Take-in waits end: a group whose members are all taken in ends itself, a failed start is asked again by the next event (issue 1585) #163

2026-11-07T11:31:27Z

mesh-tools commented

2026-11-07 11:31:27 +00:00

Issue 1585, notes 70 and 71: take-in waits that never end.

A group wait whose members are all taken in (a held member at or behind its repository's taken merge, or a member taken in by hand) ends itself at every evaluation, said as take-in.taken before its pending keys go; the merge behind it is taken in in the same evaluation (note 71: B0's group wait stood 01:48-12:25 with every member taken in).
A take-in at start that fails is said in the journal and the indexer goes on instead of exiting; what it left pending is asked again at the next event the indexer hears (merged message, toolchain report, hand take-in). No timer: ADR 0633 §5, ADR 0437 §8 (note 70).
The group-end probe is best effort: a forge that does not answer keeps that group's wait and is said; it never stops other take-ins.

Quick against thorough: quick chosen (within the records). The thorough fix, take-in items on the indexer's own work queue redelivered by the bus, needs ADR 0633 amended; it is written on issue 1585 note 73.

Tests: rows in TestTakeIn's 1585 subtest; no new test function, no module.json change. Local: go vet (with and without the integration tag) and go test ./... pass in the indexer module, but for judge's TestTheSeatsAreAClosedSetAndEachNamesItsDecision, which fails the same way on main bab05c66.

Ahead of fix-1965's registry group on main (its group merges pending.go and main.go after this lands).

The first check (at 0520c126, 11:35:21Z) passed every test and failed only its composition, which ran before the indexer's build src2:b5163570 was done (the checker's skipped build wait, issue 1921 note 74; fix-1965 fixes it). 70cd6241 is an empty commit on the reviewed head that asks the check again; it changes no file.

Reviewed-by: mesh-code-reviewer@shanks at 0520c12619
Reviewed-by: lead-47@shanks at 70cd624127 (empty commit, no diff against 0520c126)

Issue 1585, notes 70 and 71: take-in waits that never end. - A group wait whose members are all taken in (a held member at or behind its repository's taken merge, or a member taken in by hand) ends itself at every evaluation, said as take-in.taken before its pending keys go; the merge behind it is taken in in the same evaluation (note 71: B0's group wait stood 01:48-12:25 with every member taken in). - A take-in at start that fails is said in the journal and the indexer goes on instead of exiting; what it left pending is asked again at the next event the indexer hears (merged message, toolchain report, hand take-in). No timer: ADR 0633 §5, ADR 0437 §8 (note 70). - The group-end probe is best effort: a forge that does not answer keeps that group's wait and is said; it never stops other take-ins. Quick against thorough: quick chosen (within the records). The thorough fix, take-in items on the indexer's own work queue redelivered by the bus, needs ADR 0633 amended; it is written on issue 1585 note 73. Tests: rows in TestTakeIn's 1585 subtest; no new test function, no module.json change. Local: go vet (with and without the integration tag) and go test ./... pass in the indexer module, but for judge's TestTheSeatsAreAClosedSetAndEachNamesItsDecision, which fails the same way on main bab05c66. Ahead of fix-1965's registry group on main (its group merges pending.go and main.go after this lands). The first check (at 0520c126, 11:35:21Z) passed every test and failed only its composition, which ran before the indexer's build src2:b5163570 was done (the checker's skipped build wait, issue 1921 note 74; fix-1965 fixes it). 70cd6241 is an empty commit on the reviewed head that asks the check again; it changes no file. Reviewed-by: mesh-code-reviewer@shanks at 0520c126195be347b4a32383d4bd891089abcb9b Reviewed-by: lead-47@shanks at 70cd624127bf8382b49e496a12d9a97cb34c2368 (empty commit, no diff against 0520c126)

mesh-tools added 2 commits 2026-11-07 11:31:27 +00:00

A group wait whose members are all taken in ends itself, and the take-in at the start is tried again (issue 1585) a7aa836ff9

Group-end probe is best effort and a failed start is asked again by the next event, with no timer (issue 1585, review round 1)

mesh/delivery closed: a newer head of the same pull request

Details

mesh/check failed: novox: not composed on novox: mesh-delivery-indexer: frombus: mesh-delivery-indexer is wanted at src… | 0520c126195b on bab05c66b3cb

0520c12619

mesh-tools commented

2026-11-07 11:31:31 +00:00

Delivery novox/mesh-delivery@70cd624127bf — merged since 2026-11-07T11:42:24Z

Source — source.main moved to novox/mesh-delivery@b719dc0ecf. Whether it reached the machines is mesh-delivery-view's, derived from source.main, the wants and the haves.

Transitions

2026-11-07 11:38 (new) → proposed (announced): novox/mesh-delivery#163's head announced
2026-11-07 11:38 proposed → ready (accepted): a verdict names this commit and the trunk's head as its trunk commit, the gate passed or warned, and the repository's own check did not fail; or the check gates no merge (check-gates-merge false)
2026-11-07 11:42 ready → merged (merged): on the trunk its modules follow: merged there

The commit's note under refs/notes/mesh-plan keeps every transition of this change up to its merge: git log --notes=mesh-plan.

**Delivery** `novox/mesh-delivery@70cd624127bf` — **merged** since 2026-11-07T11:42:24Z **Source** — source.main moved to novox/mesh-delivery@b719dc0ecf27. Whether it reached the machines is `mesh-delivery-view`'s, derived from `source.main`, the wants and the haves. **Transitions** - 2026-11-07 11:38 (new) → proposed (announced): novox/mesh-delivery#163's head announced - 2026-11-07 11:38 proposed → ready (accepted): a verdict names this commit and the trunk's head as its trunk commit, the gate passed or warned, and the repository's own check did not fail; or the check gates no merge (check-gates-merge false) - 2026-11-07 11:42 ready → merged (merged): on the trunk its modules follow: merged there The commit's note under `refs/notes/mesh-plan` keeps every transition of this change up to its merge: `git log --notes=mesh-plan`.

jschoubben added 1 commit 2026-11-07 11:38:43 +00:00

Ask the check again: its composition ran before the indexer's build b5163570 was done, which is now (issue 1585; the checker's skipped build wait is noted on issue 1921)

mesh/delivery merged: source.main moved to novox/mesh-delivery@b719dc0ecf27

Details

mesh/check being carried to 70cd624127bf on b719dc0ecf27 | from bab05c66b3cb

70cd624127

mesh-tools commented

2026-11-07 11:38:49 +00:00

mesh-delivery: the operator's Approve of this pull request's merge into main is ended: its head 0520c12619 is no longer the pull request's head, or the pull request closed. Nothing is merged on it. The operator is asked again once the checks pass at the new head.

mesh-delivery: the operator's Approve of this pull request's merge into main is ended: its head 0520c126195b is no longer the pull request's head, or the pull request closed. Nothing is merged on it. The operator is asked again once the checks pass at the new head.

mesh-tools merged commit b719dc0ecf into main

2026-11-07 11:42:23 +00:00

mesh-tools deleted branch fix/1585-waits-end

2026-11-07 11:42:23 +00:00

jschoubben referenced this issue from a commit

2026-11-07 11:52:43 +00:00

A check composes only after the candidate's builds it waits on have ended (issue 1921)

mesh-tools referenced this pull request

2026-11-07 11:52:53 +00:00

A pull request's check composes only after the change's own builds end (issue 1921) #164

jschoubben referenced this issue from a commit

2026-11-07 13:39:05 +00:00