Fixes novox/hq issue 578: a delivery group's order was not held for mesh-controller #271 and mesh-catalog #251 on 2026-10-12 at 18:11.
Cause
The merge tool deletes the source branch after a merge, which is its default. After that, the forge reads the pull request's head.ref as refs/pull/<n>/head. Only head.label still names the branch. GiteaClient.mapPull took head.ref, so pull.merged named refs/pull/251/head as the head. A delivery group is the pull requests that share a head branch name (ADR 0249). The controller finds a merge's group by that head, and it joins a group's merges in one batch by that head too (ADR 0276 decision 5, ADR 0306 decision 2). Asked for the group refs/pull/251/head, mesh-delivery had none, and the controller logged "no member of a group is still to merge: it is cut at once" at 18:11:51. #251's walk (plan-1791821511497826695) went first. #271's merge was heard two seconds later, again under a pull ref, and was cut as a batch of its own, queued behind #251's walk.
Both walks were on the controller's own path, because #251 moves mesh-delivery. So mesh-delivery's word held neither walk. Its go came from the walk starting, not from a turn it gave: let_go is empty on both deliveries. mesh-delivery's turn logic was not at fault. The group's order for own-path members is held by the controller's batching, and that batching never saw the group.
A group merged through the tool has carried a pull ref as its head since the tool deleted branches by default. So the group wait (ADR 0306 decision 2) and the group-order edges inside a walk (ADR 0276 decision 5) did not hold for any such group.
Change
client.ts: the new headBranchOf names head.ref while the branch exists. When head.ref is a pull ref, it names head.label. When the forge gives no label, it keeps the ref and invents nothing. mapPull uses it, so pull.merged, pull.closed, pull.updated and the tools all say the branch.
Nothing else changes. With the right head, the 18:11 sequence goes as designed. #251 is heard while #271 is still ready, so it waits for the window. #271 joins #251's batch by head. The one walk puts mesh-controller and build-agent in a tier before gitea and mesh-delivery.
Test (fails on main)
test/announcing.test.ts: "a merged pull request whose branch was deleted is announced with its branch, not its pull ref". A fake forge answers #251 as the real one did after its merge (label is the branch, ref is refs/pull/251/head). On main the announcer's read gives refs/pull/251/head. With this change it gives feat/341-a-merge-check-judges-the-merge. The test also covers the cases where the branch exists, only a label is given, or only a pull ref is given.
Run locally: merge-check.sh (the Go part: checks and gitea's Go tests) passes. They ran with Node's type stripping and a resolver that maps .js to .ts: 41 of gitea's 44 TypeScript tests pass, the new one included. The other 3 test files import dist/ or the SDK, which are not on this machine. Not run locally:tsc. No TypeScript compiler is installed here, so the mesh/repo-check TypeScript part is the type check and the full test run.
Rollout
One module: gitea on novox. It does not depend on a controller or mesh-delivery change, and it can go in any order. A merge announced before it rolls still carries a pull ref.
Self-review
Can a group wait for ever? No new wait is added. A group's merges wait at most merge-window-at-most (10 minutes) for the window, as before (ADR 0276 decision 9). A failed or stopped controller member is handled as before.
Is it safe across a restart? Yes. The head is in the announced event the controller keeps with each batched merge in its store, and mesh-delivery keeps it as the delivery's branch.
Is ADR 0299's stop and walk-again behaviour unchanged? Yes. No stop, fold or walk-again code is touched.
Left as it was, and said here: when a group's later member merges more than merge-window-at-most before an earlier one, it still walks first. The window is the bound ADR 0306 chose.
Fixes novox/hq issue 578: a delivery group's order was not held for mesh-controller #271 and mesh-catalog #251 on 2026-10-12 at 18:11.
## Cause
The merge tool deletes the source branch after a merge, which is its default. After that, the forge reads the pull request's `head.ref` as `refs/pull/<n>/head`. Only `head.label` still names the branch. `GiteaClient.mapPull` took `head.ref`, so `pull.merged` named `refs/pull/251/head` as the head. A delivery group is the pull requests that share a head branch name (ADR 0249). The controller finds a merge's group by that head, and it joins a group's merges in one batch by that head too (ADR 0276 decision 5, ADR 0306 decision 2). Asked for the group `refs/pull/251/head`, mesh-delivery had none, and the controller logged "no member of a group is still to merge: it is cut at once" at 18:11:51. #251's walk (`plan-1791821511497826695`) went first. #271's merge was heard two seconds later, again under a pull ref, and was cut as a batch of its own, queued behind #251's walk.
Both walks were on the controller's own path, because #251 moves mesh-delivery. So mesh-delivery's word held neither walk. Its `go` came from the walk starting, not from a turn it gave: `let_go` is empty on both deliveries. mesh-delivery's turn logic was not at fault. The group's order for own-path members is held by the controller's batching, and that batching never saw the group.
A group merged through the tool has carried a pull ref as its head since the tool deleted branches by default. So the group wait (ADR 0306 decision 2) and the group-order edges inside a walk (ADR 0276 decision 5) did not hold for any such group.
## Change
- `client.ts`: the new `headBranchOf` names `head.ref` while the branch exists. When `head.ref` is a pull ref, it names `head.label`. When the forge gives no label, it keeps the ref and invents nothing. `mapPull` uses it, so `pull.merged`, `pull.closed`, `pull.updated` and the tools all say the branch.
- Nothing else changes. With the right head, the 18:11 sequence goes as designed. #251 is heard while #271 is still `ready`, so it waits for the window. #271 joins #251's batch by head. The one walk puts mesh-controller and build-agent in a tier before gitea and mesh-delivery.
## Test (fails on main)
- `test/announcing.test.ts`: "a merged pull request whose branch was deleted is announced with its branch, not its pull ref". A fake forge answers #251 as the real one did after its merge (`label` is the branch, `ref` is `refs/pull/251/head`). On main the announcer's read gives `refs/pull/251/head`. With this change it gives `feat/341-a-merge-check-judges-the-merge`. The test also covers the cases where the branch exists, only a label is given, or only a pull ref is given.
- Run locally: `merge-check.sh` (the Go part: checks and gitea's Go tests) passes. They ran with Node's type stripping and a resolver that maps `.js` to `.ts`: 41 of gitea's 44 TypeScript tests pass, the new one included. The other 3 test files import `dist/` or the SDK, which are not on this machine. **Not run locally:** `tsc`. No TypeScript compiler is installed here, so the `mesh/repo-check` TypeScript part is the type check and the full test run.
## Rollout
One module: gitea on novox. It does not depend on a controller or mesh-delivery change, and it can go in any order. A merge announced before it rolls still carries a pull ref.
## Self-review
- **Can a group wait for ever?** No new wait is added. A group's merges wait at most `merge-window-at-most` (10 minutes) for the window, as before (ADR 0276 decision 9). A failed or stopped controller member is handled as before.
- **Is it safe across a restart?** Yes. The head is in the announced event the controller keeps with each batched merge in its store, and mesh-delivery keeps it as the delivery's branch.
- **Is ADR 0299's stop and walk-again behaviour unchanged?** Yes. No stop, fold or walk-again code is touched.
Left as it was, and said here: when a group's later member merges more than `merge-window-at-most` before an earlier one, it still walks first. The window is the bound ADR 0306 chose.
mesh/merge-gate head 01d2ff7a46bc merged into trunk 2fe5601397ae — pass: builds gitea → novox; no bus step; every machine composes with the change as it di…
mesh/repo-check head 01d2ff7a46bc merged into trunk 2fe5601397ae — pass: its merge-check.sh passed, each part in its toolchain: merge-check.sh (go), merge-…
The merge tool deletes the branch after merging; the forge then reads head.ref as refs/pull/<n>/head and only
head.label names the branch. pull.merged carried the pull ref, so the controller found no group for the merge,
cut mesh-catalog #251's walk at once and walked it before mesh-controller #271's, against the group's order.
Deliverynovox/mesh-catalog@01d2ff7a46bc — delivered since 2026-10-12T19:09:32Z
Delivery plan — builds gitea → novox; no bus step
tier 0: gitea
novox: receives gitea
gitea provides npm-package-registry, git: its consumers are sent again after it
gitea keeps data (ADR 0233): what it holds is backed up before it moves
Machines
gitea on novox: passed — healthy 1 times over 25s
gitea on the rest running it: sent
Transitions
2026-10-12 18:57 (new) → proposed (announced): novox/mesh-catalog#255's head announced
2026-10-12 19:06 proposed → checked (checked): a verdict names this commit and the trunk's head as its trunk commit
2026-10-12 19:06 checked → ready (accepted): the gate passed or warned, and the repository's own check did not fail
2026-10-12 19:07 ready → published (merged): on the trunk its modules follow: merged there, and its walk opened — or nothing for a walk to move
2026-10-12 19:07 published → delivering (go): its walk started: let go by this owner in its group's order, by a person, or on the controller's own path — not a failed walk that carried a later commit for it, after which it waits for a walk of its own commit
2026-10-12 19:09 delivering → delivered (done): its walk is done, carrying its own commit: every machine of its deploy plan passed, or was left as its policy says
The commit's note under refs/notes/mesh-plan keeps every transition: git log --notes=mesh-plan.
<!-- mesh-delivery:view -->
**Delivery** `novox/mesh-catalog@01d2ff7a46bc` — **delivered** since 2026-10-12T19:09:32Z
**Delivery plan** — builds gitea → novox; no bus step
- tier 0: gitea
- novox: receives gitea
- gitea provides npm-package-registry, git: its consumers are sent again after it
- gitea keeps data (ADR 0233): what it holds is backed up before it moves
**Machines**
- gitea on novox: passed — healthy 1 times over 25s
- gitea on the rest running it: sent
**Transitions**
- 2026-10-12 18:57 (new) → proposed (announced): novox/mesh-catalog#255's head announced
- 2026-10-12 19:06 proposed → checked (checked): a verdict names this commit and the trunk's head as its trunk commit
- 2026-10-12 19:06 checked → ready (accepted): the gate passed or warned, and the repository's own check did not fail
- 2026-10-12 19:07 ready → published (merged): on the trunk its modules follow: merged there, and its walk opened — or nothing for a walk to move
- 2026-10-12 19:07 published → delivering (go): its walk started: let go by this owner in its group's order, by a person, or on the controller's own path — not a failed walk that carried a later commit for it, after which it waits for a walk of its own commit
- 2026-10-12 19:09 delivering → delivered (done): its walk is done, carrying its own commit: every machine of its deploy plan passed, or was left as its policy says
The commit's note under `refs/notes/mesh-plan` keeps every transition: `git log --notes=mesh-plan`.
Merge check at 01d2ff7a
Judged: head 01d2ff7a46bc merged into trunk 22071dabfc7b; read beside it: mesh-controller 84deb530ee5a, mesh-controller-main 84deb530ee5a, mesh-host 303d4037e84e, mesh-tools 954e1a53eef9
mesh/merge-gate: PASS — every machine composes with the change as it did without (4 of 4 compose) (modules: gitea)
mesh/repo-check: PASS — its merge-check.sh passed, each part in its toolchain: merge-check.sh (go), merge-check-typescript.sh (typescript)
Delivery plan — builds gitea → novox; no bus step
tier 0: gitea
novox: receives gitea
gitea provides npm-package-registry, git: its consumers are sent again after it
gitea keeps data (ADR 0233): what it holds is backed up before it moves
Run by the build seat on novox as build-1791831469096804104 (builds --log build-1791831469096804104).
--- the gate: gitea
gitea: ok, keeps data (valuable); WARNING: server stay(s) up and say(s) not how it is ready — judged by liveness alone, refused from 2026-11-18 (ADR 0240 rule 8)
long-running resources without health: 1
files asking for a setting without saying whether it is trusted: 0
1 manifest(s) checked. Judged against the seats this binary carries; a claim on one of the mesh's own seats is judged fully at registration, and a seat declared by a module not given here reads as unknown. Identities judged on a 6-character machine name, against the bounds of the providers given here
judging 1 module(s) changed against 4 machine(s), as the snapshot of 2026-10-12T18:54:57Z says they run
composing every machine as the mesh is
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
composing every machine with the change
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
merge gate: PASS — every machine composes with the change as it did without (4 of 4 compose)
judged by controller development build against the facts of 2026-10-12T18:54:57Z
modules the change moves:
- gitea (changed)
machines:
- a machine (f35): composes
- holds mesh-dns-resolver (idt): composes
- holds git, holds mesh-artifact-store, holds mesh-broker, holds mesh-catalog, holds mesh-delivery, holds mesh-dns-resolver, holds mesh-store, holds mesh-vault, holds npm-package-registry, the control node, the hub (ygcyr): composes
- a machine (yvbpsm): composes
a merge would rebuild 1 module(s) in 1 tier(s): gitea
--- its merge-check.sh (go toolchain)
ok git.novox.be/novox/mesh-catalog/checks/words 0.007s
words: 12 retired words and 70 daemons' lines, none in the 44435 strings of 701 files
ok git.novox.be/novox/mesh-catalog/checks/agenthome 0.013s
agent-home: no archive and no file written into under the agent's home
ok git.novox.be/novox/mesh-catalog/checks/paths 0.006s
paths: no module names another's state on its machine (3 known, listed to go)
long-running resources without health: 69, as kept
ok git.novox.be/novox/mesh-catalog/checks/deps 0.307s
testing modules/gitea
ok gitea/cmd/forge 1.579s
ok gitea/cmd/npm-registry 1.938s
? gitea/internal/forgehttp [no test files]
--- its merge-check-typescript.sh (typescript toolchain)
type-checking modules/gitea (TypeScript)
testing modules/gitea (TypeScript)
✔ a pull request opened after the margin is looked at, by its own update time (27.043939ms)
✔ every open pull request is read with what its base requires and what its head has (gitea_open_pulls) (104.597361ms)
✔ a pull request or issue number that is not one is refused before the forge is asked (5.492478ms)
✔ a merged pull request whose branch was deleted is announced with its branch, not its pull ref (34.825531ms)
✔ the newest status of each context is kept, with what it said, when and by whom (25.764477ms)
✔ a commit merges only when every required status is a success; a warning blocks (0.675181ms)
✔ no rule requires nothing; a rule that cannot be read says nothing, never yes (0.27536ms)
✔ a commit is contained in another when the compare finds nothing on it the other lacks (0.868582ms)
✔ a note line is appended once, as the forge's user, in its own repository (14.625581ms)
✔ a note that cannot be written is said, never read as written (0.535221ms)
✔ one view per pull request, found by its marker; only the mesh's statuses (0.762651ms)
✔ a build machine taking the check is said on the head: building on which machine, since the local time (34.49895ms)
✔ the verdict heard while a 'building' write is in flight lands last and stays (13.778091ms)
[gitea] novox/hq@a6b08de8: the verdict of build-1791680941000000000 is an older run's than the check now pending; not written
✔ the verdict written while a start's read of the forge is in flight stays the newest status (1.100371ms)
✔ the verdict heard while a 'building' write still waits its turn: the 'building' is never written (6.095229ms)
✔ a start heard after the verdict writes nothing (0.924532ms)
✔ a start from before the check now pending (a run before a recheck, the stream's history) writes nothing (0.605371ms)
✔ an old run's verdict redelivered after a recheck's pending is not written over it (1.010321ms)
[gitea] novox/hq@a6b08de8: the verdict of build-1791680941000000000 is an older run's than the check the forge holds pending; not written
[gitea] novox/hq@a6b08de8: the verdict of build-1791680941000000000 is an older run's than the check now pending; not written
✔ an old run's verdict in flight when the recheck comes lands before the recheck's pending (12.589439ms)
✔ a rechecked head says where its new run is: building on which machine (1.133642ms)
✔ the recheck's own verdict is written, even asked a little before the pending by another machine's clock (0.803091ms)
✔ after a restart, an old run's verdict is weighed against the check the forge's history holds open (2.736784ms)
✔ a recheck the forge does not take fails its caller, and the head is as it was (0.719291ms)
✔ the status tool writes the merge check's pending through the writer, and the mesh's other statuses as they were (17.995946ms)
✔ when a check was asked is read from its id; the opening of the check now pending from the gate's history (0.344321ms)
✔ a head this process did not announce (it restarted) is said only while the forge holds its gate pending (2.685984ms)
✔ a repository as the mesh clones it is told from another (0.426341ms)
✔ gitea_api refuses the forge's administration, a branch's protection and the merge check's status, and sends nothing (7.26796ms)
✔ gitea_api refuses a file written to the default branch or one a rule guards, and writes one on a branch of its own (79.497635ms)
✔ a protection rule covers a branch as the forge decides it: classes, alternatives, a plain name in any case (issue 539) (61.824369ms)
✔ gitea_api still writes what an author writes (13.298799ms)
✔ gitea_branch_protection_set only strengthens (18.359017ms)
✔ gitea_merge_pull_request merges only a head whose required statuses all succeeded, and only that head (44.980185ms)
✔ gitea_merge_pull_request refuses a verdict judged against an earlier trunk, and asks the check once (250.347322ms)
✔ gitea_merge_pull_request merges only the head it is named, and refuses one that is not the head (16.067273ms)
✔ gitea_merge_pull_request is loud when the forge merged a head other than the one it was handed (14.268941ms)
✔ gitea_merge_pull_request says the merge happened when reading back what landed fails (27.105499ms)
✔ gitea_merge_pull_request reads back a fast-forward, and says it cannot read back a squash or a rebase (56.296901ms)
✔ gitea_delete_repo deletes only the tools' own user's repository, unguarded, untouched for 30 days, named in full (19.836579ms)
✔ gitea_api sends the path it judged, and refuses moving a pull request's base (4.816187ms)
✔ gitea_merge_pull_request merges only into a branch whose protection requires the checks (7.08623ms)
✔ a rule named for a branch is not made weaker than the glob rule that guards it now (12.928169ms)
✔ a repository owner or name is never another path (0.21918ms)
✔ the judges, without a forge (0.367731ms)
✔ gitea_api never acts as another user: sudo, in any case, and any query key it does not send are refused (1.651342ms)
✔ gitea_api never approves a pull request nor requests changes: those are the operator's (2.719864ms)
✔ no tool's caller text becomes another path or another user: `..` and `?sudo=` are refused before the forge (5.486058ms)
✔ every page of a pull request's files is read, though the forge caps a page at fifty (58.082234ms)
✔ a pass asks only the repositories that moved since the last look, every one before the first (0.442511ms)
✔ the directories above a merge's files that hold a module at the commit are said, the root never (22.458123ms)
✔ a lookup the forge refuses is thrown, not read as no module (8.984983ms)
✔ a renamed file is said under its old name too, as gone (novox/hq ADR 0267) (12.860578ms)
✔ the statuses asked for replace the rule's, and an empty list requires none (7.61334ms)
✔ a new rule refuses direct pushes unless asked; an administrator's override only when said (0.417361ms)
✔ each open pull request's new head is announced once, and nothing closed or merged (19.958379ms)
✔ a verdict is the head commit's status; an error is the forge's error, never a success (1.004352ms)
✔ each layer is its own status: the gate with the modules it judged, the repository's own check beside it (0.696931ms)
✔ no verdict sets warning on a required check; only a person's decision fails (1.021532ms)
✔ a commit status is set on the commit, under the merge check's context (85.488463ms)
✔ a change plan is the gate's result: said on the status and, when it builds something, as a comment (3.651526ms)
✔ a head announced is pending in the build queue, since the local time it was announced (0.40702ms)
✔ first start: mints with the admin account, keeps the token at 0600, asks for the tools' scopes only (63.537732ms)
✔ second start: reuses the kept token, mints nothing (66.086255ms)
✔ the forge rejects the kept token (its data was restored): minted afresh, once, and the call goes through (121.928837ms)
✔ a token kept with write:admin by an earlier build is minted again without it at the next start, and the forge drops the old one (181.551862ms)
✔ a kept token lacking a scope the tools need: the forge refuses the call for the scope, the token is re-minted with the whole list, and the call goes through (18.827848ms)
✔ the kept file is gone but the forge still holds a token by that name: replaced, not refused (17.808075ms)
✔ concurrent first calls share one mint (11.330696ms)
✔ the admin account is refused: said plainly, nothing kept, and the next call fails the same way rather than crashing (11.859077ms)
✔ one process shares one source per kept file — the watcher and the tools never renew against each other (1.007552ms)
✔ a second process finds the token the first renewed, and reuses it instead of minting over it (24.644886ms)
✔ a configured token wins, and is reported rather than minted over when the forge rejects it (2.282643ms)
✔ nothing to mint with and no token: the client says what is missing (0.16301ms)
[gitea] no tools — no Gitea token, and nothing to mint one with — set MESH_GITEA_TOKEN, or MESH_GITEA_ADMIN_USER, MESH_GITEA_ADMIN_PASSWORD_FILE, MESH_GITEA_STATE_DIR
[gitea] no token kept — minting one
[gitea] the forge already holds a token named "mesh-tools" — replacing it
[gitea] minted a token for "mesh-admin" (write:repository, write:issue, read:user), kept at /tmp/gitea-LYgiwj/state/token
✔ the tools register once there is a way to a token, and the first call mints it (8.354732ms)
[gitea] no token kept — minting one
[gitea] the forge already holds a token named "mesh-tools" — replacing it
[gitea] minted a token for "mesh-admin" (write:repository, write:issue, read:user), kept at /tmp/gitea-7w0OVZ/state/token
✔ a pull request can be closed, read and cleaned up through the tools (20.34239ms)
[gitea] novox/mesh-host's main moved from a2c2c35bbbbb to d28ca95ccccc: 25 open pull request(s) into it put out of date, nothing run (novox/hq ADR 0323)
[gitea] novox/mesh-host's main moved from d28ca95ccccc to ffffffffffff: 0 open pull request(s) into it put out of date, nothing run (novox/hq ADR 0323)
[gitea] novox/mesh-host: whether a trunk moved could not be weighed, so it is weighed again on the next pass — the forge is away
[gitea] novox/mesh-host's main moved from a2c2c35bbbbb to d28ca95ccccc: 0 open pull request(s) into it put out of date, nothing run (novox/hq ADR 0323)
✔ both statuses say the head and the trunk commit the verdict judged, first, whatever the summary's length (1.438822ms)
✔ a trunk's move puts an open head's merge check out of date, naming both commits, unless a check is on its way (0.879421ms)
✔ a push to a trunk puts every open pull request into it out of date — twenty-five of them — and none into another (4.652857ms)
✔ a trunk that cannot be read is weighed again on the next pass, never recorded half (0.434001ms)
✔ the trunk's move takes its turn after a verdict in flight, and weighs what the forge holds then (0.524851ms)
✔ a verdict judged against a trunk newer than the pass's first read is not put out of date (0.29596ms)
✔ a verdict that landed out of date keeps its account on the pull request (0.295391ms)
ℹ tests 82
ℹ suites 0
ℹ pass 82
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 925.799669
Out of date when it landed: out of date: judged against trunk 22071dabfc, the trunk is now 2fe5601397ae; judged again on a recheck or when it is merged
**Merge check** at `01d2ff7a`
Judged: head `01d2ff7a46bc` merged into trunk `22071dabfc7b`; read beside it: mesh-controller `84deb530ee5a`, mesh-controller-main `84deb530ee5a`, mesh-host `303d4037e84e`, mesh-tools `954e1a53eef9`
- `mesh/merge-gate`: **PASS** — every machine composes with the change as it did without (4 of 4 compose) (modules: gitea)
- `mesh/repo-check`: **PASS** — its merge-check.sh passed, each part in its toolchain: merge-check.sh (go), merge-check-typescript.sh (typescript)
**Delivery plan** — builds gitea → novox; no bus step
- tier 0: gitea
- novox: receives gitea
- gitea provides npm-package-registry, git: its consumers are sent again after it
- gitea keeps data (ADR 0233): what it holds is backed up before it moves
Run by the build seat on novox as `build-1791831469096804104` (`builds --log build-1791831469096804104`).
```
--- the gate: gitea
gitea: ok, keeps data (valuable); WARNING: server stay(s) up and say(s) not how it is ready — judged by liveness alone, refused from 2026-11-18 (ADR 0240 rule 8)
long-running resources without health: 1
files asking for a setting without saying whether it is trusted: 0
1 manifest(s) checked. Judged against the seats this binary carries; a claim on one of the mesh's own seats is judged fully at registration, and a seat declared by a module not given here reads as unknown. Identities judged on a 6-character machine name, against the bounds of the providers given here
judging 1 module(s) changed against 4 machine(s), as the snapshot of 2026-10-12T18:54:57Z says they run
composing every machine as the mesh is
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
composing every machine with the change
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
merge gate: PASS — every machine composes with the change as it did without (4 of 4 compose)
judged by controller development build against the facts of 2026-10-12T18:54:57Z
modules the change moves:
- gitea (changed)
machines:
- a machine (f35): composes
- holds mesh-dns-resolver (idt): composes
- holds git, holds mesh-artifact-store, holds mesh-broker, holds mesh-catalog, holds mesh-delivery, holds mesh-dns-resolver, holds mesh-store, holds mesh-vault, holds npm-package-registry, the control node, the hub (ygcyr): composes
- a machine (yvbpsm): composes
a merge would rebuild 1 module(s) in 1 tier(s): gitea
--- its merge-check.sh (go toolchain)
ok git.novox.be/novox/mesh-catalog/checks/words 0.007s
words: 12 retired words and 70 daemons' lines, none in the 44435 strings of 701 files
ok git.novox.be/novox/mesh-catalog/checks/agenthome 0.013s
agent-home: no archive and no file written into under the agent's home
ok git.novox.be/novox/mesh-catalog/checks/paths 0.006s
paths: no module names another's state on its machine (3 known, listed to go)
long-running resources without health: 69, as kept
ok git.novox.be/novox/mesh-catalog/checks/deps 0.307s
testing modules/gitea
ok gitea/cmd/forge 1.579s
ok gitea/cmd/npm-registry 1.938s
? gitea/internal/forgehttp [no test files]
--- its merge-check-typescript.sh (typescript toolchain)
type-checking modules/gitea (TypeScript)
testing modules/gitea (TypeScript)
✔ a pull request opened after the margin is looked at, by its own update time (27.043939ms)
✔ every open pull request is read with what its base requires and what its head has (gitea_open_pulls) (104.597361ms)
✔ a pull request or issue number that is not one is refused before the forge is asked (5.492478ms)
✔ a merged pull request whose branch was deleted is announced with its branch, not its pull ref (34.825531ms)
✔ the newest status of each context is kept, with what it said, when and by whom (25.764477ms)
✔ a commit merges only when every required status is a success; a warning blocks (0.675181ms)
✔ no rule requires nothing; a rule that cannot be read says nothing, never yes (0.27536ms)
✔ a commit is contained in another when the compare finds nothing on it the other lacks (0.868582ms)
✔ a note line is appended once, as the forge's user, in its own repository (14.625581ms)
✔ a note that cannot be written is said, never read as written (0.535221ms)
✔ one view per pull request, found by its marker; only the mesh's statuses (0.762651ms)
✔ a build machine taking the check is said on the head: building on which machine, since the local time (34.49895ms)
✔ the verdict heard while a 'building' write is in flight lands last and stays (13.778091ms)
[gitea] novox/hq@a6b08de8: the verdict of build-1791680941000000000 is an older run's than the check now pending; not written
✔ the verdict written while a start's read of the forge is in flight stays the newest status (1.100371ms)
✔ the verdict heard while a 'building' write still waits its turn: the 'building' is never written (6.095229ms)
✔ a start heard after the verdict writes nothing (0.924532ms)
✔ a start from before the check now pending (a run before a recheck, the stream's history) writes nothing (0.605371ms)
✔ an old run's verdict redelivered after a recheck's pending is not written over it (1.010321ms)
[gitea] novox/hq@a6b08de8: the verdict of build-1791680941000000000 is an older run's than the check the forge holds pending; not written
[gitea] novox/hq@a6b08de8: the verdict of build-1791680941000000000 is an older run's than the check now pending; not written
✔ an old run's verdict in flight when the recheck comes lands before the recheck's pending (12.589439ms)
✔ a rechecked head says where its new run is: building on which machine (1.133642ms)
✔ the recheck's own verdict is written, even asked a little before the pending by another machine's clock (0.803091ms)
✔ after a restart, an old run's verdict is weighed against the check the forge's history holds open (2.736784ms)
✔ a recheck the forge does not take fails its caller, and the head is as it was (0.719291ms)
✔ the status tool writes the merge check's pending through the writer, and the mesh's other statuses as they were (17.995946ms)
✔ when a check was asked is read from its id; the opening of the check now pending from the gate's history (0.344321ms)
✔ a head this process did not announce (it restarted) is said only while the forge holds its gate pending (2.685984ms)
✔ a repository as the mesh clones it is told from another (0.426341ms)
✔ gitea_api refuses the forge's administration, a branch's protection and the merge check's status, and sends nothing (7.26796ms)
✔ gitea_api refuses a file written to the default branch or one a rule guards, and writes one on a branch of its own (79.497635ms)
✔ a protection rule covers a branch as the forge decides it: classes, alternatives, a plain name in any case (issue 539) (61.824369ms)
✔ gitea_api still writes what an author writes (13.298799ms)
✔ gitea_branch_protection_set only strengthens (18.359017ms)
✔ gitea_merge_pull_request merges only a head whose required statuses all succeeded, and only that head (44.980185ms)
✔ gitea_merge_pull_request refuses a verdict judged against an earlier trunk, and asks the check once (250.347322ms)
✔ gitea_merge_pull_request merges only the head it is named, and refuses one that is not the head (16.067273ms)
✔ gitea_merge_pull_request is loud when the forge merged a head other than the one it was handed (14.268941ms)
✔ gitea_merge_pull_request says the merge happened when reading back what landed fails (27.105499ms)
✔ gitea_merge_pull_request reads back a fast-forward, and says it cannot read back a squash or a rebase (56.296901ms)
✔ gitea_delete_repo deletes only the tools' own user's repository, unguarded, untouched for 30 days, named in full (19.836579ms)
✔ gitea_api sends the path it judged, and refuses moving a pull request's base (4.816187ms)
✔ gitea_merge_pull_request merges only into a branch whose protection requires the checks (7.08623ms)
✔ a rule named for a branch is not made weaker than the glob rule that guards it now (12.928169ms)
✔ a repository owner or name is never another path (0.21918ms)
✔ the judges, without a forge (0.367731ms)
✔ gitea_api never acts as another user: sudo, in any case, and any query key it does not send are refused (1.651342ms)
✔ gitea_api never approves a pull request nor requests changes: those are the operator's (2.719864ms)
✔ no tool's caller text becomes another path or another user: `..` and `?sudo=` are refused before the forge (5.486058ms)
✔ every page of a pull request's files is read, though the forge caps a page at fifty (58.082234ms)
✔ a pass asks only the repositories that moved since the last look, every one before the first (0.442511ms)
✔ the directories above a merge's files that hold a module at the commit are said, the root never (22.458123ms)
✔ a lookup the forge refuses is thrown, not read as no module (8.984983ms)
✔ a renamed file is said under its old name too, as gone (novox/hq ADR 0267) (12.860578ms)
✔ the statuses asked for replace the rule's, and an empty list requires none (7.61334ms)
✔ a new rule refuses direct pushes unless asked; an administrator's override only when said (0.417361ms)
✔ each open pull request's new head is announced once, and nothing closed or merged (19.958379ms)
✔ a verdict is the head commit's status; an error is the forge's error, never a success (1.004352ms)
✔ each layer is its own status: the gate with the modules it judged, the repository's own check beside it (0.696931ms)
✔ no verdict sets warning on a required check; only a person's decision fails (1.021532ms)
✔ a commit status is set on the commit, under the merge check's context (85.488463ms)
✔ a change plan is the gate's result: said on the status and, when it builds something, as a comment (3.651526ms)
✔ a head announced is pending in the build queue, since the local time it was announced (0.40702ms)
✔ first start: mints with the admin account, keeps the token at 0600, asks for the tools' scopes only (63.537732ms)
✔ second start: reuses the kept token, mints nothing (66.086255ms)
✔ the forge rejects the kept token (its data was restored): minted afresh, once, and the call goes through (121.928837ms)
✔ a token kept with write:admin by an earlier build is minted again without it at the next start, and the forge drops the old one (181.551862ms)
✔ a kept token lacking a scope the tools need: the forge refuses the call for the scope, the token is re-minted with the whole list, and the call goes through (18.827848ms)
✔ the kept file is gone but the forge still holds a token by that name: replaced, not refused (17.808075ms)
✔ concurrent first calls share one mint (11.330696ms)
✔ the admin account is refused: said plainly, nothing kept, and the next call fails the same way rather than crashing (11.859077ms)
✔ one process shares one source per kept file — the watcher and the tools never renew against each other (1.007552ms)
✔ a second process finds the token the first renewed, and reuses it instead of minting over it (24.644886ms)
✔ a configured token wins, and is reported rather than minted over when the forge rejects it (2.282643ms)
✔ nothing to mint with and no token: the client says what is missing (0.16301ms)
[gitea] no tools — no Gitea token, and nothing to mint one with — set MESH_GITEA_TOKEN, or MESH_GITEA_ADMIN_USER, MESH_GITEA_ADMIN_PASSWORD_FILE, MESH_GITEA_STATE_DIR
[gitea] no token kept — minting one
[gitea] the forge already holds a token named "mesh-tools" — replacing it
[gitea] minted a token for "mesh-admin" (write:repository, write:issue, read:user), kept at /tmp/gitea-LYgiwj/state/token
✔ the tools register once there is a way to a token, and the first call mints it (8.354732ms)
[gitea] no token kept — minting one
[gitea] the forge already holds a token named "mesh-tools" — replacing it
[gitea] minted a token for "mesh-admin" (write:repository, write:issue, read:user), kept at /tmp/gitea-7w0OVZ/state/token
✔ a pull request can be closed, read and cleaned up through the tools (20.34239ms)
[gitea] novox/mesh-host's main moved from a2c2c35bbbbb to d28ca95ccccc: 25 open pull request(s) into it put out of date, nothing run (novox/hq ADR 0323)
[gitea] novox/mesh-host's main moved from d28ca95ccccc to ffffffffffff: 0 open pull request(s) into it put out of date, nothing run (novox/hq ADR 0323)
[gitea] novox/mesh-host: whether a trunk moved could not be weighed, so it is weighed again on the next pass — the forge is away
[gitea] novox/mesh-host's main moved from a2c2c35bbbbb to d28ca95ccccc: 0 open pull request(s) into it put out of date, nothing run (novox/hq ADR 0323)
✔ both statuses say the head and the trunk commit the verdict judged, first, whatever the summary's length (1.438822ms)
✔ a trunk's move puts an open head's merge check out of date, naming both commits, unless a check is on its way (0.879421ms)
✔ a push to a trunk puts every open pull request into it out of date — twenty-five of them — and none into another (4.652857ms)
✔ a trunk that cannot be read is weighed again on the next pass, never recorded half (0.434001ms)
✔ the trunk's move takes its turn after a verdict in flight, and weighs what the forge holds then (0.524851ms)
✔ a verdict judged against a trunk newer than the pass's first read is not put out of date (0.29596ms)
✔ a verdict that landed out of date keeps its account on the pull request (0.295391ms)
ℹ tests 82
ℹ suites 0
ℹ pass 82
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 925.799669
```
**Out of date when it landed**: out of date: judged against trunk 22071dabfc7b, the trunk is now 2fe5601397ae; judged again on a recheck or when it is merged
Merge check at 01d2ff7a
Judged: head 01d2ff7a46bc merged into trunk 2fe5601397ae; read beside it: mesh-controller 84deb530ee5a, mesh-controller-main 84deb530ee5a, mesh-host 303d4037e84e, mesh-tools 954e1a53eef9
mesh/merge-gate: PASS — every machine composes with the change as it did without (4 of 4 compose) (modules: gitea)
mesh/repo-check: PASS — its merge-check.sh passed, each part in its toolchain: merge-check.sh (go), merge-check-typescript.sh (typescript)
Delivery plan — builds gitea → novox; no bus step
tier 0: gitea
novox: receives gitea
gitea provides npm-package-registry, git: its consumers are sent again after it
gitea keeps data (ADR 0233): what it holds is backed up before it moves
Run by the build seat on shanks as build-1791831837030233002 (builds --log build-1791831837030233002).
--- the gate: gitea
gitea: ok, keeps data (valuable); WARNING: server stay(s) up and say(s) not how it is ready — judged by liveness alone, refused from 2026-11-18 (ADR 0240 rule 8)
long-running resources without health: 1
files asking for a setting without saying whether it is trusted: 0
1 manifest(s) checked. Judged against the seats this binary carries; a claim on one of the mesh's own seats is judged fully at registration, and a seat declared by a module not given here reads as unknown. Identities judged on a 6-character machine name, against the bounds of the providers given here
judging 1 module(s) changed against 4 machine(s), as the snapshot of 2026-10-12T19:03:57Z says they run
composing every machine as the mesh is
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
composing every machine with the change
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
merge gate: PASS — every machine composes with the change as it did without (4 of 4 compose)
judged by controller development build against the facts of 2026-10-12T19:03:57Z
modules the change moves:
- gitea (changed)
machines:
- a machine (f35): composes
- holds mesh-dns-resolver (idt): composes
- holds git, holds mesh-artifact-store, holds mesh-broker, holds mesh-catalog, holds mesh-delivery, holds mesh-dns-resolver, holds mesh-store, holds mesh-vault, holds npm-package-registry, the control node, the hub (ygcyr): composes
- a machine (yvbpsm): composes
a merge would rebuild 1 module(s) in 1 tier(s): gitea
--- its merge-check.sh (go toolchain)
ok git.novox.be/novox/mesh-catalog/checks/words 0.003s
words: 12 retired words and 70 daemons' lines, none in the 44437 strings of 701 files
ok git.novox.be/novox/mesh-catalog/checks/agenthome 0.012s
agent-home: no archive and no file written into under the agent's home
ok git.novox.be/novox/mesh-catalog/checks/paths 0.007s
paths: no module names another's state on its machine (3 known, listed to go)
long-running resources without health: 69, as kept
ok git.novox.be/novox/mesh-catalog/checks/deps 0.252s
testing modules/gitea
ok gitea/cmd/forge 1.506s
ok gitea/cmd/npm-registry 1.881s
? gitea/internal/forgehttp [no test files]
--- its merge-check-typescript.sh (typescript toolchain)
type-checking modules/gitea (TypeScript)
testing modules/gitea (TypeScript)
✔ a pull request opened after the margin is looked at, by its own update time (12.909796ms)
✔ every open pull request is read with what its base requires and what its head has (gitea_open_pulls) (62.152058ms)
✔ a pull request or issue number that is not one is refused before the forge is asked (1.770794ms)
✔ a merged pull request whose branch was deleted is announced with its branch, not its pull ref (11.288445ms)
✔ the newest status of each context is kept, with what it said, when and by whom (51.096717ms)
✔ a commit merges only when every required status is a success; a warning blocks (0.643978ms)
✔ no rule requires nothing; a rule that cannot be read says nothing, never yes (0.274444ms)
✔ a commit is contained in another when the compare finds nothing on it the other lacks (0.999414ms)
✔ a note line is appended once, as the forge's user, in its own repository (14.116302ms)
✔ a note that cannot be written is said, never read as written (0.583608ms)
✔ one view per pull request, found by its marker; only the mesh's statuses (0.792311ms)
✔ a build machine taking the check is said on the head: building on which machine, since the local time (25.497138ms)
✔ the verdict heard while a 'building' write is in flight lands last and stays (6.472599ms)
[gitea] novox/hq@a6b08de8: the verdict of build-1791680941000000000 is an older run's than the check now pending; not written
✔ the verdict written while a start's read of the forge is in flight stays the newest status (0.841851ms)
✔ the verdict heard while a 'building' write still waits its turn: the 'building' is never written (0.434796ms)
✔ a start heard after the verdict writes nothing (0.454676ms)
✔ a start from before the check now pending (a run before a recheck, the stream's history) writes nothing (0.455477ms)
✔ an old run's verdict redelivered after a recheck's pending is not written over it (0.778111ms)
[gitea] novox/hq@a6b08de8: the verdict of build-1791680941000000000 is an older run's than the check the forge holds pending; not written
[gitea] novox/hq@a6b08de8: the verdict of build-1791680941000000000 is an older run's than the check now pending; not written
✔ an old run's verdict in flight when the recheck comes lands before the recheck's pending (2.399113ms)
✔ a rechecked head says where its new run is: building on which machine (0.858881ms)
✔ the recheck's own verdict is written, even asked a little before the pending by another machine's clock (0.652379ms)
✔ after a restart, an old run's verdict is weighed against the check the forge's history holds open (0.469437ms)
✔ a recheck the forge does not take fails its caller, and the head is as it was (0.459666ms)
✔ the status tool writes the merge check's pending through the writer, and the mesh's other statuses as they were (4.711374ms)
✔ when a check was asked is read from its id; the opening of the check now pending from the gate's history (0.526317ms)
✔ a head this process did not announce (it restarted) is said only while the forge holds its gate pending (0.422536ms)
✔ a repository as the mesh clones it is told from another (0.292404ms)
✔ gitea_api refuses the forge's administration, a branch's protection and the merge check's status, and sends nothing (7.917288ms)
✔ gitea_api refuses a file written to the default branch or one a rule guards, and writes one on a branch of its own (53.007923ms)
✔ a protection rule covers a branch as the forge decides it: classes, alternatives, a plain name in any case (issue 539) (19.246233ms)
✔ gitea_api still writes what an author writes (9.403138ms)
✔ gitea_branch_protection_set only strengthens (16.999692ms)
✔ gitea_merge_pull_request merges only a head whose required statuses all succeeded, and only that head (47.765702ms)
✔ gitea_merge_pull_request refuses a verdict judged against an earlier trunk, and asks the check once (152.54454ms)
✔ gitea_merge_pull_request merges only the head it is named, and refuses one that is not the head (19.512686ms)
✔ gitea_merge_pull_request is loud when the forge merged a head other than the one it was handed (15.33385ms)
✔ gitea_merge_pull_request says the merge happened when reading back what landed fails (30.158292ms)
✔ gitea_merge_pull_request reads back a fast-forward, and says it cannot read back a squash or a rebase (61.59406ms)
✔ gitea_delete_repo deletes only the tools' own user's repository, unguarded, untouched for 30 days, named in full (23.943336ms)
✔ gitea_api sends the path it judged, and refuses moving a pull request's base (3.168803ms)
✔ gitea_merge_pull_request merges only into a branch whose protection requires the checks (6.912384ms)
✔ a rule named for a branch is not made weaker than the glob rule that guards it now (12.834035ms)
✔ a repository owner or name is never another path (0.253213ms)
✔ the judges, without a forge (0.378245ms)
✔ gitea_api never acts as another user: sudo, in any case, and any query key it does not send are refused (1.627962ms)
✔ gitea_api never approves a pull request nor requests changes: those are the operator's (2.291311ms)
✔ no tool's caller text becomes another path or another user: `..` and `?sudo=` are refused before the forge (4.975518ms)
✔ every page of a pull request's files is read, though the forge caps a page at fifty (32.747787ms)
✔ a pass asks only the repositories that moved since the last look, every one before the first (0.468717ms)
✔ the directories above a merge's files that hold a module at the commit are said, the root never (13.263851ms)
✔ a lookup the forge refuses is thrown, not read as no module (5.310082ms)
✔ a renamed file is said under its old name too, as gone (novox/hq ADR 0267) (6.739941ms)
✔ a new head beyond the newest 20 open pull requests is announced, exactly once (50.828653ms)
✔ a pass pages until it reaches back past its floor, and no further (26.925058ms)
✔ the statuses asked for replace the rule's, and an empty list requires none (6.731991ms)
✔ a new rule refuses direct pushes unless asked; an administrator's override only when said (0.340814ms)
✔ each open pull request's new head is announced once, and nothing closed or merged (9.307127ms)
✔ a verdict is the head commit's status; an error is the forge's error, never a success (0.854122ms)
✔ each layer is its own status: the gate with the modules it judged, the repository's own check beside it (0.633079ms)
✔ no verdict sets warning on a required check; only a person's decision fails (1.057794ms)
✔ a commit status is set on the commit, under the merge check's context (38.243101ms)
✔ a change plan is the gate's result: said on the status and, when it builds something, as a comment (0.614348ms)
✔ a head announced is pending in the build queue, since the local time it was announced (0.311054ms)
✔ first start: mints with the admin account, keeps the token at 0600, asks for the tools' scopes only (36.6383ms)
✔ second start: reuses the kept token, mints nothing (17.552069ms)
✔ the forge rejects the kept token (its data was restored): minted afresh, once, and the call goes through (20.50366ms)
✔ a token kept with write:admin by an earlier build is minted again without it at the next start, and the forge drops the old one (22.70724ms)
✔ a kept token lacking a scope the tools need: the forge refuses the call for the scope, the token is re-minted with the whole list, and the call goes through (18.516192ms)
✔ the kept file is gone but the forge still holds a token by that name: replaced, not refused (18.33732ms)
✔ concurrent first calls share one mint (10.286281ms)
✔ the admin account is refused: said plainly, nothing kept, and the next call fails the same way rather than crashing (12.45969ms)
✔ one process shares one source per kept file — the watcher and the tools never renew against each other (0.942933ms)
✔ a second process finds the token the first renewed, and reuses it instead of minting over it (19.535246ms)
✔ a configured token wins, and is reported rather than minted over when the forge rejects it (2.497274ms)
✔ nothing to mint with and no token: the client says what is missing (0.206203ms)
[gitea] no tools — no Gitea token, and nothing to mint one with — set MESH_GITEA_TOKEN, or MESH_GITEA_ADMIN_USER, MESH_GITEA_ADMIN_PASSWORD_FILE, MESH_GITEA_STATE_DIR
[gitea] no token kept — minting one
[gitea] the forge already holds a token named "mesh-tools" — replacing it
[gitea] minted a token for "mesh-admin" (write:repository, write:issue, read:user), kept at /tmp/gitea-PVngJn/state/token
✔ the tools register once there is a way to a token, and the first call mints it (8.531087ms)
[gitea] no token kept — minting one
[gitea] the forge already holds a token named "mesh-tools" — replacing it
[gitea] minted a token for "mesh-admin" (write:repository, write:issue, read:user), kept at /tmp/gitea-e6cdfF/state/token
✔ a pull request can be closed, read and cleaned up through the tools (20.234186ms)
[gitea] novox/mesh-host's main moved from a2c2c35bbbbb to d28ca95ccccc: 25 open pull request(s) into it put out of date, nothing run (novox/hq ADR 0323)
[gitea] novox/mesh-host's main moved from d28ca95ccccc to ffffffffffff: 0 open pull request(s) into it put out of date, nothing run (novox/hq ADR 0323)
[gitea] novox/mesh-host: whether a trunk moved could not be weighed, so it is weighed again on the next pass — the forge is away
[gitea] novox/mesh-host's main moved from a2c2c35bbbbb to d28ca95ccccc: 0 open pull request(s) into it put out of date, nothing run (novox/hq ADR 0323)
✔ both statuses say the head and the trunk commit the verdict judged, first, whatever the summary's length (2.246481ms)
✔ a trunk's move puts an open head's merge check out of date, naming both commits, unless a check is on its way (1.327118ms)
✔ a push to a trunk puts every open pull request into it out of date — twenty-five of them — and none into another (5.88108ms)
✔ a trunk that cannot be read is weighed again on the next pass, never recorded half (0.523547ms)
✔ the trunk's move takes its turn after a verdict in flight, and weighs what the forge holds then (0.614319ms)
✔ a verdict judged against a trunk newer than the pass's first read is not put out of date (0.356505ms)
✔ a verdict that landed out of date keeps its account on the pull request (0.312625ms)
ℹ tests 84
ℹ suites 0
ℹ pass 84
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 675.256821
**Merge check** at `01d2ff7a`
Judged: head `01d2ff7a46bc` merged into trunk `2fe5601397ae`; read beside it: mesh-controller `84deb530ee5a`, mesh-controller-main `84deb530ee5a`, mesh-host `303d4037e84e`, mesh-tools `954e1a53eef9`
- `mesh/merge-gate`: **PASS** — every machine composes with the change as it did without (4 of 4 compose) (modules: gitea)
- `mesh/repo-check`: **PASS** — its merge-check.sh passed, each part in its toolchain: merge-check.sh (go), merge-check-typescript.sh (typescript)
**Delivery plan** — builds gitea → novox; no bus step
- tier 0: gitea
- novox: receives gitea
- gitea provides npm-package-registry, git: its consumers are sent again after it
- gitea keeps data (ADR 0233): what it holds is backed up before it moves
Run by the build seat on shanks as `build-1791831837030233002` (`builds --log build-1791831837030233002`).
```
--- the gate: gitea
gitea: ok, keeps data (valuable); WARNING: server stay(s) up and say(s) not how it is ready — judged by liveness alone, refused from 2026-11-18 (ADR 0240 rule 8)
long-running resources without health: 1
files asking for a setting without saying whether it is trusted: 0
1 manifest(s) checked. Judged against the seats this binary carries; a claim on one of the mesh's own seats is judged fully at registration, and a seat declared by a module not given here reads as unknown. Identities judged on a 6-character machine name, against the bounds of the providers given here
judging 1 module(s) changed against 4 machine(s), as the snapshot of 2026-10-12T19:03:57Z says they run
composing every machine as the mesh is
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
composing every machine with the change
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
merge gate: PASS — every machine composes with the change as it did without (4 of 4 compose)
judged by controller development build against the facts of 2026-10-12T19:03:57Z
modules the change moves:
- gitea (changed)
machines:
- a machine (f35): composes
- holds mesh-dns-resolver (idt): composes
- holds git, holds mesh-artifact-store, holds mesh-broker, holds mesh-catalog, holds mesh-delivery, holds mesh-dns-resolver, holds mesh-store, holds mesh-vault, holds npm-package-registry, the control node, the hub (ygcyr): composes
- a machine (yvbpsm): composes
a merge would rebuild 1 module(s) in 1 tier(s): gitea
--- its merge-check.sh (go toolchain)
ok git.novox.be/novox/mesh-catalog/checks/words 0.003s
words: 12 retired words and 70 daemons' lines, none in the 44437 strings of 701 files
ok git.novox.be/novox/mesh-catalog/checks/agenthome 0.012s
agent-home: no archive and no file written into under the agent's home
ok git.novox.be/novox/mesh-catalog/checks/paths 0.007s
paths: no module names another's state on its machine (3 known, listed to go)
long-running resources without health: 69, as kept
ok git.novox.be/novox/mesh-catalog/checks/deps 0.252s
testing modules/gitea
ok gitea/cmd/forge 1.506s
ok gitea/cmd/npm-registry 1.881s
? gitea/internal/forgehttp [no test files]
--- its merge-check-typescript.sh (typescript toolchain)
type-checking modules/gitea (TypeScript)
testing modules/gitea (TypeScript)
✔ a pull request opened after the margin is looked at, by its own update time (12.909796ms)
✔ every open pull request is read with what its base requires and what its head has (gitea_open_pulls) (62.152058ms)
✔ a pull request or issue number that is not one is refused before the forge is asked (1.770794ms)
✔ a merged pull request whose branch was deleted is announced with its branch, not its pull ref (11.288445ms)
✔ the newest status of each context is kept, with what it said, when and by whom (51.096717ms)
✔ a commit merges only when every required status is a success; a warning blocks (0.643978ms)
✔ no rule requires nothing; a rule that cannot be read says nothing, never yes (0.274444ms)
✔ a commit is contained in another when the compare finds nothing on it the other lacks (0.999414ms)
✔ a note line is appended once, as the forge's user, in its own repository (14.116302ms)
✔ a note that cannot be written is said, never read as written (0.583608ms)
✔ one view per pull request, found by its marker; only the mesh's statuses (0.792311ms)
✔ a build machine taking the check is said on the head: building on which machine, since the local time (25.497138ms)
✔ the verdict heard while a 'building' write is in flight lands last and stays (6.472599ms)
[gitea] novox/hq@a6b08de8: the verdict of build-1791680941000000000 is an older run's than the check now pending; not written
✔ the verdict written while a start's read of the forge is in flight stays the newest status (0.841851ms)
✔ the verdict heard while a 'building' write still waits its turn: the 'building' is never written (0.434796ms)
✔ a start heard after the verdict writes nothing (0.454676ms)
✔ a start from before the check now pending (a run before a recheck, the stream's history) writes nothing (0.455477ms)
✔ an old run's verdict redelivered after a recheck's pending is not written over it (0.778111ms)
[gitea] novox/hq@a6b08de8: the verdict of build-1791680941000000000 is an older run's than the check the forge holds pending; not written
[gitea] novox/hq@a6b08de8: the verdict of build-1791680941000000000 is an older run's than the check now pending; not written
✔ an old run's verdict in flight when the recheck comes lands before the recheck's pending (2.399113ms)
✔ a rechecked head says where its new run is: building on which machine (0.858881ms)
✔ the recheck's own verdict is written, even asked a little before the pending by another machine's clock (0.652379ms)
✔ after a restart, an old run's verdict is weighed against the check the forge's history holds open (0.469437ms)
✔ a recheck the forge does not take fails its caller, and the head is as it was (0.459666ms)
✔ the status tool writes the merge check's pending through the writer, and the mesh's other statuses as they were (4.711374ms)
✔ when a check was asked is read from its id; the opening of the check now pending from the gate's history (0.526317ms)
✔ a head this process did not announce (it restarted) is said only while the forge holds its gate pending (0.422536ms)
✔ a repository as the mesh clones it is told from another (0.292404ms)
✔ gitea_api refuses the forge's administration, a branch's protection and the merge check's status, and sends nothing (7.917288ms)
✔ gitea_api refuses a file written to the default branch or one a rule guards, and writes one on a branch of its own (53.007923ms)
✔ a protection rule covers a branch as the forge decides it: classes, alternatives, a plain name in any case (issue 539) (19.246233ms)
✔ gitea_api still writes what an author writes (9.403138ms)
✔ gitea_branch_protection_set only strengthens (16.999692ms)
✔ gitea_merge_pull_request merges only a head whose required statuses all succeeded, and only that head (47.765702ms)
✔ gitea_merge_pull_request refuses a verdict judged against an earlier trunk, and asks the check once (152.54454ms)
✔ gitea_merge_pull_request merges only the head it is named, and refuses one that is not the head (19.512686ms)
✔ gitea_merge_pull_request is loud when the forge merged a head other than the one it was handed (15.33385ms)
✔ gitea_merge_pull_request says the merge happened when reading back what landed fails (30.158292ms)
✔ gitea_merge_pull_request reads back a fast-forward, and says it cannot read back a squash or a rebase (61.59406ms)
✔ gitea_delete_repo deletes only the tools' own user's repository, unguarded, untouched for 30 days, named in full (23.943336ms)
✔ gitea_api sends the path it judged, and refuses moving a pull request's base (3.168803ms)
✔ gitea_merge_pull_request merges only into a branch whose protection requires the checks (6.912384ms)
✔ a rule named for a branch is not made weaker than the glob rule that guards it now (12.834035ms)
✔ a repository owner or name is never another path (0.253213ms)
✔ the judges, without a forge (0.378245ms)
✔ gitea_api never acts as another user: sudo, in any case, and any query key it does not send are refused (1.627962ms)
✔ gitea_api never approves a pull request nor requests changes: those are the operator's (2.291311ms)
✔ no tool's caller text becomes another path or another user: `..` and `?sudo=` are refused before the forge (4.975518ms)
✔ every page of a pull request's files is read, though the forge caps a page at fifty (32.747787ms)
✔ a pass asks only the repositories that moved since the last look, every one before the first (0.468717ms)
✔ the directories above a merge's files that hold a module at the commit are said, the root never (13.263851ms)
✔ a lookup the forge refuses is thrown, not read as no module (5.310082ms)
✔ a renamed file is said under its old name too, as gone (novox/hq ADR 0267) (6.739941ms)
✔ a new head beyond the newest 20 open pull requests is announced, exactly once (50.828653ms)
✔ a pass pages until it reaches back past its floor, and no further (26.925058ms)
✔ the statuses asked for replace the rule's, and an empty list requires none (6.731991ms)
✔ a new rule refuses direct pushes unless asked; an administrator's override only when said (0.340814ms)
✔ each open pull request's new head is announced once, and nothing closed or merged (9.307127ms)
✔ a verdict is the head commit's status; an error is the forge's error, never a success (0.854122ms)
✔ each layer is its own status: the gate with the modules it judged, the repository's own check beside it (0.633079ms)
✔ no verdict sets warning on a required check; only a person's decision fails (1.057794ms)
✔ a commit status is set on the commit, under the merge check's context (38.243101ms)
✔ a change plan is the gate's result: said on the status and, when it builds something, as a comment (0.614348ms)
✔ a head announced is pending in the build queue, since the local time it was announced (0.311054ms)
✔ first start: mints with the admin account, keeps the token at 0600, asks for the tools' scopes only (36.6383ms)
✔ second start: reuses the kept token, mints nothing (17.552069ms)
✔ the forge rejects the kept token (its data was restored): minted afresh, once, and the call goes through (20.50366ms)
✔ a token kept with write:admin by an earlier build is minted again without it at the next start, and the forge drops the old one (22.70724ms)
✔ a kept token lacking a scope the tools need: the forge refuses the call for the scope, the token is re-minted with the whole list, and the call goes through (18.516192ms)
✔ the kept file is gone but the forge still holds a token by that name: replaced, not refused (18.33732ms)
✔ concurrent first calls share one mint (10.286281ms)
✔ the admin account is refused: said plainly, nothing kept, and the next call fails the same way rather than crashing (12.45969ms)
✔ one process shares one source per kept file — the watcher and the tools never renew against each other (0.942933ms)
✔ a second process finds the token the first renewed, and reuses it instead of minting over it (19.535246ms)
✔ a configured token wins, and is reported rather than minted over when the forge rejects it (2.497274ms)
✔ nothing to mint with and no token: the client says what is missing (0.206203ms)
[gitea] no tools — no Gitea token, and nothing to mint one with — set MESH_GITEA_TOKEN, or MESH_GITEA_ADMIN_USER, MESH_GITEA_ADMIN_PASSWORD_FILE, MESH_GITEA_STATE_DIR
[gitea] no token kept — minting one
[gitea] the forge already holds a token named "mesh-tools" — replacing it
[gitea] minted a token for "mesh-admin" (write:repository, write:issue, read:user), kept at /tmp/gitea-PVngJn/state/token
✔ the tools register once there is a way to a token, and the first call mints it (8.531087ms)
[gitea] no token kept — minting one
[gitea] the forge already holds a token named "mesh-tools" — replacing it
[gitea] minted a token for "mesh-admin" (write:repository, write:issue, read:user), kept at /tmp/gitea-e6cdfF/state/token
✔ a pull request can be closed, read and cleaned up through the tools (20.234186ms)
[gitea] novox/mesh-host's main moved from a2c2c35bbbbb to d28ca95ccccc: 25 open pull request(s) into it put out of date, nothing run (novox/hq ADR 0323)
[gitea] novox/mesh-host's main moved from d28ca95ccccc to ffffffffffff: 0 open pull request(s) into it put out of date, nothing run (novox/hq ADR 0323)
[gitea] novox/mesh-host: whether a trunk moved could not be weighed, so it is weighed again on the next pass — the forge is away
[gitea] novox/mesh-host's main moved from a2c2c35bbbbb to d28ca95ccccc: 0 open pull request(s) into it put out of date, nothing run (novox/hq ADR 0323)
✔ both statuses say the head and the trunk commit the verdict judged, first, whatever the summary's length (2.246481ms)
✔ a trunk's move puts an open head's merge check out of date, naming both commits, unless a check is on its way (1.327118ms)
✔ a push to a trunk puts every open pull request into it out of date — twenty-five of them — and none into another (5.88108ms)
✔ a trunk that cannot be read is weighed again on the next pass, never recorded half (0.523547ms)
✔ the trunk's move takes its turn after a verdict in flight, and weighs what the forge holds then (0.614319ms)
✔ a verdict judged against a trunk newer than the pass's first read is not put out of date (0.356505ms)
✔ a verdict that landed out of date keeps its account on the pull request (0.312625ms)
ℹ tests 84
ℹ suites 0
ℹ pass 84
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 675.256821
```
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Fixes novox/hq issue 578: a delivery group's order was not held for mesh-controller #271 and mesh-catalog #251 on 2026-10-12 at 18:11.
Cause
The merge tool deletes the source branch after a merge, which is its default. After that, the forge reads the pull request's
head.refasrefs/pull/<n>/head. Onlyhead.labelstill names the branch.GiteaClient.mapPulltookhead.ref, sopull.mergednamedrefs/pull/251/headas the head. A delivery group is the pull requests that share a head branch name (ADR 0249). The controller finds a merge's group by that head, and it joins a group's merges in one batch by that head too (ADR 0276 decision 5, ADR 0306 decision 2). Asked for the grouprefs/pull/251/head, mesh-delivery had none, and the controller logged "no member of a group is still to merge: it is cut at once" at 18:11:51. #251's walk (plan-1791821511497826695) went first. #271's merge was heard two seconds later, again under a pull ref, and was cut as a batch of its own, queued behind #251's walk.Both walks were on the controller's own path, because #251 moves mesh-delivery. So mesh-delivery's word held neither walk. Its
gocame from the walk starting, not from a turn it gave:let_gois empty on both deliveries. mesh-delivery's turn logic was not at fault. The group's order for own-path members is held by the controller's batching, and that batching never saw the group.A group merged through the tool has carried a pull ref as its head since the tool deleted branches by default. So the group wait (ADR 0306 decision 2) and the group-order edges inside a walk (ADR 0276 decision 5) did not hold for any such group.
Change
client.ts: the newheadBranchOfnameshead.refwhile the branch exists. Whenhead.refis a pull ref, it nameshead.label. When the forge gives no label, it keeps the ref and invents nothing.mapPulluses it, sopull.merged,pull.closed,pull.updatedand the tools all say the branch.ready, so it waits for the window. #271 joins #251's batch by head. The one walk puts mesh-controller and build-agent in a tier before gitea and mesh-delivery.Test (fails on main)
test/announcing.test.ts: "a merged pull request whose branch was deleted is announced with its branch, not its pull ref". A fake forge answers #251 as the real one did after its merge (labelis the branch,refisrefs/pull/251/head). On main the announcer's read givesrefs/pull/251/head. With this change it givesfeat/341-a-merge-check-judges-the-merge. The test also covers the cases where the branch exists, only a label is given, or only a pull ref is given.merge-check.sh(the Go part: checks and gitea's Go tests) passes. They ran with Node's type stripping and a resolver that maps.jsto.ts: 41 of gitea's 44 TypeScript tests pass, the new one included. The other 3 test files importdist/or the SDK, which are not on this machine. Not run locally:tsc. No TypeScript compiler is installed here, so themesh/repo-checkTypeScript part is the type check and the full test run.Rollout
One module: gitea on novox. It does not depend on a controller or mesh-delivery change, and it can go in any order. A merge announced before it rolls still carries a pull ref.
Self-review
merge-window-at-most(10 minutes) for the window, as before (ADR 0276 decision 9). A failed or stopped controller member is handled as before.Left as it was, and said here: when a group's later member merges more than
merge-window-at-mostbefore an earlier one, it still walks first. The window is the bound ADR 0306 chose.Delivery
novox/mesh-catalog@01d2ff7a46bc— delivered since 2026-10-12T19:09:32ZDelivery plan — builds gitea → novox; no bus step
Machines
Transitions
The commit's note under
refs/notes/mesh-plankeeps every transition:git log --notes=mesh-plan.Merge check at
01d2ff7aJudged: head
01d2ff7a46bcmerged into trunk22071dabfc7b; read beside it: mesh-controller84deb530ee5a, mesh-controller-main84deb530ee5a, mesh-host303d4037e84e, mesh-tools954e1a53eef9mesh/merge-gate: PASS — every machine composes with the change as it did without (4 of 4 compose) (modules: gitea)mesh/repo-check: PASS — its merge-check.sh passed, each part in its toolchain: merge-check.sh (go), merge-check-typescript.sh (typescript)Delivery plan — builds gitea → novox; no bus step
Run by the build seat on novox as
build-1791831469096804104(builds --log build-1791831469096804104).Out of date when it landed: out of date: judged against trunk
22071dabfc, the trunk is now 2fe5601397ae; judged again on a recheck or when it is mergedMerge check at
01d2ff7aJudged: head
01d2ff7a46bcmerged into trunk2fe5601397ae; read beside it: mesh-controller84deb530ee5a, mesh-controller-main84deb530ee5a, mesh-host303d4037e84e, mesh-tools954e1a53eef9mesh/merge-gate: PASS — every machine composes with the change as it did without (4 of 4 compose) (modules: gitea)mesh/repo-check: PASS — its merge-check.sh passed, each part in its toolchain: merge-check.sh (go), merge-check-typescript.sh (typescript)Delivery plan — builds gitea → novox; no bus step
Run by the build seat on shanks as
build-1791831837030233002(builds --log build-1791831837030233002).