Repository navigation
v1.94.20.0 test(review): one file per review and harness pin, so independent repins merge cleanly (GST-1084) - #3903
Merged
Merged
Conversation
…ins merge cleanly (GST-1084)
The reviewer-brief pins move from the table in test/review-packets-removed.test.ts to
test/fixtures/review-pins/{files,replays,brief-bodies}/*.json, and the harness pins from
scripts/harness-version.json to scripts/harness-pins/<path>.json. Every entry is ported
unchanged. New tests prove a stale pin still fails and that two repins of different pins
merge cleanly with a real git merge-tree.
Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
|
😎 This pull request was merged. |
- unadmitted() and the specialist coverage use Object.hasOwn, so an inherited name such as toString is not an admitted replay - MUST_PIN names every non-specialist reviewer source; deleting one of those pins now fails - writeManifest writes the pins before the manifest head; pin parse errors name the file - mergePinEdits removes its repository when setup throws, and the merge tests clean up every merge they made Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
…-file-per-review Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> # Conflicts: # test/review-packets-removed.test.ts
…-file-per-review Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> # Conflicts: # scripts/harness-version.json
…(GST-1084) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
7 tasks done
…rift (GST-1084) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
…-file-per-review Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> # Conflicts: # test/review-packets-removed.test.ts
garrytan
added a commit
that referenced
this pull request
Nov 3, 2026
Resolve scripts/harness-version.json against #3903: take main's manifest and carry this branch's test-pr-profile.ts pin into scripts/harness-pins/scripts/test-pr-profile.ts.json. Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
garrytan
added a commit
that referenced
this pull request
Nov 3, 2026
test/review-packets-removed.test.ts takes main's version (#3903 moved the review pins to test/fixtures/review-pins/). This branch's two pin entries become pin files with GSTA-202's recipe: a new replays/GST-508+GST-464+GST-967+GSTA-258-L4.json and the files/ship/sections/review-army.md.json repin to 22c37a64. Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
6 tasks done
This was referenced Nov 3, 2026
Merged
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What changed
The review pins and the harness pins are now one file per pin, so two PRs that repin different files never touch the same lines and merge without a conflict.
NO_LOSS_REPLAYS,REVIEWER_BRIEFSandBRIEF_BODIEStables intest/review-packets-removed.test.tsmoved totest/fixtures/review-pins/. Each pinned file hasfiles/<pinned path>.jsonwith its sha and replay. Each replay hasreplays/<slug>.jsonwith its id and written no-loss justification. Each brief body hasbrief-bodies/<name>.jsonwith its sha.scripts/harness-version.jsonnow holds onlyversionandpatterns. Each harness pin moved toscripts/harness-pins/<path>.json, andscripts/bump-harness-version.tsreads and writes those files.lib/ship-release.ts:/ship's post-review route (GST-1031) reads the per-file harness pins, so a non-behavioral re-pin after review still counts as release metadata. A manifest that regains afilestable counts as behavior.test/helpers/pin-merge-fixture.tsruns a realgit merge-tree --write-treeon two branches in a fresh repository.Why
The single pin table caused at least five main-merge conflicts on 2026-11-02, two of them on #3852 (GST-1084). Git treats changes on adjacent lines as a conflict, so sorted one-line entries would still conflict for neighboring pins. Separate files never share lines.
The repro ran two branches that repin the adjacent review-army.md and adversarial.md pins and the adjacent floor.ts and observation.ts pins. On main 14d7414 both files conflict (merge-tree exit 1). On this branch the same edits merge cleanly (exit 0).
Blast radius
This changes test and merge-check infrastructure only, plus the post-review route in
lib/ship-release.ts. No reviewer text, sha, replay or harness blob changed. All 21 replays, 20 file pins, 6 brief bodies and 42 harness entries are ported unchanged, including the combined GST-508 entries from #3852.HARNESS_VERSIONand its readers (eval-trial-series.ts,ship-measure.ts) are unchanged.The checks are stricter than on main. Every pin still needs its sha and a replay with a written no-loss justification, and the brief-body check still guards the bodies. These now fail too. They are a missing pin for any reviewer source (
MUST_PINplus every specialist checklist), a replay name that only exists on the object prototype (toString), a malformed or misnamed pin file, a stray file in a pin directory, and afilesblock back in the manifest.Approvals
No approval class applies (
APPROVAL_CLASSES: none). GStack Minder granted two main merges (5f705f3 and f6d6215) and a fresh /ship at f0613db with one fix round.Pre-Landing Review
REVIEW_LOOP: passes=1 blocking_found=2 deferred=1 noted=2 fixed=1 scope=frozen ledgers_read=4 fix_checks=2 dispatches=4
Review-loop exception: accepted by GStack Minder (GSTA-36) 2026-11-03 via GSTA-36 review_loop_exception grant for #3903: deferred outside-voice CRITICAL in gate code (partial --bump leaves pins ahead of the manifest) fixed with a failing-first test and one scoped fix-check, no further rounds (fix-check at ecd3caa, 0 findings)
Gate edits: the GST-1031 test fixture in
test/ship-release.test.tsmoved to the per-file pin layout with the same four cases, plus one new negative case. It fails on the oldharnessRepinand passes on the fix.harnessRepinonly read the oldfilestable, so every post-review re-pin stopped finish. Fixed in 876b064.Review notes
scripts/bump-harness-version.ts:28: pin files underscripts/harness-pins/test/helpers/classify as tests inlib/review-depth.ts, while the old manifest classified as code, so a PR that only re-pins a harness helper plans at a lower review depth. Changing the classifier touches a pinned file and needs its own no-loss replay.lib/ship-release.ts:186: deleting a pin after review routes as metadata, as the old table rule did; CI's harness-version test still catches it.lib/ship-release.ts:181: two different unparsable pin states compare equal;readManifestthrows on them in CI.readPinsthrows on untracked files such as.DS_Storeor.originscripts/harness-pins/; pin contents are only checked to parse; a stale comment intest/harness-version.test.ts.lib/ship-*changed) failed in 13 files outside this diff (browse daemon, design-lint browser tier, CSO ordinary repos, autoplan gate judgment); Free Tests run 41610864341 passed at the same tree.Documentation
Status: updated. One factual fix and no blockers.
Scope: base f6d6215, head 527c7f6, branch capy/one-file-per-review (GST-1084). 101 changed files: review pins and harness pins split into one file per pin, plus
lib/ship-release.ts,scripts/bump-harness-version.tsand their tests. Nothing was staged, unstaged or untracked. The three docs the diff touches were read in full. A repo-widegit grepof docs for the changed paths found no other stale references.Documentation health
Coverage debt: no how-to or tutorial for repinning a review file (edit its
files/pin and add the replay file it names); the changelog fragment is the only place it appears.bump-harness-version.tshas no standalone reference page. Diagram drift: none.Test plan
bump-harness-version --checkfails on drift,--non-behavioralrewrites only the touched pin, and a removed file drops its pinDeferred findings
None open. The outside voice's CRITICAL (
scripts/bump-harness-version.ts:109: a--bumpthat fails partway left pins ahead of the manifest version while the drift check passed) was deferred after the fix round, then fixed in ecd3caa under GStack Minder's review-loop exception:harnessDriftcounts a pin whosesinceis above the manifest version as changed. A new test makes the realwriteManifestfail on its second pin; it fails at 37b736e and passes on the fix. The scoped fix-check found nothing. The receipt'sdeferred=1counts that earlier deferral.