Builds the forge's and mesh-delivery's part of novox/hq ADR 0323 (issue 341). Issue 406's rebase symptom is fixed with it. The controller's part is mesh-controller#271: there the build seat judges the head merged into the trunk's head and the verdict names the trunk commit.
Merge after mesh-controller#271 is delivered to the controller and every build seat holder (see Rollout).
gitea (TypeScript)
trunk.ts (new): what a verdict judged, in words (head … merged into trunk …). Also the out-of-date words, which statuses a trunk's move resets, the merge tool's comparison, the comment of a verdict that landed out of date, the record of how long a repository's last check took, and the pull.updated announcement, shared by the watcher and the merge tool. The announcement now also carries base_sha, the base branch's head; the controller names it as the trunk commit of a verdict no build judges.
pulls.ts: both statuses' descriptions begin with head <12> merged into trunk <12> — when the verdict names a trunk, so the 140-character clip never drops it. The comment says what was judged and which commits were read beside it.
index.ts: on each pass, every trunk an open pull request merges into is read. The open pull requests are paged up to 1000, and the trunk heads are kept in trunks-seen.json. When a trunk moved, mesh/merge-gate and mesh/repo-check of every open pull request into it go back to pending with "out of date: judged against trunk X, the trunk is now Y; …". The trunk is read again in each head's own write turn, so a verdict judged against a newer head is left alone. Nothing runs. A verdict that lands judged against a trunk that has moved since is written as out of date, and its comment is still posted. Each verdict's took is kept for the merge tool.
merge.ts (new): gitea_merge_pull_request, moved here from tools/index.ts and served from the watcher's bundle beside the merge check's one writer (issue 456). It reads the trunk commit the head's verdict names and the trunk's head. When they differ, nothing is merged and both commits are named. The statuses are set pending through the one writer ("waiting in the build queue … asked by the merge tool"), so "building on …" is said and no duplicate check is asked. The head is announced once, and the refusal says how long the last check of the repository took. It does not ask again while a check is on its way, puts the out-of-date words back if the announcement fails, and never merges by itself. The refusal names mesh-delivery.recheck for a check that never answers, and says that in a busy repository it may be refused and asked once more. A verdict that names no trunk commit is refused the same way.
gate.ts: outOfDate takes the head's turn on the one writer's chain and decides from what the forge holds at that moment. A verdict still in flight lands first.
New table row: ready → proposed on trunk moved. The guard is "its verdict names a trunk commit that is not its trunk's head", and the reason carries both commits.
The guard of proposed → checked is now "a verdict names this commit and the trunk's head". A late verdict for an old trunk is kept and shown but does not count, so the table reaches a fixed point.
A rejected delivery stays rejected. A verdict for it judged against another trunk is taken as its recheck.
An out-of-date proposed delivery has no bound and is not listed by stalled. Its one-hour bound starts when its head is announced again (the merge tool's ask). A forge status that reads "out of date:" is not said unanswered.
The verdict keeps head, trunk, beside commits and took. checks gives judged and out_of_date, show gives the verdict whole, and the kept plan records trunk.
Groups: the composed check keeps each member's trunk commit. A merge into a member's trunk that is not a member puts the composed check out of date: the group reads checking with the reason, and mesh/delivery-group goes pending on every member's head. It runs again once a member is judged again. A member rebased or pushed after the group closed takes its old head's place. The composed check is asked again for the new heads, counting members already merged or delivered, the group reads checking, and nobody is held.
mesh-delivery's trunk head is the last merge it heard into that trunk.
Tests (each fails on main)
gitea test/trunk.test.ts: the status words; 25 open pull requests into main all put out of date, one into release left; a first sight puts nothing out of date; a trunk that cannot be weighed is not recorded; a verdict in flight is weighed in its turn; a verdict judged against a trunk newer than the pass's first read is left; a late verdict keeps its account.
gitea test/guard.test.ts: the merge tool refuses naming both commits, asks once with base_sha in the announcement, writes its pending through the one writer, says how long the last check took and names mesh-delivery.recheck. It restores the out-of-date words when the announcement fails, and merges when the verdict names the trunk's head. test/token.test.ts lists the tools' bundle without the merge tool.
mesh-delivery issue341_test.go: the new row, rejected stays, stalled, the late verdict's fixed point, checks/show/plan, a non-member's merge against a member's own, the rebased member (also after another member was delivered), and the bound from the ask. table_test.go walks the new row.
Run: merge-check.sh (gitea and mesh-delivery, race) and merge-check-typescript.sh (81 of 81), against real copies of node_modules.
Rollout
mesh-controller#271 first: the controller and the build agent, on every build seat holder.
Then this PR. If gitea rolls first, the merge tool refuses every verdict that names no trunk, and #271 itself could not be merged through the tool: only a person's merge on the forge page would get it in. A build machine that sleeps through #271's rollout wakes with the old agent. Its verdicts name no trunk and are refused until it updates. mesh-delivery works in either order: a verdict naming no trunk counts as before.
When it goes live, nothing flips at once: the first pass records each trunk's head. After the next merge into a repository, its open pull requests read pending "out of date …". The open pull requests' trunk-less verdicts are then read out of date by the forge while mesh-delivery keeps their deliveries ready, until each is judged again. The merge tool asks one check for each of them at its merge.
Not here
The forge's holder does not write mesh/delivery-group. mesh-delivery is its one writer, and a second writer would race it (issue 446). The group's out-of-date state is mesh-delivery's, and is said through that status.
Builds the forge's and mesh-delivery's part of novox/hq ADR 0323 (issue 341). Issue 406's rebase symptom is fixed with it. The controller's part is mesh-controller#271: there the build seat judges the head merged into the trunk's head and the verdict names the trunk commit.
**Merge after mesh-controller#271 is delivered to the controller and every build seat holder** (see Rollout).
## gitea (TypeScript)
- `trunk.ts` (new): what a verdict judged, in words (`head … merged into trunk …`). Also the out-of-date words, which statuses a trunk's move resets, the merge tool's comparison, the comment of a verdict that landed out of date, the record of how long a repository's last check took, and the `pull.updated` announcement, shared by the watcher and the merge tool. The announcement now also carries `base_sha`, the base branch's head; the controller names it as the trunk commit of a verdict no build judges.
- `pulls.ts`: both statuses' descriptions begin with `head <12> merged into trunk <12> — ` when the verdict names a trunk, so the 140-character clip never drops it. The comment says what was judged and which commits were read beside it.
- `index.ts`: on each pass, every trunk an open pull request merges into is read. The open pull requests are paged up to 1000, and the trunk heads are kept in `trunks-seen.json`. When a trunk moved, `mesh/merge-gate` and `mesh/repo-check` of every open pull request into it go back to pending with "out of date: judged against trunk X, the trunk is now Y; …". The trunk is read again in each head's own write turn, so a verdict judged against a newer head is left alone. Nothing runs. A verdict that lands judged against a trunk that has moved since is written as out of date, and its comment is still posted. Each verdict's `took` is kept for the merge tool.
- `merge.ts` (new): `gitea_merge_pull_request`, moved here from `tools/index.ts` and served from the watcher's bundle beside the merge check's one writer (issue 456). It reads the trunk commit the head's verdict names and the trunk's head. When they differ, nothing is merged and both commits are named. The statuses are set pending through the one writer ("waiting in the build queue … asked by the merge tool"), so "building on …" is said and no duplicate check is asked. The head is announced once, and the refusal says how long the last check of the repository took. It does not ask again while a check is on its way, puts the out-of-date words back if the announcement fails, and never merges by itself. The refusal names `mesh-delivery.recheck` for a check that never answers, and says that in a busy repository it may be refused and asked once more. A verdict that names no trunk commit is refused the same way.
- `gate.ts`: `outOfDate` takes the head's turn on the one writer's chain and decides from what the forge holds at that moment. A verdict still in flight lands first.
- `client.ts`: `listOpenPulls` (paged), `branchHead`, `getRepo`.
## mesh-delivery (Go)
- New table row: `ready` → `proposed` on `trunk moved`. The guard is "its verdict names a trunk commit that is not its trunk's head", and the reason carries both commits.
- The guard of `proposed` → `checked` is now "a verdict names this commit and the trunk's head". A late verdict for an old trunk is kept and shown but does not count, so the table reaches a fixed point.
- A `rejected` delivery stays `rejected`. A verdict for it judged against another trunk is taken as its recheck.
- An out-of-date `proposed` delivery has no bound and is not listed by `stalled`. Its one-hour bound starts when its head is announced again (the merge tool's ask). A forge status that reads "out of date:" is not said `unanswered`.
- The verdict keeps head, trunk, beside commits and took. `checks` gives `judged` and `out_of_date`, `show` gives the verdict whole, and the kept plan records `trunk`.
- Groups: the composed check keeps each member's trunk commit. A merge into a member's trunk that is not a member puts the composed check out of date: the group reads `checking` with the reason, and `mesh/delivery-group` goes pending on every member's head. It runs again once a member is judged again. A member rebased or pushed after the group closed takes its old head's place. The composed check is asked again for the new heads, counting members already merged or delivered, the group reads `checking`, and nobody is held.
- mesh-delivery's trunk head is the last merge it heard into that trunk.
## Tests (each fails on main)
- gitea `test/trunk.test.ts`: the status words; 25 open pull requests into `main` all put out of date, one into `release` left; a first sight puts nothing out of date; a trunk that cannot be weighed is not recorded; a verdict in flight is weighed in its turn; a verdict judged against a trunk newer than the pass's first read is left; a late verdict keeps its account.
- gitea `test/guard.test.ts`: the merge tool refuses naming both commits, asks once with `base_sha` in the announcement, writes its pending through the one writer, says how long the last check took and names `mesh-delivery.recheck`. It restores the out-of-date words when the announcement fails, and merges when the verdict names the trunk's head. `test/token.test.ts` lists the tools' bundle without the merge tool.
- mesh-delivery `issue341_test.go`: the new row, rejected stays, `stalled`, the late verdict's fixed point, `checks`/`show`/plan, a non-member's merge against a member's own, the rebased member (also after another member was delivered), and the bound from the ask. `table_test.go` walks the new row.
- Run: `merge-check.sh` (gitea and mesh-delivery, race) and `merge-check-typescript.sh` (81 of 81), against real copies of node_modules.
## Rollout
1. **mesh-controller#271 first**: the controller and the build agent, on every build seat holder.
2. **Then this PR.** If gitea rolls first, the merge tool refuses every verdict that names no trunk, and #271 itself could not be merged through the tool: only a person's merge on the forge page would get it in. A build machine that sleeps through #271's rollout wakes with the old agent. Its verdicts name no trunk and are refused until it updates. mesh-delivery works in either order: a verdict naming no trunk counts as before.
3. **When it goes live**, nothing flips at once: the first pass records each trunk's head. After the next merge into a repository, its open pull requests read pending "out of date …". The open pull requests' trunk-less verdicts are then read out of date by the forge while mesh-delivery keeps their deliveries `ready`, until each is judged again. The merge tool asks one check for each of them at its merge.
## Not here
The forge's holder does not write `mesh/delivery-group`. mesh-delivery is its one writer, and a second writer would race it (issue 446). The group's out-of-date state is mesh-delivery's, and is said through that status.
A merge check's verdict now names the trunk commit it judged (novox/hq ADR
0323, issue 341), but the forge kept it green after another merge moved the
trunk, the merge tool merged on it, and mesh-delivery kept the delivery ready.
The forge's holder puts both statuses of every open pull request into a moved
trunk back to pending with the reason and runs nothing; the merge tool compares
the verdict's trunk commit with the trunk's head, asks the check once and never
merges by itself; mesh-delivery sends a ready delivery back to proposed, counts
only a verdict for the trunk's head, and composes a group again for a rebased
member instead of holding it.
composed together: pass — every machine composes with the change as it did without (4 of 4 compose)
Machines
gitea on novox: passed — healthy 3 times over 2m36s
mesh-delivery on novox: passed — judged with gitea: healthy 3 times over 2m36s
gitea on the rest running it: sent
mesh-delivery on the rest running it: sent
Transitions
2026-10-12 15:52 (new) → proposed (announced): novox/mesh-catalog#251's head announced
2026-10-12 15:56 proposed → checked (checked): the verdict names this commit
2026-10-12 15:56 checked → ready (accepted): the gate passed or warned, and the repository's own check did not fail
2026-10-12 16:11 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 16:11 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 16:15 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@41b6e10077ef` — **delivered** since 2026-10-12T16:15:24Z
**Delivery plan** — builds gitea, mesh-delivery → novox; no bus step
- tier 0: gitea, mesh-delivery
- novox: receives gitea, mesh-delivery
- 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
**Group** `feat/341-a-merge-check-judges-the-merge`, in order: novox/hq@09916e2bbc07 → novox/mesh-controller@875ebc093e18 → novox/mesh-catalog@41b6e10077ef → novox/mesh-lab@79108a72a0df
- novox/mesh-controller@875ebc093e18 before novox/mesh-catalog@41b6e10077ef: built by
- composed together: pass — every machine composes with the change as it did without (4 of 4 compose)
**Machines**
- gitea on novox: passed — healthy 3 times over 2m36s
- mesh-delivery on novox: passed — judged with gitea: healthy 3 times over 2m36s
- gitea on the rest running it: sent
- mesh-delivery on the rest running it: sent
**Transitions**
- 2026-10-12 15:52 (new) → proposed (announced): novox/mesh-catalog#251's head announced
- 2026-10-12 15:56 proposed → checked (checked): the verdict names this commit
- 2026-10-12 15:56 checked → ready (accepted): the gate passed or warned, and the repository's own check did not fail
- 2026-10-12 16:11 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 16:11 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 16:15 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/merge-gate: PASS — every machine composes with the change as it did without (4 of 4 compose) (modules: gitea, mesh-delivery)
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, mesh-delivery → novox; no bus step
tier 0: gitea, mesh-delivery
novox: receives gitea, mesh-delivery
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
read by no module's build: modules/mesh-delivery/cmd/mesh-delivery/covered_test.go, modules/mesh-delivery/cmd/mesh-delivery/issue341_test.go, modules/mesh-delivery/cmd/mesh-delivery/table_test.go
Run by the build seat on ace as build-1791819071374743686 (builds --log build-1791819071374743686).
--- the gate: gitea, mesh-delivery
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)
mesh-delivery: ok, 1 tool(s), invokes seat:mesh-controller.delivery-plan, seat:mesh-controller.delivery-order, seat:mesh-controller.delivery-check, seat:mesh-controller.deliver, seat:mesh-controller.delivery-stop, seat:mesh-controller.delivery-walks, gitea.gitea_note_append, gitea.gitea_delivery_view, gitea.gitea_commit_status, gitea.gitea_commit_statuses, gitea.gitea_contains, gitea.gitea_open_pulls, keeps state deliveries, groups
long-running resources without health: 1
files asking for a setting without saying whether it is trusted: 0
2 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 2026-10-12T14:54:57Z 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 2026-10-12T14:54:57Z
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 2 module(s) in 1 tier(s): gitea, mesh-delivery
notes:
- modules/mesh-delivery/cmd/mesh-delivery/covered_test.go, modules/mesh-delivery/cmd/mesh-delivery/issue341_test.go and modules/mesh-delivery/cmd/mesh-delivery/table_test.go read by no module's build, and rebuild nothing (issue 280)
--- its merge-check.sh (go toolchain)
ok git.novox.be/novox/mesh-catalog/checks/words 0.002s
words: 12 retired words and 70 daemons' lines, none in the 44397 strings of 700 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 0.005s
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 0.206s
testing modules/gitea
ok gitea/cmd/forge 1.419s
ok gitea/cmd/npm-registry 1.740s
? gitea/internal/forgehttp [no test files]
testing modules/mesh-delivery
ok mesh-delivery/cmd/mesh-delivery 2.406s
--- 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 (8.732882ms)
✔ every open pull request is read with what its base requires and what its head has (gitea_open_pulls) (43.865101ms)
✔ a pull request or issue number that is not one is refused before the forge is asked (1.623935ms)
✔ the newest status of each context is kept, with what it said, when and by whom (18.003182ms)
✔ a commit merges only when every required status is a success; a warning blocks (0.489381ms)
✔ no rule requires nothing; a rule that cannot be read says nothing, never yes (0.217499ms)
✔ a commit is contained in another when the compare finds nothing on it the other lacks (0.642339ms)
✔ a note line is appended once, as the forge's user, in its own repository (9.591596ms)
✔ a note that cannot be written is said, never read as written (0.474704ms)
✔ one view per pull request, found by its marker; only the mesh's statuses (0.901776ms)
✔ a build machine taking the check is said on the head: building on which machine, since the local time (11.395149ms)
✔ the verdict heard while a 'building' write is in flight lands last and stays (6.228398ms)
[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 (1.008117ms)
✔ the verdict heard while a 'building' write still waits its turn: the 'building' is never written (0.514358ms)
✔ a start heard after the verdict writes nothing (0.674359ms)
✔ a start from before the check now pending (a run before a recheck, the stream's history) writes nothing (0.56324ms)
✔ an old run's verdict redelivered after a recheck's pending is not written over it (0.939357ms)
[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 (2.797824ms)
✔ a rechecked head says where its new run is: building on which machine (0.869576ms)
✔ the recheck's own verdict is written, even asked a little before the pending by another machine's clock (0.734532ms)
✔ after a restart, an old run's verdict is weighed against the check the forge's history holds open (2.264901ms)
✔ a recheck the forge does not take fails its caller, and the head is as it was (0.414871ms)
✔ the status tool writes the merge check's pending through the writer, and the mesh's other statuses as they were (3.559718ms)
✔ when a check was asked is read from its id; the opening of the check now pending from the gate's history (0.154712ms)
✔ a head this process did not announce (it restarted) is said only while the forge holds its gate pending (0.275398ms)
✔ a repository as the mesh clones it is told from another (0.186802ms)
✔ gitea_api refuses the forge's administration, a branch's protection and the merge check's status, and sends nothing (4.73035ms)
✔ gitea_api refuses a file written to the default branch or one a rule guards, and writes one on a branch of its own (43.614041ms)
✔ a protection rule covers a branch as the forge decides it: classes, alternatives, a plain name in any case (issue 539) (16.840444ms)
✔ gitea_api still writes what an author writes (7.855712ms)
✔ gitea_branch_protection_set only strengthens (13.965947ms)
✔ gitea_merge_pull_request merges only a head whose required statuses all succeeded, and only that head (39.285596ms)
✔ gitea_merge_pull_request refuses a verdict judged against an earlier trunk, and asks the check once (132.838186ms)
✔ gitea_merge_pull_request merges only the head it is named, and refuses one that is not the head (16.743523ms)
✔ gitea_merge_pull_request is loud when the forge merged a head other than the one it was handed (12.613974ms)
✔ gitea_merge_pull_request says the merge happened when reading back what landed fails (25.380722ms)
✔ gitea_merge_pull_request reads back a fast-forward, and says it cannot read back a squash or a rebase (54.96659ms)
✔ gitea_delete_repo deletes only the tools' own user's repository, unguarded, untouched for 30 days, named in full (22.185733ms)
✔ gitea_api sends the path it judged, and refuses moving a pull request's base (3.737351ms)
✔ gitea_merge_pull_request merges only into a branch whose protection requires the checks (6.600981ms)
✔ a rule named for a branch is not made weaker than the glob rule that guards it now (13.007153ms)
✔ a repository owner or name is never another path (0.248858ms)
✔ the judges, without a forge (0.416404ms)
✔ gitea_api never acts as another user: sudo, in any case, and any query key it does not send are refused (2.777506ms)
✔ gitea_api never approves a pull request nor requests changes: those are the operator's (1.959968ms)
✔ no tool's caller text becomes another path or another user: `..` and `?sudo=` are refused before the forge (4.065009ms)
✔ every page of a pull request's files is read, though the forge caps a page at fifty (22.129165ms)
✔ a pass asks only the repositories that moved since the last look, every one before the first (0.355719ms)
✔ the directories above a merge's files that hold a module at the commit are said, the root never (10.140941ms)
✔ a lookup the forge refuses is thrown, not read as no module (2.731109ms)
✔ a renamed file is said under its old name too, as gone (novox/hq ADR 0267) (3.684883ms)
✔ the statuses asked for replace the rule's, and an empty list requires none (4.749286ms)
✔ a new rule refuses direct pushes unless asked; an administrator's override only when said (0.224232ms)
✔ each open pull request's new head is announced once, and nothing closed or merged (6.501803ms)
✔ a verdict is the head commit's status; an error is the forge's error, never a success (0.64278ms)
✔ each layer is its own status: the gate with the modules it judged, the repository's own check beside it (0.494781ms)
✔ no verdict sets warning on a required check; only a person's decision fails (0.707331ms)
✔ a commit status is set on the commit, under the merge check's context (27.351272ms)
✔ a change plan is the gate's result: said on the status and, when it builds something, as a comment (0.395454ms)
✔ a head announced is pending in the build queue, since the local time it was announced (0.197271ms)
✔ first start: mints with the admin account, keeps the token at 0600, asks for the tools' scopes only (28.832359ms)
✔ second start: reuses the kept token, mints nothing (17.324847ms)
✔ the forge rejects the kept token (its data was restored): minted afresh, once, and the call goes through (20.854868ms)
✔ 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 (19.142205ms)
✔ 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 (16.987653ms)
✔ the kept file is gone but the forge still holds a token by that name: replaced, not refused (15.396357ms)
✔ concurrent first calls share one mint (9.309839ms)
✔ the admin account is refused: said plainly, nothing kept, and the next call fails the same way rather than crashing (9.796893ms)
✔ one process shares one source per kept file — the watcher and the tools never renew against each other (0.849869ms)
✔ a second process finds the token the first renewed, and reuses it instead of minting over it (16.917089ms)
✔ a configured token wins, and is reported rather than minted over when the forge rejects it (2.233002ms)
✔ nothing to mint with and no token: the client says what is missing (0.158689ms)
[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-D1wNGs/state/token
✔ the tools register once there is a way to a token, and the first call mints it (8.1852ms)
[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-vIgv46/state/token
✔ a pull request can be closed, read and cleaned up through the tools (18.716175ms)
[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
✔ both statuses say the head and the trunk commit the verdict judged, first, whatever the summary's length (1.378794ms)
✔ a trunk's move puts an open head's merge check out of date, naming both commits, unless a check is on its way (0.953654ms)
✔ a push to a trunk puts every open pull request into it out of date — twenty-five of them — and none into another (3.161358ms)
✔ a trunk that cannot be read is weighed again on the next pass, never recorded half (0.291779ms)
✔ the trunk's move takes its turn after a verdict in flight, and weighs what the forge holds then (0.373182ms)
ℹ tests 79
ℹ suites 0
ℹ pass 79
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 561.634108
**Merge check** at `82cca650`
- `mesh/merge-gate`: **PASS** — every machine composes with the change as it did without (4 of 4 compose) (modules: gitea, mesh-delivery)
- `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, mesh-delivery → novox; no bus step
- tier 0: gitea, mesh-delivery
- novox: receives gitea, mesh-delivery
- 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
- read by no module's build: modules/mesh-delivery/cmd/mesh-delivery/covered_test.go, modules/mesh-delivery/cmd/mesh-delivery/issue341_test.go, modules/mesh-delivery/cmd/mesh-delivery/table_test.go
Run by the build seat on ace as `build-1791819071374743686` (`builds --log build-1791819071374743686`).
```
--- the gate: gitea, mesh-delivery
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)
mesh-delivery: ok, 1 tool(s), invokes seat:mesh-controller.delivery-plan, seat:mesh-controller.delivery-order, seat:mesh-controller.delivery-check, seat:mesh-controller.deliver, seat:mesh-controller.delivery-stop, seat:mesh-controller.delivery-walks, gitea.gitea_note_append, gitea.gitea_delivery_view, gitea.gitea_commit_status, gitea.gitea_commit_statuses, gitea.gitea_contains, gitea.gitea_open_pulls, keeps state deliveries, groups
long-running resources without health: 1
files asking for a setting without saying whether it is trusted: 0
2 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 2026-10-12T14:54:57Z 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 2026-10-12T14:54:57Z
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 2 module(s) in 1 tier(s): gitea, mesh-delivery
notes:
- modules/mesh-delivery/cmd/mesh-delivery/covered_test.go, modules/mesh-delivery/cmd/mesh-delivery/issue341_test.go and modules/mesh-delivery/cmd/mesh-delivery/table_test.go read by no module's build, and rebuild nothing (issue 280)
--- its merge-check.sh (go toolchain)
ok git.novox.be/novox/mesh-catalog/checks/words 0.002s
words: 12 retired words and 70 daemons' lines, none in the 44397 strings of 700 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 0.005s
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 0.206s
testing modules/gitea
ok gitea/cmd/forge 1.419s
ok gitea/cmd/npm-registry 1.740s
? gitea/internal/forgehttp [no test files]
testing modules/mesh-delivery
ok mesh-delivery/cmd/mesh-delivery 2.406s
--- 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 (8.732882ms)
✔ every open pull request is read with what its base requires and what its head has (gitea_open_pulls) (43.865101ms)
✔ a pull request or issue number that is not one is refused before the forge is asked (1.623935ms)
✔ the newest status of each context is kept, with what it said, when and by whom (18.003182ms)
✔ a commit merges only when every required status is a success; a warning blocks (0.489381ms)
✔ no rule requires nothing; a rule that cannot be read says nothing, never yes (0.217499ms)
✔ a commit is contained in another when the compare finds nothing on it the other lacks (0.642339ms)
✔ a note line is appended once, as the forge's user, in its own repository (9.591596ms)
✔ a note that cannot be written is said, never read as written (0.474704ms)
✔ one view per pull request, found by its marker; only the mesh's statuses (0.901776ms)
✔ a build machine taking the check is said on the head: building on which machine, since the local time (11.395149ms)
✔ the verdict heard while a 'building' write is in flight lands last and stays (6.228398ms)
[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 (1.008117ms)
✔ the verdict heard while a 'building' write still waits its turn: the 'building' is never written (0.514358ms)
✔ a start heard after the verdict writes nothing (0.674359ms)
✔ a start from before the check now pending (a run before a recheck, the stream's history) writes nothing (0.56324ms)
✔ an old run's verdict redelivered after a recheck's pending is not written over it (0.939357ms)
[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 (2.797824ms)
✔ a rechecked head says where its new run is: building on which machine (0.869576ms)
✔ the recheck's own verdict is written, even asked a little before the pending by another machine's clock (0.734532ms)
✔ after a restart, an old run's verdict is weighed against the check the forge's history holds open (2.264901ms)
✔ a recheck the forge does not take fails its caller, and the head is as it was (0.414871ms)
✔ the status tool writes the merge check's pending through the writer, and the mesh's other statuses as they were (3.559718ms)
✔ when a check was asked is read from its id; the opening of the check now pending from the gate's history (0.154712ms)
✔ a head this process did not announce (it restarted) is said only while the forge holds its gate pending (0.275398ms)
✔ a repository as the mesh clones it is told from another (0.186802ms)
✔ gitea_api refuses the forge's administration, a branch's protection and the merge check's status, and sends nothing (4.73035ms)
✔ gitea_api refuses a file written to the default branch or one a rule guards, and writes one on a branch of its own (43.614041ms)
✔ a protection rule covers a branch as the forge decides it: classes, alternatives, a plain name in any case (issue 539) (16.840444ms)
✔ gitea_api still writes what an author writes (7.855712ms)
✔ gitea_branch_protection_set only strengthens (13.965947ms)
✔ gitea_merge_pull_request merges only a head whose required statuses all succeeded, and only that head (39.285596ms)
✔ gitea_merge_pull_request refuses a verdict judged against an earlier trunk, and asks the check once (132.838186ms)
✔ gitea_merge_pull_request merges only the head it is named, and refuses one that is not the head (16.743523ms)
✔ gitea_merge_pull_request is loud when the forge merged a head other than the one it was handed (12.613974ms)
✔ gitea_merge_pull_request says the merge happened when reading back what landed fails (25.380722ms)
✔ gitea_merge_pull_request reads back a fast-forward, and says it cannot read back a squash or a rebase (54.96659ms)
✔ gitea_delete_repo deletes only the tools' own user's repository, unguarded, untouched for 30 days, named in full (22.185733ms)
✔ gitea_api sends the path it judged, and refuses moving a pull request's base (3.737351ms)
✔ gitea_merge_pull_request merges only into a branch whose protection requires the checks (6.600981ms)
✔ a rule named for a branch is not made weaker than the glob rule that guards it now (13.007153ms)
✔ a repository owner or name is never another path (0.248858ms)
✔ the judges, without a forge (0.416404ms)
✔ gitea_api never acts as another user: sudo, in any case, and any query key it does not send are refused (2.777506ms)
✔ gitea_api never approves a pull request nor requests changes: those are the operator's (1.959968ms)
✔ no tool's caller text becomes another path or another user: `..` and `?sudo=` are refused before the forge (4.065009ms)
✔ every page of a pull request's files is read, though the forge caps a page at fifty (22.129165ms)
✔ a pass asks only the repositories that moved since the last look, every one before the first (0.355719ms)
✔ the directories above a merge's files that hold a module at the commit are said, the root never (10.140941ms)
✔ a lookup the forge refuses is thrown, not read as no module (2.731109ms)
✔ a renamed file is said under its old name too, as gone (novox/hq ADR 0267) (3.684883ms)
✔ the statuses asked for replace the rule's, and an empty list requires none (4.749286ms)
✔ a new rule refuses direct pushes unless asked; an administrator's override only when said (0.224232ms)
✔ each open pull request's new head is announced once, and nothing closed or merged (6.501803ms)
✔ a verdict is the head commit's status; an error is the forge's error, never a success (0.64278ms)
✔ each layer is its own status: the gate with the modules it judged, the repository's own check beside it (0.494781ms)
✔ no verdict sets warning on a required check; only a person's decision fails (0.707331ms)
✔ a commit status is set on the commit, under the merge check's context (27.351272ms)
✔ a change plan is the gate's result: said on the status and, when it builds something, as a comment (0.395454ms)
✔ a head announced is pending in the build queue, since the local time it was announced (0.197271ms)
✔ first start: mints with the admin account, keeps the token at 0600, asks for the tools' scopes only (28.832359ms)
✔ second start: reuses the kept token, mints nothing (17.324847ms)
✔ the forge rejects the kept token (its data was restored): minted afresh, once, and the call goes through (20.854868ms)
✔ 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 (19.142205ms)
✔ 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 (16.987653ms)
✔ the kept file is gone but the forge still holds a token by that name: replaced, not refused (15.396357ms)
✔ concurrent first calls share one mint (9.309839ms)
✔ the admin account is refused: said plainly, nothing kept, and the next call fails the same way rather than crashing (9.796893ms)
✔ one process shares one source per kept file — the watcher and the tools never renew against each other (0.849869ms)
✔ a second process finds the token the first renewed, and reuses it instead of minting over it (16.917089ms)
✔ a configured token wins, and is reported rather than minted over when the forge rejects it (2.233002ms)
✔ nothing to mint with and no token: the client says what is missing (0.158689ms)
[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-D1wNGs/state/token
✔ the tools register once there is a way to a token, and the first call mints it (8.1852ms)
[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-vIgv46/state/token
✔ a pull request can be closed, read and cleaned up through the tools (18.716175ms)
[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
✔ both statuses say the head and the trunk commit the verdict judged, first, whatever the summary's length (1.378794ms)
✔ a trunk's move puts an open head's merge check out of date, naming both commits, unless a check is on its way (0.953654ms)
✔ a push to a trunk puts every open pull request into it out of date — twenty-five of them — and none into another (3.161358ms)
✔ a trunk that cannot be read is weighed again on the next pass, never recorded half (0.291779ms)
✔ the trunk's move takes its turn after a verdict in flight, and weighs what the forge holds then (0.373182ms)
ℹ tests 79
ℹ suites 0
ℹ pass 79
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 561.634108
```
The merge tool's pending bypassed the merge check's one writer from the tools'
bundle (issue 456); it is now served beside it. A verdict no build judges names
the announced base head, so the merge tool no longer refuses it for ever. A
late verdict keeps its account, a trunk is read again in each head's turn so a
fresh verdict is not put out of date, and a delivered member counts when a
rebased member's group is composed again (novox/hq ADR 0323).
mesh/merge-gate: PASS — every machine composes with the change as it did without (4 of 4 compose) (modules: gitea, mesh-delivery)
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, mesh-delivery → novox; no bus step
tier 0: gitea, mesh-delivery
novox: receives gitea, mesh-delivery
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
read by no module's build: modules/mesh-delivery/cmd/mesh-delivery/covered_test.go, modules/mesh-delivery/cmd/mesh-delivery/issue341_test.go, modules/mesh-delivery/cmd/mesh-delivery/table_test.go
Run by the build seat on g14 as build-1791820016890812212 (builds --log build-1791820016890812212).
--- the gate: gitea, mesh-delivery
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)
mesh-delivery: ok, 1 tool(s), invokes seat:mesh-controller.delivery-plan, seat:mesh-controller.delivery-order, seat:mesh-controller.delivery-check, seat:mesh-controller.deliver, seat:mesh-controller.delivery-stop, seat:mesh-controller.delivery-walks, gitea.gitea_note_append, gitea.gitea_delivery_view, gitea.gitea_commit_status, gitea.gitea_commit_statuses, gitea.gitea_contains, gitea.gitea_open_pulls, keeps state deliveries, groups
long-running resources without health: 1
files asking for a setting without saying whether it is trusted: 0
2 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 2026-10-12T15:47:39Z 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 2026-10-12T15:47:39Z
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 2 module(s) in 1 tier(s): gitea, mesh-delivery
notes:
- modules/mesh-delivery/cmd/mesh-delivery/covered_test.go, modules/mesh-delivery/cmd/mesh-delivery/issue341_test.go and modules/mesh-delivery/cmd/mesh-delivery/table_test.go read by no module's build, and rebuild nothing (issue 280)
--- its merge-check.sh (go toolchain)
ok git.novox.be/novox/mesh-catalog/checks/words 0.002s
words: 12 retired words and 70 daemons' lines, none in the 44405 strings of 701 files
ok git.novox.be/novox/mesh-catalog/checks/agenthome 0.011s
agent-home: no archive and no file written into under the agent's home
ok git.novox.be/novox/mesh-catalog/checks/paths 0.005s
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 0.280s
testing modules/gitea
ok gitea/cmd/forge 1.426s
ok gitea/cmd/npm-registry 1.736s
? gitea/internal/forgehttp [no test files]
testing modules/mesh-delivery
ok mesh-delivery/cmd/mesh-delivery 2.964s
--- 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 (11.112343ms)
✔ every open pull request is read with what its base requires and what its head has (gitea_open_pulls) (51.319649ms)
✔ a pull request or issue number that is not one is refused before the forge is asked (1.317982ms)
✔ the newest status of each context is kept, with what it said, when and by whom (23.14552ms)
✔ a commit merges only when every required status is a success; a warning blocks (0.693065ms)
✔ no rule requires nothing; a rule that cannot be read says nothing, never yes (0.305786ms)
✔ a commit is contained in another when the compare finds nothing on it the other lacks (0.857905ms)
✔ a note line is appended once, as the forge's user, in its own repository (10.216616ms)
✔ a note that cannot be written is said, never read as written (0.528125ms)
✔ one view per pull request, found by its marker; only the mesh's statuses (1.277034ms)
✔ a build machine taking the check is said on the head: building on which machine, since the local time (20.550674ms)
✔ the verdict heard while a 'building' write is in flight lands last and stays (9.07092ms)
[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 (1.130389ms)
✔ the verdict heard while a 'building' write still waits its turn: the 'building' is never written (0.563822ms)
✔ a start heard after the verdict writes nothing (0.569934ms)
✔ a start from before the check now pending (a run before a recheck, the stream's history) writes nothing (1.472172ms)
✔ an old run's verdict redelivered after a recheck's pending is not written over it (1.021313ms)
[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 (2.265776ms)
✔ a rechecked head says where its new run is: building on which machine (0.925934ms)
✔ the recheck's own verdict is written, even asked a little before the pending by another machine's clock (0.753759ms)
✔ after a restart, an old run's verdict is weighed against the check the forge's history holds open (0.623884ms)
✔ a recheck the forge does not take fails its caller, and the head is as it was (0.559864ms)
✔ the status tool writes the merge check's pending through the writer, and the mesh's other statuses as they were (4.954488ms)
✔ when a check was asked is read from its id; the opening of the check now pending from the gate's history (0.256974ms)
✔ a head this process did not announce (it restarted) is said only while the forge holds its gate pending (0.348907ms)
✔ a repository as the mesh clones it is told from another (0.251093ms)
✔ gitea_api refuses the forge's administration, a branch's protection and the merge check's status, and sends nothing (4.479083ms)
✔ gitea_api refuses a file written to the default branch or one a rule guards, and writes one on a branch of its own (49.952375ms)
✔ a protection rule covers a branch as the forge decides it: classes, alternatives, a plain name in any case (issue 539) (15.957566ms)
✔ gitea_api still writes what an author writes (7.893523ms)
✔ gitea_branch_protection_set only strengthens (14.556618ms)
✔ gitea_merge_pull_request merges only a head whose required statuses all succeeded, and only that head (39.624078ms)
✔ gitea_merge_pull_request refuses a verdict judged against an earlier trunk, and asks the check once (140.143145ms)
✔ gitea_merge_pull_request merges only the head it is named, and refuses one that is not the head (14.017263ms)
✔ gitea_merge_pull_request is loud when the forge merged a head other than the one it was handed (14.007064ms)
✔ gitea_merge_pull_request says the merge happened when reading back what landed fails (24.455908ms)
✔ gitea_merge_pull_request reads back a fast-forward, and says it cannot read back a squash or a rebase (56.197103ms)
✔ gitea_delete_repo deletes only the tools' own user's repository, unguarded, untouched for 30 days, named in full (20.499269ms)
✔ gitea_api sends the path it judged, and refuses moving a pull request's base (3.970696ms)
✔ gitea_merge_pull_request merges only into a branch whose protection requires the checks (6.496261ms)
✔ a rule named for a branch is not made weaker than the glob rule that guards it now (15.243552ms)
✔ a repository owner or name is never another path (0.780049ms)
✔ the judges, without a forge (0.374044ms)
✔ gitea_api never acts as another user: sudo, in any case, and any query key it does not send are refused (1.850794ms)
✔ gitea_api never approves a pull request nor requests changes: those are the operator's (2.927542ms)
✔ no tool's caller text becomes another path or another user: `..` and `?sudo=` are refused before the forge (5.317963ms)
✔ every page of a pull request's files is read, though the forge caps a page at fifty (24.667236ms)
✔ a pass asks only the repositories that moved since the last look, every one before the first (0.364797ms)
✔ the directories above a merge's files that hold a module at the commit are said, the root never (10.966088ms)
✔ a lookup the forge refuses is thrown, not read as no module (3.29784ms)
✔ a renamed file is said under its old name too, as gone (novox/hq ADR 0267) (4.563102ms)
✔ the statuses asked for replace the rule's, and an empty list requires none (5.526917ms)
✔ a new rule refuses direct pushes unless asked; an administrator's override only when said (0.359927ms)
✔ each open pull request's new head is announced once, and nothing closed or merged (12.795984ms)
✔ a verdict is the head commit's status; an error is the forge's error, never a success (1.027014ms)
✔ each layer is its own status: the gate with the modules it judged, the repository's own check beside it (0.732218ms)
✔ no verdict sets warning on a required check; only a person's decision fails (1.023517ms)
✔ a commit status is set on the commit, under the merge check's context (34.14956ms)
✔ a change plan is the gate's result: said on the status and, when it builds something, as a comment (0.527133ms)
✔ a head announced is pending in the build queue, since the local time it was announced (0.232849ms)
✔ first start: mints with the admin account, keeps the token at 0600, asks for the tools' scopes only (32.99151ms)
✔ second start: reuses the kept token, mints nothing (13.941661ms)
✔ the forge rejects the kept token (its data was restored): minted afresh, once, and the call goes through (19.007639ms)
✔ 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 (22.745237ms)
✔ 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 (17.643ms)
✔ the kept file is gone but the forge still holds a token by that name: replaced, not refused (16.335658ms)
✔ concurrent first calls share one mint (10.683456ms)
✔ the admin account is refused: said plainly, nothing kept, and the next call fails the same way rather than crashing (13.356558ms)
✔ one process shares one source per kept file — the watcher and the tools never renew against each other (1.134246ms)
✔ a second process finds the token the first renewed, and reuses it instead of minting over it (24.437874ms)
✔ a configured token wins, and is reported rather than minted over when the forge rejects it (3.012823ms)
✔ nothing to mint with and no token: the client says what is missing (0.248639ms)
[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-XXF2gQ/state/token
✔ the tools register once there is a way to a token, and the first call mints it (8.379698ms)
[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-kdrJs0/state/token
✔ a pull request can be closed, read and cleaned up through the tools (19.738956ms)
[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 (1.545771ms)
✔ a trunk's move puts an open head's merge check out of date, naming both commits, unless a check is on its way (0.980336ms)
✔ a push to a trunk puts every open pull request into it out of date — twenty-five of them — and none into another (4.750775ms)
✔ a trunk that cannot be read is weighed again on the next pass, never recorded half (0.388221ms)
✔ the trunk's move takes its turn after a verdict in flight, and weighs what the forge holds then (0.432223ms)
✔ a verdict judged against a trunk newer than the pass's first read is not put out of date (0.726518ms)
✔ a verdict that landed out of date keeps its account on the pull request (0.926074ms)
ℹ tests 81
ℹ suites 0
ℹ pass 81
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 606.783472
**Merge check** at `434506f9`
- `mesh/merge-gate`: **PASS** — every machine composes with the change as it did without (4 of 4 compose) (modules: gitea, mesh-delivery)
- `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, mesh-delivery → novox; no bus step
- tier 0: gitea, mesh-delivery
- novox: receives gitea, mesh-delivery
- 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
- read by no module's build: modules/mesh-delivery/cmd/mesh-delivery/covered_test.go, modules/mesh-delivery/cmd/mesh-delivery/issue341_test.go, modules/mesh-delivery/cmd/mesh-delivery/table_test.go
Run by the build seat on g14 as `build-1791820016890812212` (`builds --log build-1791820016890812212`).
```
--- the gate: gitea, mesh-delivery
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)
mesh-delivery: ok, 1 tool(s), invokes seat:mesh-controller.delivery-plan, seat:mesh-controller.delivery-order, seat:mesh-controller.delivery-check, seat:mesh-controller.deliver, seat:mesh-controller.delivery-stop, seat:mesh-controller.delivery-walks, gitea.gitea_note_append, gitea.gitea_delivery_view, gitea.gitea_commit_status, gitea.gitea_commit_statuses, gitea.gitea_contains, gitea.gitea_open_pulls, keeps state deliveries, groups
long-running resources without health: 1
files asking for a setting without saying whether it is trusted: 0
2 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 2026-10-12T15:47:39Z 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 2026-10-12T15:47:39Z
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 2 module(s) in 1 tier(s): gitea, mesh-delivery
notes:
- modules/mesh-delivery/cmd/mesh-delivery/covered_test.go, modules/mesh-delivery/cmd/mesh-delivery/issue341_test.go and modules/mesh-delivery/cmd/mesh-delivery/table_test.go read by no module's build, and rebuild nothing (issue 280)
--- its merge-check.sh (go toolchain)
ok git.novox.be/novox/mesh-catalog/checks/words 0.002s
words: 12 retired words and 70 daemons' lines, none in the 44405 strings of 701 files
ok git.novox.be/novox/mesh-catalog/checks/agenthome 0.011s
agent-home: no archive and no file written into under the agent's home
ok git.novox.be/novox/mesh-catalog/checks/paths 0.005s
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 0.280s
testing modules/gitea
ok gitea/cmd/forge 1.426s
ok gitea/cmd/npm-registry 1.736s
? gitea/internal/forgehttp [no test files]
testing modules/mesh-delivery
ok mesh-delivery/cmd/mesh-delivery 2.964s
--- 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 (11.112343ms)
✔ every open pull request is read with what its base requires and what its head has (gitea_open_pulls) (51.319649ms)
✔ a pull request or issue number that is not one is refused before the forge is asked (1.317982ms)
✔ the newest status of each context is kept, with what it said, when and by whom (23.14552ms)
✔ a commit merges only when every required status is a success; a warning blocks (0.693065ms)
✔ no rule requires nothing; a rule that cannot be read says nothing, never yes (0.305786ms)
✔ a commit is contained in another when the compare finds nothing on it the other lacks (0.857905ms)
✔ a note line is appended once, as the forge's user, in its own repository (10.216616ms)
✔ a note that cannot be written is said, never read as written (0.528125ms)
✔ one view per pull request, found by its marker; only the mesh's statuses (1.277034ms)
✔ a build machine taking the check is said on the head: building on which machine, since the local time (20.550674ms)
✔ the verdict heard while a 'building' write is in flight lands last and stays (9.07092ms)
[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 (1.130389ms)
✔ the verdict heard while a 'building' write still waits its turn: the 'building' is never written (0.563822ms)
✔ a start heard after the verdict writes nothing (0.569934ms)
✔ a start from before the check now pending (a run before a recheck, the stream's history) writes nothing (1.472172ms)
✔ an old run's verdict redelivered after a recheck's pending is not written over it (1.021313ms)
[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 (2.265776ms)
✔ a rechecked head says where its new run is: building on which machine (0.925934ms)
✔ the recheck's own verdict is written, even asked a little before the pending by another machine's clock (0.753759ms)
✔ after a restart, an old run's verdict is weighed against the check the forge's history holds open (0.623884ms)
✔ a recheck the forge does not take fails its caller, and the head is as it was (0.559864ms)
✔ the status tool writes the merge check's pending through the writer, and the mesh's other statuses as they were (4.954488ms)
✔ when a check was asked is read from its id; the opening of the check now pending from the gate's history (0.256974ms)
✔ a head this process did not announce (it restarted) is said only while the forge holds its gate pending (0.348907ms)
✔ a repository as the mesh clones it is told from another (0.251093ms)
✔ gitea_api refuses the forge's administration, a branch's protection and the merge check's status, and sends nothing (4.479083ms)
✔ gitea_api refuses a file written to the default branch or one a rule guards, and writes one on a branch of its own (49.952375ms)
✔ a protection rule covers a branch as the forge decides it: classes, alternatives, a plain name in any case (issue 539) (15.957566ms)
✔ gitea_api still writes what an author writes (7.893523ms)
✔ gitea_branch_protection_set only strengthens (14.556618ms)
✔ gitea_merge_pull_request merges only a head whose required statuses all succeeded, and only that head (39.624078ms)
✔ gitea_merge_pull_request refuses a verdict judged against an earlier trunk, and asks the check once (140.143145ms)
✔ gitea_merge_pull_request merges only the head it is named, and refuses one that is not the head (14.017263ms)
✔ gitea_merge_pull_request is loud when the forge merged a head other than the one it was handed (14.007064ms)
✔ gitea_merge_pull_request says the merge happened when reading back what landed fails (24.455908ms)
✔ gitea_merge_pull_request reads back a fast-forward, and says it cannot read back a squash or a rebase (56.197103ms)
✔ gitea_delete_repo deletes only the tools' own user's repository, unguarded, untouched for 30 days, named in full (20.499269ms)
✔ gitea_api sends the path it judged, and refuses moving a pull request's base (3.970696ms)
✔ gitea_merge_pull_request merges only into a branch whose protection requires the checks (6.496261ms)
✔ a rule named for a branch is not made weaker than the glob rule that guards it now (15.243552ms)
✔ a repository owner or name is never another path (0.780049ms)
✔ the judges, without a forge (0.374044ms)
✔ gitea_api never acts as another user: sudo, in any case, and any query key it does not send are refused (1.850794ms)
✔ gitea_api never approves a pull request nor requests changes: those are the operator's (2.927542ms)
✔ no tool's caller text becomes another path or another user: `..` and `?sudo=` are refused before the forge (5.317963ms)
✔ every page of a pull request's files is read, though the forge caps a page at fifty (24.667236ms)
✔ a pass asks only the repositories that moved since the last look, every one before the first (0.364797ms)
✔ the directories above a merge's files that hold a module at the commit are said, the root never (10.966088ms)
✔ a lookup the forge refuses is thrown, not read as no module (3.29784ms)
✔ a renamed file is said under its old name too, as gone (novox/hq ADR 0267) (4.563102ms)
✔ the statuses asked for replace the rule's, and an empty list requires none (5.526917ms)
✔ a new rule refuses direct pushes unless asked; an administrator's override only when said (0.359927ms)
✔ each open pull request's new head is announced once, and nothing closed or merged (12.795984ms)
✔ a verdict is the head commit's status; an error is the forge's error, never a success (1.027014ms)
✔ each layer is its own status: the gate with the modules it judged, the repository's own check beside it (0.732218ms)
✔ no verdict sets warning on a required check; only a person's decision fails (1.023517ms)
✔ a commit status is set on the commit, under the merge check's context (34.14956ms)
✔ a change plan is the gate's result: said on the status and, when it builds something, as a comment (0.527133ms)
✔ a head announced is pending in the build queue, since the local time it was announced (0.232849ms)
✔ first start: mints with the admin account, keeps the token at 0600, asks for the tools' scopes only (32.99151ms)
✔ second start: reuses the kept token, mints nothing (13.941661ms)
✔ the forge rejects the kept token (its data was restored): minted afresh, once, and the call goes through (19.007639ms)
✔ 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 (22.745237ms)
✔ 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 (17.643ms)
✔ the kept file is gone but the forge still holds a token by that name: replaced, not refused (16.335658ms)
✔ concurrent first calls share one mint (10.683456ms)
✔ the admin account is refused: said plainly, nothing kept, and the next call fails the same way rather than crashing (13.356558ms)
✔ one process shares one source per kept file — the watcher and the tools never renew against each other (1.134246ms)
✔ a second process finds the token the first renewed, and reuses it instead of minting over it (24.437874ms)
✔ a configured token wins, and is reported rather than minted over when the forge rejects it (3.012823ms)
✔ nothing to mint with and no token: the client says what is missing (0.248639ms)
[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-XXF2gQ/state/token
✔ the tools register once there is a way to a token, and the first call mints it (8.379698ms)
[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-kdrJs0/state/token
✔ a pull request can be closed, read and cleaned up through the tools (19.738956ms)
[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 (1.545771ms)
✔ a trunk's move puts an open head's merge check out of date, naming both commits, unless a check is on its way (0.980336ms)
✔ a push to a trunk puts every open pull request into it out of date — twenty-five of them — and none into another (4.750775ms)
✔ a trunk that cannot be read is weighed again on the next pass, never recorded half (0.388221ms)
✔ the trunk's move takes its turn after a verdict in flight, and weighs what the forge holds then (0.432223ms)
✔ a verdict judged against a trunk newer than the pass's first read is not put out of date (0.726518ms)
✔ a verdict that landed out of date keeps its account on the pull request (0.926074ms)
ℹ tests 81
ℹ suites 0
ℹ pass 81
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 606.783472
```
mesh/merge-gate: PASS — every machine composes with the change as it did without (4 of 4 compose) (modules: gitea, mesh-delivery)
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, mesh-delivery → novox; no bus step
tier 0: gitea, mesh-delivery
novox: receives gitea, mesh-delivery
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
read by no module's build: modules/mesh-delivery/cmd/mesh-delivery/covered_test.go, modules/mesh-delivery/cmd/mesh-delivery/issue341_test.go, modules/mesh-delivery/cmd/mesh-delivery/table_test.go
Run by the build seat on shanks as build-1791820355799476047 (builds --log build-1791820355799476047).
--- the gate: gitea, mesh-delivery
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)
mesh-delivery: ok, 1 tool(s), invokes seat:mesh-controller.delivery-plan, seat:mesh-controller.delivery-order, seat:mesh-controller.delivery-check, seat:mesh-controller.deliver, seat:mesh-controller.delivery-stop, seat:mesh-controller.delivery-walks, gitea.gitea_note_append, gitea.gitea_delivery_view, gitea.gitea_commit_status, gitea.gitea_commit_statuses, gitea.gitea_contains, gitea.gitea_open_pulls, keeps state deliveries, groups
long-running resources without health: 1
files asking for a setting without saying whether it is trusted: 0
2 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 2026-10-12T15:47:39Z 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 2026-10-12T15:47:39Z
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 2 module(s) in 1 tier(s): gitea, mesh-delivery
notes:
- modules/mesh-delivery/cmd/mesh-delivery/covered_test.go, modules/mesh-delivery/cmd/mesh-delivery/issue341_test.go and modules/mesh-delivery/cmd/mesh-delivery/table_test.go read by no module's build, and rebuild nothing (issue 280)
--- 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 44409 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 0.215s
testing modules/gitea
ok gitea/cmd/forge 1.500s
ok gitea/cmd/npm-registry 1.918s
? gitea/internal/forgehttp [no test files]
testing modules/mesh-delivery
ok mesh-delivery/cmd/mesh-delivery 3.036s
--- 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 (9.463686ms)
✔ every open pull request is read with what its base requires and what its head has (gitea_open_pulls) (47.046103ms)
✔ a pull request or issue number that is not one is refused before the forge is asked (1.313851ms)
✔ the newest status of each context is kept, with what it said, when and by whom (20.13084ms)
✔ a commit merges only when every required status is a success; a warning blocks (0.729832ms)
✔ no rule requires nothing; a rule that cannot be read says nothing, never yes (0.324535ms)
✔ a commit is contained in another when the compare finds nothing on it the other lacks (1.081848ms)
✔ a note line is appended once, as the forge's user, in its own repository (11.330386ms)
✔ a note that cannot be written is said, never read as written (0.370056ms)
✔ one view per pull request, found by its marker; only the mesh's statuses (0.59188ms)
✔ a build machine taking the check is said on the head: building on which machine, since the local time (11.0132ms)
✔ the verdict heard while a 'building' write is in flight lands last and stays (4.88876ms)
[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 (0.766572ms)
✔ the verdict heard while a 'building' write still waits its turn: the 'building' is never written (0.461987ms)
✔ a start heard after the verdict writes nothing (0.506569ms)
✔ a start from before the check now pending (a run before a recheck, the stream's history) writes nothing (0.563589ms)
✔ an old run's verdict redelivered after a recheck's pending is not written over it (0.769132ms)
[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 (1.599566ms)
✔ a rechecked head says where its new run is: building on which machine (0.700371ms)
✔ the recheck's own verdict is written, even asked a little before the pending by another machine's clock (0.58932ms)
✔ after a restart, an old run's verdict is weighed against the check the forge's history holds open (0.501928ms)
✔ a recheck the forge does not take fails its caller, and the head is as it was (0.580619ms)
✔ the status tool writes the merge check's pending through the writer, and the mesh's other statuses as they were (4.743348ms)
✔ when a check was asked is read from its id; the opening of the check now pending from the gate's history (0.213864ms)
✔ a head this process did not announce (it restarted) is said only while the forge holds its gate pending (0.332786ms)
✔ a repository as the mesh clones it is told from another (0.234524ms)
✔ gitea_api refuses the forge's administration, a branch's protection and the merge check's status, and sends nothing (5.210455ms)
✔ gitea_api refuses a file written to the default branch or one a rule guards, and writes one on a branch of its own (49.186898ms)
✔ a protection rule covers a branch as the forge decides it: classes, alternatives, a plain name in any case (issue 539) (20.368684ms)
✔ gitea_api still writes what an author writes (8.160514ms)
✔ gitea_branch_protection_set only strengthens (14.698801ms)
✔ gitea_merge_pull_request merges only a head whose required statuses all succeeded, and only that head (43.526474ms)
✔ gitea_merge_pull_request refuses a verdict judged against an earlier trunk, and asks the check once (142.400327ms)
✔ gitea_merge_pull_request merges only the head it is named, and refuses one that is not the head (18.228829ms)
✔ gitea_merge_pull_request is loud when the forge merged a head other than the one it was handed (13.527932ms)
✔ gitea_merge_pull_request says the merge happened when reading back what landed fails (24.168456ms)
✔ gitea_merge_pull_request reads back a fast-forward, and says it cannot read back a squash or a rebase (52.460301ms)
✔ gitea_delete_repo deletes only the tools' own user's repository, unguarded, untouched for 30 days, named in full (22.53929ms)
✔ gitea_api sends the path it judged, and refuses moving a pull request's base (4.28968ms)
✔ gitea_merge_pull_request merges only into a branch whose protection requires the checks (6.72678ms)
✔ a rule named for a branch is not made weaker than the glob rule that guards it now (13.410821ms)
✔ a repository owner or name is never another path (0.219424ms)
✔ the judges, without a forge (0.363526ms)
✔ gitea_api never acts as another user: sudo, in any case, and any query key it does not send are refused (2.41278ms)
✔ gitea_api never approves a pull request nor requests changes: those are the operator's (2.173466ms)
✔ no tool's caller text becomes another path or another user: `..` and `?sudo=` are refused before the forge (3.70636ms)
✔ every page of a pull request's files is read, though the forge caps a page at fifty (25.191173ms)
✔ a pass asks only the repositories that moved since the last look, every one before the first (0.441047ms)
✔ the directories above a merge's files that hold a module at the commit are said, the root never (13.798226ms)
✔ a lookup the forge refuses is thrown, not read as no module (4.827119ms)
✔ a renamed file is said under its old name too, as gone (novox/hq ADR 0267) (5.921857ms)
✔ the statuses asked for replace the rule's, and an empty list requires none (6.852523ms)
✔ a new rule refuses direct pushes unless asked; an administrator's override only when said (0.289194ms)
✔ each open pull request's new head is announced once, and nothing closed or merged (9.614118ms)
✔ a verdict is the head commit's status; an error is the forge's error, never a success (1.037497ms)
✔ each layer is its own status: the gate with the modules it judged, the repository's own check beside it (0.753912ms)
✔ no verdict sets warning on a required check; only a person's decision fails (1.162449ms)
✔ a commit status is set on the commit, under the merge check's context (36.942757ms)
✔ a change plan is the gate's result: said on the status and, when it builds something, as a comment (0.475017ms)
✔ a head announced is pending in the build queue, since the local time it was announced (0.250854ms)
✔ first start: mints with the admin account, keeps the token at 0600, asks for the tools' scopes only (29.521974ms)
✔ second start: reuses the kept token, mints nothing (15.676967ms)
✔ the forge rejects the kept token (its data was restored): minted afresh, once, and the call goes through (18.974142ms)
✔ 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 (28.156212ms)
✔ 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 (20.641649ms)
✔ the kept file is gone but the forge still holds a token by that name: replaced, not refused (19.420089ms)
✔ concurrent first calls share one mint (10.564123ms)
✔ the admin account is refused: said plainly, nothing kept, and the next call fails the same way rather than crashing (12.265732ms)
✔ one process shares one source per kept file — the watcher and the tools never renew against each other (0.893284ms)
✔ a second process finds the token the first renewed, and reuses it instead of minting over it (19.661513ms)
✔ a configured token wins, and is reported rather than minted over when the forge rejects it (2.651384ms)
✔ nothing to mint with and no token: the client says what is missing (0.211723ms)
[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-tEfcNI/state/token
✔ the tools register once there is a way to a token, and the first call mints it (7.679085ms)
[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-Anoju5/state/token
✔ a pull request can be closed, read and cleaned up through the tools (20.680459ms)
[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 (1.495875ms)
✔ a trunk's move puts an open head's merge check out of date, naming both commits, unless a check is on its way (0.983026ms)
✔ a push to a trunk puts every open pull request into it out of date — twenty-five of them — and none into another (4.532595ms)
✔ a trunk that cannot be read is weighed again on the next pass, never recorded half (0.342846ms)
✔ the trunk's move takes its turn after a verdict in flight, and weighs what the forge holds then (0.443428ms)
✔ a verdict judged against a trunk newer than the pass's first read is not put out of date (0.235484ms)
✔ a verdict that landed out of date keeps its account on the pull request (0.272954ms)
ℹ tests 81
ℹ suites 0
ℹ pass 81
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 609.835087
**Merge check** at `41b6e100`
- `mesh/merge-gate`: **PASS** — every machine composes with the change as it did without (4 of 4 compose) (modules: gitea, mesh-delivery)
- `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, mesh-delivery → novox; no bus step
- tier 0: gitea, mesh-delivery
- novox: receives gitea, mesh-delivery
- 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
- read by no module's build: modules/mesh-delivery/cmd/mesh-delivery/covered_test.go, modules/mesh-delivery/cmd/mesh-delivery/issue341_test.go, modules/mesh-delivery/cmd/mesh-delivery/table_test.go
Run by the build seat on shanks as `build-1791820355799476047` (`builds --log build-1791820355799476047`).
```
--- the gate: gitea, mesh-delivery
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)
mesh-delivery: ok, 1 tool(s), invokes seat:mesh-controller.delivery-plan, seat:mesh-controller.delivery-order, seat:mesh-controller.delivery-check, seat:mesh-controller.deliver, seat:mesh-controller.delivery-stop, seat:mesh-controller.delivery-walks, gitea.gitea_note_append, gitea.gitea_delivery_view, gitea.gitea_commit_status, gitea.gitea_commit_statuses, gitea.gitea_contains, gitea.gitea_open_pulls, keeps state deliveries, groups
long-running resources without health: 1
files asking for a setting without saying whether it is trusted: 0
2 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 2026-10-12T15:47:39Z 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 2026-10-12T15:47:39Z
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 2 module(s) in 1 tier(s): gitea, mesh-delivery
notes:
- modules/mesh-delivery/cmd/mesh-delivery/covered_test.go, modules/mesh-delivery/cmd/mesh-delivery/issue341_test.go and modules/mesh-delivery/cmd/mesh-delivery/table_test.go read by no module's build, and rebuild nothing (issue 280)
--- 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 44409 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 0.215s
testing modules/gitea
ok gitea/cmd/forge 1.500s
ok gitea/cmd/npm-registry 1.918s
? gitea/internal/forgehttp [no test files]
testing modules/mesh-delivery
ok mesh-delivery/cmd/mesh-delivery 3.036s
--- 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 (9.463686ms)
✔ every open pull request is read with what its base requires and what its head has (gitea_open_pulls) (47.046103ms)
✔ a pull request or issue number that is not one is refused before the forge is asked (1.313851ms)
✔ the newest status of each context is kept, with what it said, when and by whom (20.13084ms)
✔ a commit merges only when every required status is a success; a warning blocks (0.729832ms)
✔ no rule requires nothing; a rule that cannot be read says nothing, never yes (0.324535ms)
✔ a commit is contained in another when the compare finds nothing on it the other lacks (1.081848ms)
✔ a note line is appended once, as the forge's user, in its own repository (11.330386ms)
✔ a note that cannot be written is said, never read as written (0.370056ms)
✔ one view per pull request, found by its marker; only the mesh's statuses (0.59188ms)
✔ a build machine taking the check is said on the head: building on which machine, since the local time (11.0132ms)
✔ the verdict heard while a 'building' write is in flight lands last and stays (4.88876ms)
[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 (0.766572ms)
✔ the verdict heard while a 'building' write still waits its turn: the 'building' is never written (0.461987ms)
✔ a start heard after the verdict writes nothing (0.506569ms)
✔ a start from before the check now pending (a run before a recheck, the stream's history) writes nothing (0.563589ms)
✔ an old run's verdict redelivered after a recheck's pending is not written over it (0.769132ms)
[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 (1.599566ms)
✔ a rechecked head says where its new run is: building on which machine (0.700371ms)
✔ the recheck's own verdict is written, even asked a little before the pending by another machine's clock (0.58932ms)
✔ after a restart, an old run's verdict is weighed against the check the forge's history holds open (0.501928ms)
✔ a recheck the forge does not take fails its caller, and the head is as it was (0.580619ms)
✔ the status tool writes the merge check's pending through the writer, and the mesh's other statuses as they were (4.743348ms)
✔ when a check was asked is read from its id; the opening of the check now pending from the gate's history (0.213864ms)
✔ a head this process did not announce (it restarted) is said only while the forge holds its gate pending (0.332786ms)
✔ a repository as the mesh clones it is told from another (0.234524ms)
✔ gitea_api refuses the forge's administration, a branch's protection and the merge check's status, and sends nothing (5.210455ms)
✔ gitea_api refuses a file written to the default branch or one a rule guards, and writes one on a branch of its own (49.186898ms)
✔ a protection rule covers a branch as the forge decides it: classes, alternatives, a plain name in any case (issue 539) (20.368684ms)
✔ gitea_api still writes what an author writes (8.160514ms)
✔ gitea_branch_protection_set only strengthens (14.698801ms)
✔ gitea_merge_pull_request merges only a head whose required statuses all succeeded, and only that head (43.526474ms)
✔ gitea_merge_pull_request refuses a verdict judged against an earlier trunk, and asks the check once (142.400327ms)
✔ gitea_merge_pull_request merges only the head it is named, and refuses one that is not the head (18.228829ms)
✔ gitea_merge_pull_request is loud when the forge merged a head other than the one it was handed (13.527932ms)
✔ gitea_merge_pull_request says the merge happened when reading back what landed fails (24.168456ms)
✔ gitea_merge_pull_request reads back a fast-forward, and says it cannot read back a squash or a rebase (52.460301ms)
✔ gitea_delete_repo deletes only the tools' own user's repository, unguarded, untouched for 30 days, named in full (22.53929ms)
✔ gitea_api sends the path it judged, and refuses moving a pull request's base (4.28968ms)
✔ gitea_merge_pull_request merges only into a branch whose protection requires the checks (6.72678ms)
✔ a rule named for a branch is not made weaker than the glob rule that guards it now (13.410821ms)
✔ a repository owner or name is never another path (0.219424ms)
✔ the judges, without a forge (0.363526ms)
✔ gitea_api never acts as another user: sudo, in any case, and any query key it does not send are refused (2.41278ms)
✔ gitea_api never approves a pull request nor requests changes: those are the operator's (2.173466ms)
✔ no tool's caller text becomes another path or another user: `..` and `?sudo=` are refused before the forge (3.70636ms)
✔ every page of a pull request's files is read, though the forge caps a page at fifty (25.191173ms)
✔ a pass asks only the repositories that moved since the last look, every one before the first (0.441047ms)
✔ the directories above a merge's files that hold a module at the commit are said, the root never (13.798226ms)
✔ a lookup the forge refuses is thrown, not read as no module (4.827119ms)
✔ a renamed file is said under its old name too, as gone (novox/hq ADR 0267) (5.921857ms)
✔ the statuses asked for replace the rule's, and an empty list requires none (6.852523ms)
✔ a new rule refuses direct pushes unless asked; an administrator's override only when said (0.289194ms)
✔ each open pull request's new head is announced once, and nothing closed or merged (9.614118ms)
✔ a verdict is the head commit's status; an error is the forge's error, never a success (1.037497ms)
✔ each layer is its own status: the gate with the modules it judged, the repository's own check beside it (0.753912ms)
✔ no verdict sets warning on a required check; only a person's decision fails (1.162449ms)
✔ a commit status is set on the commit, under the merge check's context (36.942757ms)
✔ a change plan is the gate's result: said on the status and, when it builds something, as a comment (0.475017ms)
✔ a head announced is pending in the build queue, since the local time it was announced (0.250854ms)
✔ first start: mints with the admin account, keeps the token at 0600, asks for the tools' scopes only (29.521974ms)
✔ second start: reuses the kept token, mints nothing (15.676967ms)
✔ the forge rejects the kept token (its data was restored): minted afresh, once, and the call goes through (18.974142ms)
✔ 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 (28.156212ms)
✔ 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 (20.641649ms)
✔ the kept file is gone but the forge still holds a token by that name: replaced, not refused (19.420089ms)
✔ concurrent first calls share one mint (10.564123ms)
✔ the admin account is refused: said plainly, nothing kept, and the next call fails the same way rather than crashing (12.265732ms)
✔ one process shares one source per kept file — the watcher and the tools never renew against each other (0.893284ms)
✔ a second process finds the token the first renewed, and reuses it instead of minting over it (19.661513ms)
✔ a configured token wins, and is reported rather than minted over when the forge rejects it (2.651384ms)
✔ nothing to mint with and no token: the client says what is missing (0.211723ms)
[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-tEfcNI/state/token
✔ the tools register once there is a way to a token, and the first call mints it (7.679085ms)
[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-Anoju5/state/token
✔ a pull request can be closed, read and cleaned up through the tools (20.680459ms)
[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 (1.495875ms)
✔ a trunk's move puts an open head's merge check out of date, naming both commits, unless a check is on its way (0.983026ms)
✔ a push to a trunk puts every open pull request into it out of date — twenty-five of them — and none into another (4.532595ms)
✔ a trunk that cannot be read is weighed again on the next pass, never recorded half (0.342846ms)
✔ the trunk's move takes its turn after a verdict in flight, and weighs what the forge holds then (0.443428ms)
✔ a verdict judged against a trunk newer than the pass's first read is not put out of date (0.235484ms)
✔ a verdict that landed out of date keeps its account on the pull request (0.272954ms)
ℹ tests 81
ℹ suites 0
ℹ pass 81
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 609.835087
```
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.
Builds the forge's and mesh-delivery's part of novox/hq ADR 0323 (issue 341). Issue 406's rebase symptom is fixed with it. The controller's part is mesh-controller#271: there the build seat judges the head merged into the trunk's head and the verdict names the trunk commit.
Merge after mesh-controller#271 is delivered to the controller and every build seat holder (see Rollout).
gitea (TypeScript)
trunk.ts(new): what a verdict judged, in words (head … merged into trunk …). Also the out-of-date words, which statuses a trunk's move resets, the merge tool's comparison, the comment of a verdict that landed out of date, the record of how long a repository's last check took, and thepull.updatedannouncement, shared by the watcher and the merge tool. The announcement now also carriesbase_sha, the base branch's head; the controller names it as the trunk commit of a verdict no build judges.pulls.ts: both statuses' descriptions begin withhead <12> merged into trunk <12> —when the verdict names a trunk, so the 140-character clip never drops it. The comment says what was judged and which commits were read beside it.index.ts: on each pass, every trunk an open pull request merges into is read. The open pull requests are paged up to 1000, and the trunk heads are kept intrunks-seen.json. When a trunk moved,mesh/merge-gateandmesh/repo-checkof every open pull request into it go back to pending with "out of date: judged against trunk X, the trunk is now Y; …". The trunk is read again in each head's own write turn, so a verdict judged against a newer head is left alone. Nothing runs. A verdict that lands judged against a trunk that has moved since is written as out of date, and its comment is still posted. Each verdict'stookis kept for the merge tool.merge.ts(new):gitea_merge_pull_request, moved here fromtools/index.tsand served from the watcher's bundle beside the merge check's one writer (issue 456). It reads the trunk commit the head's verdict names and the trunk's head. When they differ, nothing is merged and both commits are named. The statuses are set pending through the one writer ("waiting in the build queue … asked by the merge tool"), so "building on …" is said and no duplicate check is asked. The head is announced once, and the refusal says how long the last check of the repository took. It does not ask again while a check is on its way, puts the out-of-date words back if the announcement fails, and never merges by itself. The refusal namesmesh-delivery.recheckfor a check that never answers, and says that in a busy repository it may be refused and asked once more. A verdict that names no trunk commit is refused the same way.gate.ts:outOfDatetakes the head's turn on the one writer's chain and decides from what the forge holds at that moment. A verdict still in flight lands first.client.ts:listOpenPulls(paged),branchHead,getRepo.mesh-delivery (Go)
ready→proposedontrunk moved. The guard is "its verdict names a trunk commit that is not its trunk's head", and the reason carries both commits.proposed→checkedis now "a verdict names this commit and the trunk's head". A late verdict for an old trunk is kept and shown but does not count, so the table reaches a fixed point.rejecteddelivery staysrejected. A verdict for it judged against another trunk is taken as its recheck.proposeddelivery has no bound and is not listed bystalled. Its one-hour bound starts when its head is announced again (the merge tool's ask). A forge status that reads "out of date:" is not saidunanswered.checksgivesjudgedandout_of_date,showgives the verdict whole, and the kept plan recordstrunk.checkingwith the reason, andmesh/delivery-groupgoes pending on every member's head. It runs again once a member is judged again. A member rebased or pushed after the group closed takes its old head's place. The composed check is asked again for the new heads, counting members already merged or delivered, the group readschecking, and nobody is held.Tests (each fails on main)
test/trunk.test.ts: the status words; 25 open pull requests intomainall put out of date, one intoreleaseleft; a first sight puts nothing out of date; a trunk that cannot be weighed is not recorded; a verdict in flight is weighed in its turn; a verdict judged against a trunk newer than the pass's first read is left; a late verdict keeps its account.test/guard.test.ts: the merge tool refuses naming both commits, asks once withbase_shain the announcement, writes its pending through the one writer, says how long the last check took and namesmesh-delivery.recheck. It restores the out-of-date words when the announcement fails, and merges when the verdict names the trunk's head.test/token.test.tslists the tools' bundle without the merge tool.issue341_test.go: the new row, rejected stays,stalled, the late verdict's fixed point,checks/show/plan, a non-member's merge against a member's own, the rebased member (also after another member was delivered), and the bound from the ask.table_test.gowalks the new row.merge-check.sh(gitea and mesh-delivery, race) andmerge-check-typescript.sh(81 of 81), against real copies of node_modules.Rollout
ready, until each is judged again. The merge tool asks one check for each of them at its merge.Not here
The forge's holder does not write
mesh/delivery-group. mesh-delivery is its one writer, and a second writer would race it (issue 446). The group's out-of-date state is mesh-delivery's, and is said through that status.Delivery
novox/mesh-catalog@41b6e10077ef— delivered since 2026-10-12T16:15:24ZDelivery plan — builds gitea, mesh-delivery → novox; no bus step
Group
feat/341-a-merge-check-judges-the-merge, in order: novox/hq@09916e2bbc → novox/mesh-controller@875ebc093e → novox/mesh-catalog@41b6e10077 → novox/mesh-lab@79108a72a0Machines
Transitions
The commit's note under
refs/notes/mesh-plankeeps every transition:git log --notes=mesh-plan.Merge check at
82cca650mesh/merge-gate: PASS — every machine composes with the change as it did without (4 of 4 compose) (modules: gitea, mesh-delivery)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, mesh-delivery → novox; no bus step
Run by the build seat on ace as
build-1791819071374743686(builds --log build-1791819071374743686).Merge check at
434506f9mesh/merge-gate: PASS — every machine composes with the change as it did without (4 of 4 compose) (modules: gitea, mesh-delivery)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, mesh-delivery → novox; no bus step
Run by the build seat on g14 as
build-1791820016890812212(builds --log build-1791820016890812212).Merge check at
41b6e100mesh/merge-gate: PASS — every machine composes with the change as it did without (4 of 4 compose) (modules: gitea, mesh-delivery)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, mesh-delivery → novox; no bus step
Run by the build seat on shanks as
build-1791820355799476047(builds --log build-1791820355799476047).