Two steps publish a commit and treated every failed push as a moved branch:
the release review job (release-review.yml) retried three times with "dev
went ahead", and the review document step (_process.yml) rebased and pushed
again. A refusal by GitHub itself - a token without the workflow right, a
branch rule, a hook - cannot be cured by a retry or a rebase, and the step
never said what GitHub answered.
Both pushes now keep stderr and hand it to the #705 classifier through the
same CLI the rebase guard uses (merge-candidate.mjs --push-refusal). Only a
stale lease (rejected / fetch first / stale info) keeps the old retry or
rebase. Any other outcome stops the step at once, without retries: the log
gets the git answer and the step summary gets the reason and the git answer,
both passed through redactSecrets (token, credential URL, Authorization).
The review document step takes the classifier from dev, as the rebase guard
does: a task branch behind dev may not carry it.
The summary text is written by the new --summary option (refusalSummary),
not by a multi-line string in run:, and both commit messages are now built
line by line into a file instead of a heredoc (PROCESS.md §10.4 item 4).
release-review.yml is dispatch-only and is not mirrored to main. PROCESS.md
names the rule next to the rebase guard; the #638 trailer witness in
test/release-review.test.mjs follows the line-by-line message.
test/publish-push-refusal.test.mjs runs both steps as they are with real
bash and real git in temporary repositories; only the push transport is
replaced: a moved branch is a real neighbour push, a GitHub refusal is a
recorded stderr carrying a token, a credential URL and an Authorization
header. On the old steps 9 of its 11 tests fail.
Issue: #723
User-Visible: no
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
merge-candidate treated any push stderr containing "rejected" as a stale
lease. A `! [remote rejected]` from GitHub itself - in #700 the rebased
candidate changed .github/workflows/ and the conveyor token has no workflow
permission (runs 36484993494, 36487044060) - became "the branch moved after
the reviewed material (#312)", and the stderr was never printed, so the
author was sent to look for a commit that did not exist.
classifyPushRefusal now tells three outcomes apart: a stale lease
(`[rejected] (stale info)`, `fetch first`, a server-side lock race) keeps
the old behaviour; GitHub's workflow refusal (PAT, OAuth App, GitHub App,
bot and integration wordings) and any other `[remote rejected]` get their
own outcome, S6-in-progress and a comment naming the reason. The workflow
comment says what to do: the author rebases and pushes, or the owner grants
the permission. The git answer goes to the log and the comment with tokens
and credential URLs cut out; the merge-step failure comment is redacted too.
The rebase guard in _process.yml parses its push refusal with the same code
(`merge-candidate.mjs --push-refusal`): a stale lease is the old error, a
workflow refusal returns the task to S6 without review like a conflict, and
material/reuse/gate skip the rebase that never reached the branch.
Mutant push-refusal-kinds-glued restores the old regex; guard: #705 AC1.
Issue: #705
User-Visible: no
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
370 merged issue/* branches sat on origin; the branch list stopped meaning
anything and an agent looking a branch up by number could take a stale one.
- merge-candidate.mjs: after a successful push to dev the task branch is
deleted with --force-with-lease on the tip the merge saw last — the
candidate published into the branch, or the material on fast-forward
(the index commit lives only in dev). A commit that landed after the
merge keeps the branch, and the merge comment says so; a failed delete
never undoes the merge. Failed, stale and conflicting merges keep it.
- The one-time cleanup of the already merged branches is not in this
commit: the list goes to the owner first.
Canon: PROCESS.md §10.4 (exact-candidate merge).
Issue: #702
User-Visible: no
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
14 of 48 returns in #600–#691 were rebase or merge conflicts on shared
files where the two edits do not contradict each other.
- .gitattributes: docs/CHANGELOG.md and docs/CHANGELOG.ru.md use the
built-in merge=union driver — both tasks' lines in ## Unreleased survive
a rebase, a merge and git merge-tree (#696's clean-merge test) without a
stop.
- rebase-generated.mjs: UPSTREAM_WINS — on a conflict in
scripts/monolith-baseline.json the rebase takes dev's side; the band test
on the candidate's Validate judges the merged tree (#699). Any other
conflicting path aborts exactly as before, with the full list.
- merge-candidate.mjs: the candidate's patch-id excludes the changelogs and
the monolith baseline next to docs/reviews, so a neighbour's line next to
the task entry does not re-send a green task to review.
- screenshots.json needs nothing: after #697 task branches do not commit it.
Canon: PROCESS.md, the rebase paragraph of the review index (#643).
Issue: #698
User-Visible: no
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
show/ship stop paying for diff mutants and for every move of dev:
- scripts/process-track.mjs resolves the track from the current labels and
the diff (show for unlabelled infra, ask for unlabelled product work) and
checks the mechanical ship limits; outside them the pipeline comments and
relabels track:ship -> track:show in the same round.
- Validate on the review material is light on show/ship: a completed push
run on the exact SHA is proof, a dispatch asks mutants=false. ask and the
ci:mutants label keep the mutant dispatch.
- show/ship skip the pre-review rebase when git merge-tree with dev is
clean; the candidate is rebased once at merge and still passes Validate
before the push to dev. The light merge waits for the push run of the
candidate and dispatches only when none appears.
- ship inside the limits merges after the light Validate without a model
review; the issue gets a machine marker hp:ship-merge.
- ship-review.yml + scripts/ship-review.mjs read the code of all ship
tasks of a beta range in one model session and publish
docs/reviews/SHIP-REVIEW-<tag>.md; both beta publication paths refuse a
range with ship tasks the document does not cover or that carries a High.
- show reviews judge correctness and AC; the spec review installs neither
npm ci nor Chromium, the show review installs Chromium only when the issue
names a smoke.
Canon: PROCESS.md §5, §5.1, §10.4, new §11.7; REVIEWER.md, AUTHOR.md and
AGENTS.md digests.
Issue: #696
User-Visible: no
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
Волна 3 эпика #674. AGENTS.md 650 → 187 строк: карта пакета, маршрут чтения,
правило №1, классы и треки одной строкой со ссылками, трейлеры, рабочие
деревья, хендофф и ожидание вердикта; пересказы PROCESS.md — ссылками на
разделы. Неверный список «Gate jobs» снят (списки jobs не копируются в прозу,
шапка PROCESS.md). Правила, жившие только в AGENTS, получили дом: жёлтый
вердикт при выполненных AC — PROCESS §2.7; свежесть бандла, съёмка только в
Linux (#455, HP_ALLOW_FOREIGN_CAPTURE) и смоки из AC до S7 (#151) —
TESTING.md; причуда демо-стенда и среда-зависимый smoke_opening_measure —
DEVELOPMENT › Smoke tests; отказ публикации без `Release:` и при несвежем
отпечатке бандла, отмена Validate новым пушем, кандидат беты не
promotion-only, fail-closed реестра Labs — DEVELOPMENT; предупреждение и
ошибка свежести скриншотов — CONTRIBUTING.
PROCESS.md: §13 (внедрение с открытым ⏳), §14 (блок со ссылкой на
несуществующий docs/PROCESS.md) и §7.3 (история) удалены. Ссылки «§7.2» на
правило полного разбора после ребейза ведут в §2.10, на сверку SHA перед
выводом — в §2.7; то же в сообщениях scripts/branch-state.mjs,
merge-candidate.mjs, review-doc-guard.mjs, pre-push-gate.mjs, в промпте
_process.yml и TESTING.md. Число `any` в прозе → `node scripts/no-new-any.mjs
--total` (новый режим, юнит-тест; было «1034 в 49 файлах», сейчас 862 в 52),
дата-число замороженного списка якорей монолита снято. Устаревшая команда
пересъёмки скриншотов в §8 заменена ссылкой на действующий путь.
STATUS.md 113 → 61 строка: сгенерированный снимок, текущий цикл и девять
строк решений; Workflow, CI, Toolchain, Tests, Scope, open items и политика
документации — ссылками (PROCESS §2.6, DEVELOPMENT › Release, TESTING);
локали en/ru/de/fr; закрытые «coverage, mypy strict» сняты.
DEVELOPMENT.md: file-sync и «Reproducible scripts» (прототип) удалены;
раздел Release — единственный дом релизной механики: введение, правила
тела стабильного релиза (#328, release:notes), шаг continuity:screencast,
источники версии по release-contract. CONTRIBUTING: ссылка на Release вместо
пересказа, замеры клона без чисел. TESTING: any-гейт — ссылкой на PROCESS §8.
entry-cost: автор 11 125 → 5 407 слов, ревьюер 8 464 → 4 285.
Issue: #680
User-Visible: no
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
The task branch no longer carries docs/reviews/INDEX.md (1b), and
merge-candidate rebuilt it only inside rebaseOnto. When dev did not move
the fast-forward pushed the stale index and reviews_index would turn dev
red. freshIndex(tip) commits the rebuilt index on top of the material
before the push; the merge stays a fast-forward.
Real-git test: fast-forward, then reviews-index --check on the dev head
is green. Mutant merge-ff-skips-review-index.
Issue: #657
User-Visible: no
A pipeline doc commit carries the review document and the rebuilt
INDEX.md; while the task waits, dev receives other tasks' documents with
their own INDEX.md, and the rebase of the branch conflicts in the index
every time. 24.09 this bounced green #617, #618, #629, #642 to S6.
scripts/rebase-generated.mjs: shared rebase helper. At every stop, if ALL
conflicting paths are docs/reviews/INDEX.md (or paths the caller
resolves itself), the index is rebuilt from the directory in the stop
tree, staged, and the rebase continues; any other path aborts and
returns the full list. CLI exit 3 = refusal with paths on stdout.
Wired into process.yml «Привести ветку к dev» (helper taken from dev via
git archive; conflict/conflicts outputs, lease, ref wait and
--commit-if-stale kept), merge-candidate rebaseOnto (claude[bot]
identity, --commit-if-stale kept) and rebase-on-dev.mjs (index next to
GENERATED_ROOTS; bundle still dev copy + rebuild).
Issue: #643
User-Visible: no
r2 H1: INDEX.md — снимок каталога, и ребейз ветки на dev, получивший чужие
документы ревью, устаревал его молча. Теперь `--commit-if-stale` пересобирает
и коммитит индекс коммитом конвейера после приведения к dev (process.yml) и
после ребейза кандидата (merge-candidate.mjs); тест «индекс свеж» сравнивает
закоммиченный файл с пересборкой и красит Validate при расхождении.
r2 M1: находка без заголовка — первый абзац секции, склеенный из перенесённых
строк, без маркера буллета и кода `**M1.**`; «не найдено», служебные скобки
«(унаследовано…)» — не находка. Нумерованные пункты тоже забирают перенесённые
строки. Мутант reviews-index-paragraph-tail. PROCESS.md §2.10 дополнен.
Issue: #635
User-Visible: no
Шаг слияния конвейера считает patch-id материала и кандидата через
`git diff --full-index`, когда `dev` сдвинулся за время ревью. Дифф задачи,
пересобирающей бандл, несёт три копии `houseplan-card.js` — у #594 это
7 103 616 байт. `sh()` не задавал `maxBuffer`, умолчание `spawnSync` — 1 МиБ,
и процесс убивался по ENOBUFS на 1 027 868 байтах.
Вторая половина дефекта прятала первую: `r.error` нигде не читался, а
`status: r.status ?? 1` выдавало `null` за «git вернул 1». `must()` брал
`r.stderr || r.stdout`, stderr был пуст — и в issue уезжало
«Слияние не выполнено: diff: <огрызок диффа>». Огрызок выглядит осмысленным
и уводит разбор в сторону.
Путь исполняется только при сдвинувшемся `dev`, поэтому дефект дремал. Но
конвейер сам пушит в `dev` документ ревью на каждом раунде, так что
`devMoved` — норма, а не исключение.
Что изменено:
- `MAX_COMMAND_OUTPUT_BYTES` (256 МиБ) и явный `maxBuffer` в `sh()`;
- `sh()` называет сбой запуска в `stderr` результата вместо молчания;
- `sh()` экспортирован, у обоих пунктов есть исполнимый свидетель;
- мутант `merge-candidate-truncates-the-candidate-diff` возвращает 1 МиБ.
Issue: #596
User-Visible: no
The candidate is the branch tip, which already carries the round's
CODE-REVIEW-N-rK.md; the material the reviewer read does not. With
docs/reviews in the diff the two patch-ids never matched once dev had
moved, so every green candidate went back to review whenever another
task published its own document in the meantime — #514 looped twice on
09.09 and #508 only merged when dev happened to stand still. The
patch-id now excludes docs/reviews, exactly like `reviewedFresh` next to
it; a real change of the patch under rebase still returns the task.
Mutant: merge-rereviews-own-review-doc.
Issue: #516
User-Visible: no
Code review r2 (M1): the cancelled-run filter in merge-candidate's
waitValidate had no test or mutant — every test replaced ops.waitValidate
with a fake. realOps now takes an injectable `exec` (default: the same
spawnSync wrapper) so the real implementation runs against scripted
`gh run list` answers: a cancelled dispatch is skipped and its
replacement followed; a lone cancelled run ends in `missing`, never red.
Mutant: merge-trusts-cancelled-dispatch.
Issue: #510
User-Visible: no
Code review r1 (M1): validate-gate.mjs and merge-candidate's waitValidate
read a `cancelled` dispatch run on the material as red, so a dispatch
replaced by the next one in the `validate-dispatch-<ref>` concurrency
group would have returned the task S7→S6 for nothing — the same class
#511 fixed in release-gate.mjs. Cancelled runs are now ignored: the gate
follows the replacement dispatch, or starts its own when there is none.
Mutant: review-returns-task-on-cancelled-dispatch.
Issue: #510
User-Visible: no
Validate ran the three "Мутанты по диффу" shards on every push of every
branch: 48 of 56 job-hours on 08–09.09, most of them cancelled by the
next push. Mutants now run when asked — pull requests, the nightly
schedule, a push carrying a `Release:` trailer, or a dispatch with
`mutants=true` (classify-changes.mjs → `mutants_requested`); an ordinary
push runs the light checks only.
The proof moves to where it is consumed. process.yml gets a gate after
the #499 reuse step: on the code stage it looks for a dispatch Validate
run on the exact material SHA whose mutant jobs executed and passed
(scripts/validate-gate.mjs); none → it dispatches one and waits; red or
missing → the task goes back S7→S6 with the run link and the review
cycle is not spent. Spec stage and the reuse fast-path skip the gate
(`proceed=true`); all later steps branch on `proceed` in place of the
old conflict conjunct only. merge-candidate.mjs dispatches Validate on
the pushed candidate and waits for that dispatch run.
PROCESS.md/AGENTS.md: review does not start on red code; one handoff —
one push.
Mutants: mutants-run-on-every-push, review-starts-on-red-validate,
review-trusts-push-run-without-mutants, merge-waits-push-run-without-mutants.
Issue: #510
User-Visible: no
Windows portability (the three red tests on the owner's machine at green CI):
- scripts/spawn-portable.mjs: isMainModule via pathToFileURL (the
`file://${argv[1]}` form gives file:///C:/C:/... and the CLI stays silent);
portableCommand — a shell only for npm/npx/.cmd, node and git run directly
(spawn via shell dropped the quotes of `node -e "…"`). Applied to
classify-changes, mutation-gate-report, review-doc-guard, check-inputs,
merge-candidate, gate-small, rebase-on-dev.
- the rebase-on-dev test pins core.autocrlf=false / core.eol=lf through
GIT_CONFIG_* for its temp repository instead of touching the user's git
config; test/windows-portability.test.mjs forbids both anti-patterns.
Pins: scripts/toolchain-pins.mjs reads Node/Python from validate.yml, the HA
stack from tests_backend/requirements.txt, Playwright/Chromium from the
lockfile — no second dictionary; `npm run toolchain:check` compares the
machine; .nvmrc/.python-version are derived and tested equal;
scripts/wsl-setup.sh provisions WSL/Linux with those pins.
scripts/task-packet.mjs: one derived view of an issue (status, rights, owner
decisions, branch vs dev, Validate on the tip, previous verdict with recorded
tree, AC → evidence, unwitnessed). scripts/wait-verdict.mjs: polls labels,
pipeline comments and optionally Validate, prints only on state change, exit
0/3/4, writes nothing.
gate:small --smokes: after build the browser phase runs bundle-sync and then
the directly matched and registered smokes, two at a time; broad matches stay
with the reviewer. package.json changed, so the bundle is rebuilt here.
Issue: #496
User-Visible: no
scripts/merge-candidate.mjs owns the review pipeline's merge: when dev
moved during review, the rebased candidate is pushed to the issue branch,
its diff is compared to the reviewed one by patch-id, Validate on that SHA
is awaited, and only then dev is advanced with --force-with-lease on the
base the candidate was built on — a rejected lease restarts, at most three
times. Every non-merge outcome moves the label with a comment, so the
"label always changes" invariant holds. nightly.yml now finds the Validate
run it dispatched and inherits its conclusion. Three mutants guard this.
Issue: #492
User-Visible: no
