The release job's checkout `ref` is asymmetric — `BRANCH` (already bumped) on real runs vs `BASE_TAG` (older version) on dry-runs — which is why dry-run never caught the bug. Both `npm version` calls in that step now pass `--allow-same-version`, matching the existing pattern in `release.yml:326`. (#2976)
|
||||
- **`gsd-sdk query agent-skills` emits raw `<agent_skills>` block instead of JSON-wrapped string** — workflows that embed via `$(gsd-sdk query agent-skills <agent>)` were receiving a JSON-quoted string literal mid-prompt (e.g. `"<agent_skills>\n…"`), silently breaking all `<agent_skills>` injection into spawned subagents. The CLI dispatcher now honors an opt-in `format: 'text'` field on `QueryResult` and writes such results raw via `process.stdout.write`; `--pick` always returns JSON regardless. (#2917)
|
||||
|
||||
Reference in New Issue
Block a user