Repository navigation
v1.94.17.0 ci-carry: guard carry, jobs-mapped lanes, aliases, per-file carry of runner-skipped files, path-skipped jobs (GST-1186, GST-1189, GST-1190) - #3929
Merged
Conversation
…CI or the ledger (GST-1186) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
…ulate proof, policy-base check on carried (GST-1189, GST-1190) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
…tree.json records skipped_files (GST-1190) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
… find --policy-from head --simulate (GST-1190) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
…rom #3923 per GSTA-258) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
…(gate inputs, passed ignores CI rows, simulate consent, pull_request-only skips, guard not-owed) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
…nnot print Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
…h per_file Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
…ns must still stand (l6c-p2) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
…'s count must come from this run's row (l6c-p2 red-team) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
|
😎 This pull request was merged. |
…check ratchet Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
3 tasks done
…iles with per_file" This reverts commit fafbae5.
…r_file move to the adoption PR; Step 17 keeps its refs/heads marker Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
…oubleshooting (docs audit l6cdocs2) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
…mainder run recheck (#3929 second fix-check) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
3 tasks done
Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> # Conflicts: # lib/pregate/index.ts # lib/pregate/run-check.ts
…n-fixture siblings (the path-scoped Windows check runs it at its first spawn) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
2 tasks done
garrytan
added a commit
that referenced
this pull request
Nov 3, 2026
lib/run-evidence.ts keeps #3929's skipped_files, job_skipped, skipped and skipped_jobs evidence and its verifyRun unchanged option next to #3930's sizing_experiment field and refusal and the ignoreRun option. docs/pregate.md run_profiles row keeps #3930's wording and adds skippable?. 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
…GSTA-258 L4) The rule 19 merge of main 1ccab0d put #3929's push-guards wording (61,418 on main) together with J1's head=<sha> in the GSTACK_RESULT line, measuring 61,429. Three rewordings beside that line bring the skeleton back to 61,418; the branch ledger entry drops by 11 and the ship goldens are regenerated. Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
capy-ai Bot
added a commit
that referenced
this pull request
Nov 3, 2026
…ry warning and record-ci recheck order (GSTA-258 L6 follow-up) (#3934) * feat(pregate): a guard that already passed at this tree carries from CI or the ledger (GST-1186) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * feat(ci-carry): jobs-mapped lanes, aliases, narrow-lane counts, --simulate proof, policy-base check on carried (GST-1189, GST-1190) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * fix(ci-carry): a file whose every test skipped on CI is not carried; tree.json records skipped_files (GST-1190) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * feat(ci-carry): path-skipped jobs count only under an unchanged gate; find --policy-from head --simulate (GST-1190) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * ship: name the pre-push guards in Step 17 and the early push (moved from #3923 per GSTA-258) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * ci-carry: per-file lane carry for runner-skipped files; review fixes (gate inputs, passed ignores CI rows, simulate consent, pull_request-only skips, guard not-owed) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * ship: Step 17 names the pre-push guards without a code this branch cannot print Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * pregate: gstack's bun run test lane runs its runner-skipped files with per_file Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * ci-carry: a per_file remainder must run tests here and the carried runs must still stand (l6c-p2) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * ci-carry: read record-ci from its own last record line; the remainder's count must come from this run's row (l6c-p2 red-team) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * test(ci-tree-reuse): type the skipped_files fixture for the test typecheck ratchet Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * Revert "pregate: gstack's bun run test lane runs its runner-skipped files with per_file" This reverts commit fafbae5. * ci-carry: ship the skipped_files reader inert; gstack's writer and per_file move to the adoption PR; Step 17 keeps its refs/heads marker Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * docs: guard carry, per_file and the too-old message in pregate and troubleshooting (docs audit l6cdocs2) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * gstack-evidence record-ci: recheck the working tree after the post-remainder run recheck (#3929 second fix-check) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * gstack-evidence: sum bun --shard processes in a lane's tests_ran; docs: dispatch-carry warning and record-ci recheck order (GBEV-3, GSTA-258 L6 follow-up) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * gstack-evidence: keep shard and count lines that leave the output tail; sum only shards 1..N in order; soften the dispatch-carry mechanism claim (l6f-p1 fixes) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * ci-carry: a pull_request run whose base moved names the one-command dispatch fix (TRUT-7) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * docs: the tree_mismatch refusal names its fix command (l6fdocs3) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * ci-carry: a pull_request run that recorded its merge tree names the base-moved fix (the real tree_mismatch path; #3934 fix-check) 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>
capy-ai Bot
added a commit
that referenced
this pull request
Nov 3, 2026
…eck, lighter READY, head= on the result line (GSTA-258 L4) (#3926) * fix(ship): receipt counts bind by finding id, every run kept, scoped re-check after a refused carry, READY tail, head= on the result line (GSTA-258 L4; GST-1136, GST-1168, GST-1179) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * fix(ship): goldens, head= in the base-red line test, Severity field after a closing quote (ggstack#71); repin ship/sections/review-army.md under GST-508+GST-464+GST-967+GSTA-258-L4 (brief-identity no-loss check: every cut brief unchanged) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * fix: pre-landing review fixes (J1 never re-checks a legacy receipt, a cut path walk or moved base context; legacy READY asks for the cross-family read) Review-Finding: p1/specialist-batched#1 reader=security Review-Finding: p1/specialist-batched#2 reader=testing Review-Finding: p1/red-team#1 reader=red-team Review-Finding: p1/red-team#2 reader=red-team Review-Finding: p1/red-team#3 reader=red-team Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * fix: pre-landing review fixes (Step 11: re-check refuses binary paths and packets over 64 KB; fingerprint trailers bind findings with a ledger_id; the runs fence closes only on its own line) Review-Finding: v/outside-voice#1 reader=outside-voice Review-Finding: v/outside-voice#2 reader=outside-voice Review-Finding: v/adversarial-native#1 reader=adversarial-native Review-Finding: v/targeted#1 reader=targeted Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * docs: J1 re-check eligibility names every refusal (review fixes) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * test: explicit timeouts on the re-check and result-head CLI tests (each spawns several git and bun runs; the carry tests ran past bun's 5 s default locally) Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com> * ship: trim 11 skeleton bytes so the merged tree fits the 61,420 cap (GSTA-258 L4) The rule 19 merge of main 1ccab0d put #3929's push-guards wording (61,418 on main) together with J1's head=<sha> in the GSTACK_RESULT line, measuring 61,429. Three rewordings beside that line bring the skeleton back to 61,418; the branch ledger entry drops by 11 and the ship goldens are regenerated. 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>
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.
Why
Pull requests kept rerunning work CI had already done, or carrying lanes CI never really ran. Guard-only repos like founderstar reran about six minutes of suites per /ship. Lanes with no test files, like gbrain's verify and typecheck (run 41564913688), could not carry at all. alphaclaw's 46-test lane recorded the run's 9677. harness-evals' 75 runner-skipped tests counted as passed. truther's path-filtered jobs could not be carried safely. This PR makes CI carry cover those cases and fail closed everywhere else.
Summary
ship_lanes_frommaps is carried from the verified CI run at this tree, through the samegstack-evidence record-cipath /ship lanes use. A local pass of the same command at the same tree is reused. CI rows never count through the ledger, because a CI pass counts only throughrecord-ci's live check.ship_lanes_fromentry can name the CI jobs that run a lane ("jobs": ["verify", "typecheck"]) instead of an inventory. It carries only when every named job succeeded in the newest attempt at HEAD's tree, and reportstests_ran=n/a.aliasesdeclare other exact spellings of a lane.--policy-from head --simulateproves a new map without writing or clearing anything. It needs--allow-repo-commandsto run a branch inventory.tree.jsonskipped_filesare treated as uncovered. A lane whose entry declaresper_filecarries the covered files and runs only the skipped ones here. Withoutper_filethe lane refuses asinventory_skippedand runs whole.skippablelets a path-skipped job count only in a pull_request run whose gate job succeeded and whose gate inputs are unchanged from base.How the rollout is split. This PR ships the reader inert for gstack itself. gstack's own
.gstack/pregate.jsondeclares none of the new keys, and gstack's own Free Tests writer (scripts/ci-tree-reuse.ts, a gstack-internal producer) does not emitskipped_filesyet. So until the adoption PR lands, gstack's own lane behaves exactly as it does today. The skipped-files warning (N inventory file(s) skipped every test in CI) needsskipped_filesin tree.json, so it starts printing for gstack with the adoption PR's writer, not with this one. A follow-up adoption PR turns on the writer, declaresper_filefor gstack'sbun run testlane and raisesmin_gstack_versiontogether. It waits until the pull requests now queued on older gstack (#3911, #3920, #3926, #3917) have landed, because an older gstack rejects a new key in main's policy.Blast radius
Repos that map lanes or guards in
ship_lanes_from. Carry now also refuses for a skipped inventory file, a gate input that changed, a non-pull_request skip, or a zero-test remainder. Each refusal makes the lane run locally, never pass. Security notes (R45):record-cican now run the base policy'sper_filecommand locally. The skippable check reads workflow and script blobs from git.--simulateruns a branch's inventory command only with--allow-repo-commands. Nothing reads secrets or widens permissions.Test Coverage
test/ship-ci-carry.test.ts(+451 lines): guard carry in the founderstar shape (guards only, no lanes); jobs-mapped lanes (all jobs green carries; a missing, skipped or failed job refuses); aliases; narrow file counts;--simulatewrites and clears nothing and runs a branch inventory only with--allow-repo-commands;skipped_fileswith and withoutper_file(a failing or zero-test remainder carries nothing;run_changedafter the remainder);skippablegate inputs through${{ github.workspace }},working-directoryand package scripts;passednever counts a CI row.test/ci-carry.test.ts,test/evidence-reuse.test.ts,test/run-evidence.test.ts,test/helpers/fake-gh-runs.ts: the reader's new fields and refusal codes.a working-tree edit during the post-remainder run recheck carries nothing(0f6ad9d): before the fix, record-ci exited 0 and wrote the CI row; after it,CARRY_NONE: tree_changed.Pre-Landing Review
REVIEW_LOOP: passes=2 blocking_found=8 deferred=0 noted=6 fixed=12 scope=frozen unrecorded=3 ledgers_read=3 fix_checks=3 dispatches=4
Full depth (gate class). Round 1 found three CRITICAL findings and one Medium (missed gate inputs,
--simulaterunning a branch inventory without consent,passedcounting CI rows, push-event skips); all were fixed. Round 2 found two CRITICAL findings (a zero-test remainder carried, a run could move to a new attempt during the remainder); both were fixed. A scoped fix-check (grant a-21d3128888cd) found one more CRITICAL: record-ci checked the working-tree fingerprint before its post-remainderrunChanged()calls, so an edit made during them still carried. A second scoped fix-check (grant a-860fbd99026c) covered its fix in 0f6ad9d, which moves the check after those calls, and found nothing at the floor. After the base merge, one scoped reader (grant a-53beaec71a54) read the windows-curation entry at 5024bbb and found nothing.Base merge (rule 19, GStack Minder's grant at f6d6215). b01ed69 merges main. #3911 conflicted in lib/pregate/index.ts and lib/pregate/run-check.ts, and both sides are kept:
RunCheckhas bothunchanged?andfailures?,ClearDecisiontakes main'sexception, andbuildRunCheckpasses bothfailures: ghArtifactDeps(repo)andunchanged. 5024bbb then excludes test/ship-ci-carry.test.ts from the Windows-safe list, like its ship-bin-fixture siblings, because #3904's path-scoped Windows check ran it and every case fails at its first spawn there (windows-scoped-shard, run 41607840098).Review-loop exception: accepted by GStack Minder 2026-11-03 via GSTA-258 (grant a-53beaec71a54): scoped reader on the windows-curation exclusion for test/ship-ci-carry.test.ts (windows-scoped-shard run 41607840098) (fix-check at 5024bbb, 0 findings)
Open notes (not fixed in this PR):
skipped_zero_runinstead oftree_changed. It still fails closed and writes no CI row.tree_changedor the order of the rechecks; that goes in the follow-up L6 PR (v1.94.26.0 gstack-evidence: sum sharded bun lanes; docs: dispatch-carry warning and record-ci recheck order (GSTA-258 L6 follow-up) #3934).recordCiResultdrops the CARRY_NONE reason when a look-alike record line comes before a refusal (lib/ci-carry.ts:165), plus two notes at bin/gstack-evidence:717 and :719.spawnSyncwithout process-group supervision.tree_changedduring the remainder run.skipped_jobs; a skipped job normally uploads no tree.json, so the inventory check refuses it anyway.Issues
Closes GST-1186, GST-1189 and GST-1190 (Capy issues).
Refs #3923
Refs #3929
Documentation
Scope: docs inputs changed by the f6d6215 base merge (bin/gstack-ship-receipt, docs/pregate.md, lib/pregate/index.ts, lib/pregate/run-check.ts, lib/run-evidence.ts) and the record-ci recheck order at 0f6ad9d, after audit l6cdocs2 covered a8d2fee (docs/pregate.md, docs/troubleshooting.md, ship/SKILL.md.tmpl Step 17, ship/sections/tests.md.tmpl, CONTRIBUTING.md, ship/DRIVER.md.tmpl, docs/ship-policy.md, docs/standalone-binaries.md).
Result: blocked. Audit l6cdocs2 updated docs/pregate.md (#ci-carry, #ship-lanes-jobs, #ship-lanes-aliases, #prove-a-mapping, #skipped-files, #skippable, #guard-carry) and docs/troubleshooting.md (#pregate-gstack-too-old) at a8d2fee, with no blockers. The later fix at 0f6ad9d moves one record-ci refusal check that no doc describes (accepted by GSTA-258 as a named risk), and the base merge then changed docs inputs; the audit budget was used.
Documentation exception: accepted by GSTA-258 2026-11-03 via GSTA-258 message to L6 (docs named risk, #3929 base merge): the f6d6215 base merge changed docs inputs (main's #3911 edits and the both-sides resolution); #3929's own docs are unchanged since l6cdocs2
Debt: no how-to yet for adopting jobs, skippable and per_file (suggest
/document-generate); docs/standalone-binaries.md and ship/sections/tests.md do not mentiongstack-evidence passed/carriedor the narrowtests_ran (files)count.Test plan
LANE: free carried run=41609778840/1 tests_ran=34727; every check green there, windows-scoped-shard includedgstack-pregateverify ok on the stamped tree; remote gate cleared from run 41609778840 (skipped=113); macOS lanes from run 41609778840, Windows smoke from run 41609778927bun run typecheckandtypecheck:testcleangstack: ok 1.94.0.0 for=ship