Skip to content

Commit b721ca6

Browse files
mchwangclaude
andcommitted
Scan from the original base for a pre-v19 opening, never another branch's check
No record shows which base branch a check recorded before schema v19 scanned: a publish can record its check and stop before opening, so the task's oldest clear check may belong to another base branch. Start the legacy scan at the task's original base instead, which covers every base commit, and drop the now-unused Store.oldestClearAlreadyFixed. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
1 parent f684da4 commit b721ca6

5 files changed

Lines changed: 35 additions & 11 deletions

File tree

‎docs/implementation/guarded-merge.md‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -27,7 +27,7 @@ The target is resolved again for each validation pass. Both passes, and the thir
2727

2828
Merging blocks when any plan item is unreviewed or stale, an attributed file is outside that item's declared plan scope, an Ambiguous or Unplanned segment remains, a current change request is open, or a `cmd:` acceptance check lacks a passing result for the current head. Approval does not override an out-of-scope file; the plan must be amended. Historical change requests from an earlier revision or snapshot remain visible but do not block the current revision.
2929

30-
The GitHub adapter reads the current PR base/head, effective branch rulesets, classic branch protection, required check contexts and app identities, and issue cross-references. It ignores review records that are not required checks. The already-fixed check is the pre-PR check, `GhAlreadyFixedGateway` in `github/already-fixed.ts`, with the same rules (see “The check” in `pull-request-opening.md`): a closed issue, another open or merged PR that would close the issue or is linked manually, or a new commit on the base branch that mentions the issue. A PR that only mentions the issue does not count, in this repository or another; when the PR's base is not the default branch, a PR in this repository into that same base that references the issue counts too (decided 2026-09-30). The merge passes this PR, and every other PR the task records in this repository (#121), as the task's own PRs, matched by repository and number: each is excluded while open and counts once merged, as in the pre-PR check. Without a runner block only this PR is passed. Without a runner block, the commit scan starts at the PR's base commit on its base branch, and no own commits are passed. For a runner-published PR, it starts at the base-branch tip that the pre-PR already-fixed check of the task's first opening into the current base branch scanned through (each opening records its check from schema v19; for an earlier first opening, the task's oldest clear check stands in) when that check scanned from the task's original base, and otherwise at the task's original base, so base commits that a pre-merge rebase brought under the PR are covered too, whichever of the task's openings the merge follows (F6b of #22, `merge-handoff.md`); the ledger's owned commits are passed as the task's own, as in the pre-PR check. Inspections that follow a submitting, queued or merged attempt's pinned pull request scan from the PR's base, as without a runner block; they only show status, since no merge is asked for while an attempt is active. Any result the check cannot complete is unknown and blocks merging, including a base branch name outside the pattern the check accepts. The check starts as soon as the PR is read, alongside the rule reads, and stops 0.65 seconds before the inspection's deadline (`MERGE_CHECK_SETTLE_MS` in `github/merge.ts`: both grace periods of the merge's `gh` runner, which the check also uses, plus a margin), so its processes settle before the inspection ends; a check stopped that way is unknown. If the rule reads fail, the check is stopped and awaited before the inspection rejects. When the PR read ends too late to leave that much time, the check is not started and is unknown. A check answer that arrives after the check was stopped is unknown, and an inspection whose caller aborted or whose deadline passed rejects even when its reads answer late. An injected check must name the merge repository. The blocker names what the check found (at most five closes, PRs and commits, by repository, number, state and short SHA, with a closing commit's SHA shortened too; never a commit message) or why it could not finish. Once this PR has merged, the check counts it as the fix, so the coordinator shows only the `pr-state` blocker, not an already-fixed one. Malformed pagination, rules without a type, malformed protection objects, missing or malformed check-app identities, and inconsistent check status/conclusion pairs fail closed, while an explicit `null` remains an unbound check. Display status is cached for five seconds and the combined inspection has a 12-second deadline. Each of the two sequential merge-validation inspections has a six-second deadline, keeping their combined validation budget below the 15-second serving request budget. A deadline aborts the underlying GitHub CLI process before releasing the shared in-flight request. A caller's own abort during the already-fixed check waits for its processes like any other `gh` call of the gateway, up to 0.4 seconds. An unreadable or incomplete rule source, a pending/missing/failing check, an already-fixed match or unknown result, a conflict, or a closed PR blocks merging.
30+
The GitHub adapter reads the current PR base/head, effective branch rulesets, classic branch protection, required check contexts and app identities, and issue cross-references. It ignores review records that are not required checks. The already-fixed check is the pre-PR check, `GhAlreadyFixedGateway` in `github/already-fixed.ts`, with the same rules (see “The check” in `pull-request-opening.md`): a closed issue, another open or merged PR that would close the issue or is linked manually, or a new commit on the base branch that mentions the issue. A PR that only mentions the issue does not count, in this repository or another; when the PR's base is not the default branch, a PR in this repository into that same base that references the issue counts too (decided 2026-09-30). The merge passes this PR, and every other PR the task records in this repository (#121), as the task's own PRs, matched by repository and number: each is excluded while open and counts once merged, as in the pre-PR check. Without a runner block only this PR is passed. Without a runner block, the commit scan starts at the PR's base commit on its base branch, and no own commits are passed. For a runner-published PR, it starts at the base-branch tip that the pre-PR already-fixed check of the task's first opening into the current base branch scanned through (each opening records its check from schema v19) when that check scanned from the task's original base, and otherwise, including a first opening recorded before v19, at the task's original base, so base commits that a pre-merge rebase brought under the PR are covered too, whichever of the task's openings the merge follows (F6b of #22, `merge-handoff.md`); the ledger's owned commits are passed as the task's own, as in the pre-PR check. Inspections that follow a submitting, queued or merged attempt's pinned pull request scan from the PR's base, as without a runner block; they only show status, since no merge is asked for while an attempt is active. Any result the check cannot complete is unknown and blocks merging, including a base branch name outside the pattern the check accepts. The check starts as soon as the PR is read, alongside the rule reads, and stops 0.65 seconds before the inspection's deadline (`MERGE_CHECK_SETTLE_MS` in `github/merge.ts`: both grace periods of the merge's `gh` runner, which the check also uses, plus a margin), so its processes settle before the inspection ends; a check stopped that way is unknown. If the rule reads fail, the check is stopped and awaited before the inspection rejects. When the PR read ends too late to leave that much time, the check is not started and is unknown. A check answer that arrives after the check was stopped is unknown, and an inspection whose caller aborted or whose deadline passed rejects even when its reads answer late. An injected check must name the merge repository. The blocker names what the check found (at most five closes, PRs and commits, by repository, number, state and short SHA, with a closing commit's SHA shortened too; never a commit message) or why it could not finish. Once this PR has merged, the check counts it as the fix, so the coordinator shows only the `pr-state` blocker, not an already-fixed one. Malformed pagination, rules without a type, malformed protection objects, missing or malformed check-app identities, and inconsistent check status/conclusion pairs fail closed, while an explicit `null` remains an unbound check. Display status is cached for five seconds and the combined inspection has a 12-second deadline. Each of the two sequential merge-validation inspections has a six-second deadline, keeping their combined validation budget below the 15-second serving request budget. A deadline aborts the underlying GitHub CLI process before releasing the shared in-flight request. A caller's own abort during the already-fixed check waits for its processes like any other `gh` call of the gateway, up to 0.4 seconds. An unreadable or incomplete rule source, a pending/missing/failing check, an already-fixed match or unknown result, a conflict, or a closed PR blocks merging.
3131

3232
Automatic merging also requires strict required status checks as a server-enforced current-base policy. A client-side fetch cannot supply that guarantee. Merge-queue branches were blocked in this increment because `gh pr merge` can report a successful enqueue before the pull request is merged. Queue support has since landed (#24; see `merge-queue.md`): a queue-capable adapter tracks each attempt until GitHub confirms the merge, removal or failure, and queue rules count as the server-side current-base guard. An adapter without queue inspection still fails closed with a `merge-queue` blocker. The coordinator performs the complete gate twice without using the display cache, verifies the same base/head pair, then re-reads the local review generation immediately before invoking `gh pr merge --match-head-commit` with literal argv. Each `gh` process gets only the allowlisted variables in `github/gh-env.ts`, plus fixed settings that turn off prompts, the pager, colour and update checks. No other server variables are passed on. A timeout or abort sends the `gh` process SIGTERM, then SIGKILL after a quarter of a second, and the call returns only after that process has exited (or 0.15 s after it exits, if a process it started keeps its output open). So a merge call aborted at its 14-second deadline settles by 14.4 s. *Later (#48, #170):* a click can also be waiting on its issue-trust read, whose stop wait is 0.75 s, or on a review load, whose history worker gets 1 s after the abort before it is terminated; so it settles about 0.75 s after its deadline, or about 1 s or a little more after it when a review load was running, past the 14.5-second shutdown drain (traced, not tested); see [merge-queue.md](merge-queue.md). Stopping `gh` cannot recall a merge request it had already sent: GitHub can still apply it, so a cancelled merge is reported as unknown and its attempt stays in flight, with the action disabled. A direct attempt is settled only when GitHub shows the pull request merged; a queued attempt also settles when GitHub reports it removed or failed (`merge-queue.md`). The adapter invalidates its pre-action cache before the command and after either success or refusal. A generation guard prevents inspections started before or during the command from restoring stale cache entries afterward. After command success, the server returns “merge submitted” even if its best-effort status reload fails, and the browser keeps the action disabled until Refresh confirms GitHub state. GitHub's command refusal text is returned to the local UI, and a failed attempt also remains disabled until Refresh loads fresh state. There is no “merge anyway” path around missing atomic protection. Demo mode never constructs the coordinator, including when a gateway is injected.
3333

‎docs/implementation/merge-handoff.md‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -53,7 +53,7 @@ Moving the task to **approved, merge blocked** advances its state version, so th
5353

5454
## The already-fixed range
5555

56-
A rebase moves the pull request's base. The guarded merge would then scan only the base commits after the new base and miss the ones the rebase brought in. The scan must cover every base commit after the task's original base (the base of its first snapshot), since a rebase can bring any of them under the pull request. Each opening records the pre-PR check it followed (schema v19). That check of the task's **first opening into the current base branch** scanned the branch from its snapshot's base to the branch's tip then, and found nothing. When it scanned from the task's original base, each merge inspection scans from that tip to the branch's tip now. A later rebase onto the branch can bring in any commit after the original base: those up to that tip were scanned by that check, and later ones are in the merge's scan, so the scan covers them whichever of the task's openings into that branch the merge follows (after a re-publish recorded a newer check from a rebased base, and after a person closed the pull request and a new one was opened from a rebased base). A newer check is not a safe start: it scanned only from a rebased base, and a compare range excludes its own start commit. An opening into another base branch, left by a change of the `github.baseBranch` setting, says nothing about this branch and is not used. A first opening recorded before schema v19 has no check; the task's oldest clear check stands in. When there is no such check, or it scanned from a rebased base (the base branch setting changed after a rebase), the scan starts at the task's original base. If that start is not on the base branch (for example, it was force-pushed), the answer is unknown and merging stays blocked. The check reads at most 250 base commits: if more landed since the scan's start, its answer is unknown and merging stays blocked, as for a pull request never rebased. There is no person override yet (the Continue action of `pull-request-opening.md` is a later slice). The task's own commits, the ledger's owned entries, are excluded, as in the pre-PR check. The status shown while a merge attempt is submitting, queued or merged scans from the pull request's base, as before: those inspections follow the attempt's pinned pull request, and no merge is asked for while an attempt is active.
56+
A rebase moves the pull request's base. The guarded merge would then scan only the base commits after the new base and miss the ones the rebase brought in. The scan must cover every base commit after the task's original base (the base of its first snapshot), since a rebase can bring any of them under the pull request. Each opening records the pre-PR check it followed (schema v19). That check of the task's **first opening into the current base branch** scanned the branch from its snapshot's base to the branch's tip then, and found nothing. When it scanned from the task's original base, each merge inspection scans from that tip to the branch's tip now. A later rebase onto the branch can bring in any commit after the original base: those up to that tip were scanned by that check, and later ones are in the merge's scan, so the scan covers them whichever of the task's openings into that branch the merge follows (after a re-publish recorded a newer check from a rebased base, and after a person closed the pull request and a new one was opened from a rebased base). A newer check is not a safe start: it scanned only from a rebased base, and a compare range excludes its own start commit. An opening into another base branch, left by a change of the `github.baseBranch` setting, says nothing about this branch and is not used. A first opening recorded before schema v19 has no check, and no record shows which base branch an older check scanned (a publish can record its check and stop before opening), so no other check stands in. Then, and when the first opening's check scanned from a rebased base (the base branch setting changed after a rebase), the scan starts at the task's original base. If that start is not on the base branch (for example, it was force-pushed), the answer is unknown and merging stays blocked. The check reads at most 250 base commits: if more landed since the scan's start, its answer is unknown and merging stays blocked, as for a pull request never rebased. There is no person override yet (the Continue action of `pull-request-opening.md` is a later slice). The task's own commits, the ledger's owned entries, are excluded, as in the pre-PR check. The status shown while a merge attempt is submitting, queued or merged scans from the pull request's base, as before: those inspections follow the attempt's pinned pull request, and no merge is asked for while an attempt is active.
5757

5858
## Cancel, shutdown and restart
5959

‎runner/merge.ts‎

Lines changed: 4 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -130,14 +130,15 @@ export class MergeCoordinator {
130130
* nothing; when that base is the task's original base, the scan starts at that tip: commits a later rebase onto the
131131
* branch brings in up to that tip were scanned by that check, and later ones are in this scan, whichever of the
132132
* task's openings the merge follows. An opening into another base branch says nothing about this one. A first
133-
* opening recorded before schema v19 has no check: the task's oldest clear check stands in. Otherwise (no such check, or one that scanned from a rebased base), the scan starts at the
134-
* task's original base. The task's own commits are excluded.
133+
* opening recorded before schema v19 has no check, and no record shows which base branch an older check scanned (a
134+
* publish may record a check and stop before opening), so no other check stands in. Then, and when the check scanned
135+
* from a rebased base, the scan starts at the task's original base. The task's own commits are excluded.
135136
*/
136137
#alreadyFixedRange(here: TaskPullRequest[], base: string): { alreadyFixedSince: string; ownCommits: string[] } {
137138
const { store, config } = this.service, identity = config.identity;
138139
const original = store.firstSnapshot(identity).base;
139140
const first = here.find(pr => pr.base === base);
140-
const check = first?.checkId ? store.getAlreadyFixed(identity, first.checkId) : store.oldestClearAlreadyFixed(identity);
141+
const check = first?.checkId ? store.getAlreadyFixed(identity, first.checkId) : null;
141142
const ownCommits = store.getLedger(identity).filter(entry => entry.origin === 'owned').map(entry => entry.sha);
142143
if (check?.result.outcome !== 'clear') return { alreadyFixedSince: original, ownCommits };
143144
let scannedFrom: string | null = null;

‎runner/store.ts‎

Lines changed: 0 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -2356,10 +2356,6 @@ export class Store {
23562356
* which did not record its own check. A compare range excludes its start commit, so only a scan from this check's tip
23572357
* onward is provably continuous with what was scanned before.
23582358
*/
2359-
oldestClearAlreadyFixed(identity: PlanIdentity): AlreadyFixedCheck | null {
2360-
const row = this.#get(`SELECT * FROM already_fixed_checks WHERE plan_key=? AND outcome='clear' ORDER BY rowid ASC LIMIT 1`, identityKey(identity));
2361-
return row ? { id: row.id as string, snapshotId: row.snapshot_id as string, result: decode<AlreadyFixedResult>(row.result), stateVersion: row.state_version as number, reviewVersion: row.review_version as number, checkedAt: row.checked_at as string } : null;
2362-
}
23632359
getAlreadyFixed(identity: PlanIdentity, id: string): AlreadyFixedCheck | null {
23642360
const row = this.#get('SELECT * FROM already_fixed_checks WHERE plan_key=? AND id=?', identityKey(identity), id);
23652361
return row ? { id: row.id as string, snapshotId: row.snapshot_id as string, result: decode<AlreadyFixedResult>(row.result), stateVersion: row.state_version as number, reviewVersion: row.review_version as number, checkedAt: row.checked_at as string } : null;

0 commit comments

Comments
 (0)