Fixes novox/hq issue 445 (a group member superseded by a walk that later carries it still stops the members after it).
Acceptance: in a delivery group [A, B] where A's walk was superseded by a later merge C of A's repository, B stays published while C can still carry A, takes its turn once C's walk is delivered and covers A, and is stopped with a reason naming A only when no later merge can carry it.
What changes
groupTurns (holder.go): a superseded member with CoveredBy set counts as reached, as GroupState already does since #197 (ADR 0270 decision 2: a group counts a covered member as delivered). A superseded member not yet covered holds the members after it while carryAwaited names a later merge that can still carry it; otherwise they are stopped with its group G stopped: A superseded, and no later merge into main can carry M: <why>. Failed and stopped members stop the rest as before.
carryAwaited / mayCarry: a candidate carrier is a later merge of the same repository into the same trunk, on a passing check, whose plan moves the module, and which the forge has not said leaves A out. Every module A moves needs one.
What bounds the wait
An open carrier counts only while it is not held for a person and is within its own state's bound (Bounds: published 30 min, delivering 2 h; a merged delivery with no walk is held after 10 min). Past that bound it no longer counts and B is stopped, naming the bound (ADR 0239 decision 9, ADR 0276 decision 9).
A delivered carrier whose containment the forge has not answered counts for containsGrace = 30 minutes (the published bound) after it was delivered; then B is stopped saying the forge did not say. Without this, a forge that never answers would hold B for ever. I chose 30 minutes myself; it is a small design limit.
A carrier superseded in turn ends; the later merge that took it over is a carrier of its own, so the wait follows real merges, each bounded.
Follows from decisions, no new one
ADR 0239 decision 4 stops the group for a failed or stopped member and otherwise lets the next start when the one before is delivered; superseded is not named as a stop. ADR 0270 decision 2 counts a covered member as delivered so the member after it gets its turn. ADR 0276 says every wait has a bound. The wait is a turn waiting on a cover the mesh already computes.
Tests
issue445_test.go, the live shape of group fix/assign-says-why-a-module-is-not-there used by #197's tests: (a) a later walk open: B waits; (b) that walk delivered and covering A: B is let go; (c) the forge says no, or the later walk failed: B is stopped saying why; (d) the later walk past its own bound, and a forge that never answers past containsGrace: B is stopped. All six subtests failed on main (B stopped at once, or stopped without a reason). go vet ./..., go test -race ./... -count=1 in modules/mesh-delivery, and the words check all pass. Does not touch verbs.go, waiting_checks.go (#201) or modules/restic (#200).
Fixes novox/hq issue 445 (a group member superseded by a walk that later carries it still stops the members after it).
**Acceptance:** in a delivery group [A, B] where A's walk was superseded by a later merge C of A's repository, B stays published while C can still carry A, takes its turn once C's walk is delivered and covers A, and is stopped with a reason naming A only when no later merge can carry it.
## What changes
- `groupTurns` (holder.go): a superseded member with `CoveredBy` set counts as reached, as `GroupState` already does since #197 (ADR 0270 decision 2: a group counts a covered member as delivered). A superseded member not yet covered holds the members after it while `carryAwaited` names a later merge that can still carry it; otherwise they are stopped with `its group G stopped: A superseded, and no later merge into main can carry M: <why>`. Failed and stopped members stop the rest as before.
- `carryAwaited` / `mayCarry`: a candidate carrier is a later merge of the same repository into the same trunk, on a passing check, whose plan moves the module, and which the forge has not said leaves A out. Every module A moves needs one.
## What bounds the wait
- An open carrier counts only while it is not held for a person and is within its own state's bound (`Bounds`: published 30 min, delivering 2 h; a merged delivery with no walk is held after 10 min). Past that bound it no longer counts and B is stopped, naming the bound (ADR 0239 decision 9, ADR 0276 decision 9).
- A delivered carrier whose containment the forge has not answered counts for `containsGrace` = 30 minutes (the published bound) after it was delivered; then B is stopped saying the forge did not say. Without this, a forge that never answers would hold B for ever. I chose 30 minutes myself; it is a small design limit.
- A carrier superseded in turn ends; the later merge that took it over is a carrier of its own, so the wait follows real merges, each bounded.
## Follows from decisions, no new one
ADR 0239 decision 4 stops the group for a failed or stopped member and otherwise lets the next start when the one before is delivered; superseded is not named as a stop. ADR 0270 decision 2 counts a covered member as delivered so the member after it gets its turn. ADR 0276 says every wait has a bound. The wait is a turn waiting on a cover the mesh already computes.
## Tests
`issue445_test.go`, the live shape of group fix/assign-says-why-a-module-is-not-there used by #197's tests: (a) a later walk open: B waits; (b) that walk delivered and covering A: B is let go; (c) the forge says no, or the later walk failed: B is stopped saying why; (d) the later walk past its own bound, and a forge that never answers past `containsGrace`: B is stopped. All six subtests failed on main (B stopped at once, or stopped without a reason). `go vet ./...`, `go test -race ./... -count=1` in modules/mesh-delivery, and the words check all pass. Does not touch verbs.go, waiting_checks.go (#201) or modules/restic (#200).
Superseded is no failure: the later merge that took its walk over may still
deliver its change, and a covered member counts as delivered (ADR 0270). So the
members after it wait while such a merge is open within its own bound, take
their turn once its delivered walk covers the member, and are stopped, saying
why, only when none can carry it (novox/hq issue 445).
Deliverynovox/mesh-catalog@15a24fb63ec0 — delivered since 2026-10-11T02:26:03Z
Delivery plan — builds mesh-delivery → novox; no bus step
tier 0: mesh-delivery
novox: receives mesh-delivery
Machines
mesh-delivery on novox: passed — carried by release-1791685274961255927
mesh-controller on novox: passed — healthy 3 times over 3m38s
mesh-controller on the rest running it: sent
mesh-delivery on the rest running it: sent
Transitions
2026-10-11 01:47 (new) → proposed (announced): novox/mesh-catalog#203's head announced
2026-10-11 01:48 proposed → checked (checked): the verdict names this commit
2026-10-11 01:48 checked → ready (accepted): the gate passed or warned, and the repository's own check did not fail
2026-10-11 02:20 ready → published (merged): on the trunk its modules follow: merged there, and its walk opened — or nothing for a walk to move
2026-10-11 02:20 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
2026-10-11 02:26 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@15a24fb63ec0` — **delivered** since 2026-10-11T02:26:03Z
**Delivery plan** — builds mesh-delivery → novox; no bus step
- tier 0: mesh-delivery
- novox: receives mesh-delivery
**Machines**
- mesh-delivery on novox: passed — carried by release-1791685274961255927
- mesh-controller on novox: passed — healthy 3 times over 3m38s
- mesh-controller on the rest running it: sent
- mesh-delivery on the rest running it: sent
**Transitions**
- 2026-10-11 01:47 (new) → proposed (announced): novox/mesh-catalog#203's head announced
- 2026-10-11 01:48 proposed → checked (checked): the verdict names this commit
- 2026-10-11 01:48 checked → ready (accepted): the gate passed or warned, and the repository's own check did not fail
- 2026-10-11 02:20 ready → published (merged): on the trunk its modules follow: merged there, and its walk opened — or nothing for a walk to move
- 2026-10-11 02:20 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
- 2026-10-11 02:26 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: 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 mesh-delivery → novox; no bus step
tier 0: mesh-delivery
novox: receives mesh-delivery
Run by the build seat on novox as build-1791682479675297115 (builds --log build-1791682479675297115).
--- the gate: mesh-delivery
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: 0
files asking for a setting without saying whether it is trusted: 0
1 manifest(s) checked. Judged against the seats this binary carries; a claim on one of the mesh's own seats is judged fully at registration, and a seat declared by a module not given here reads as unknown. Identities judged on a 6-character machine name, against the bounds of the providers given here
judging 0 module(s) changed against 4 machine(s), as the snapshot of 2026-10-11T01:27:53Z 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 2026-10-11T01:27:53Z
machines:
- a machine (f35): composes
- holds mesh-dns-resolver (idt): composes
- holds git, holds mesh-artifact-store, holds mesh-broker, holds mesh-catalog, holds mesh-delivery, holds mesh-dns-resolver, holds mesh-store, holds mesh-vault, holds npm-package-registry, the control node, the hub (ygcyr): composes
- a machine (yvbpsm): composes
a merge would rebuild 1 module(s) in 1 tier(s): mesh-delivery
notes:
- modules/mesh-delivery/cmd/mesh-delivery/issue445_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 0.006s
words: 12 retired words and 70 daemons' lines, none in the 39472 strings of 655 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 0.006s
paths: no module names another's state on its machine (3 known, listed to go)
long-running resources without health: 69, as kept
testing modules/mesh-delivery
ok mesh-delivery/cmd/mesh-delivery 2.056s
--- its merge-check-typescript.sh (typescript toolchain)
**Merge check** at `51a85c59`
- `mesh/merge-gate`: **PASS** — every machine composes with the change as it did without (4 of 4 compose) (modules: 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 mesh-delivery → novox; no bus step
- tier 0: mesh-delivery
- novox: receives mesh-delivery
Run by the build seat on novox as `build-1791682479675297115` (`builds --log build-1791682479675297115`).
```
--- the gate: mesh-delivery
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: 0
files asking for a setting without saying whether it is trusted: 0
1 manifest(s) checked. Judged against the seats this binary carries; a claim on one of the mesh's own seats is judged fully at registration, and a seat declared by a module not given here reads as unknown. Identities judged on a 6-character machine name, against the bounds of the providers given here
judging 0 module(s) changed against 4 machine(s), as the snapshot of 2026-10-11T01:27:53Z 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 2026-10-11T01:27:53Z
machines:
- a machine (f35): composes
- holds mesh-dns-resolver (idt): composes
- holds git, holds mesh-artifact-store, holds mesh-broker, holds mesh-catalog, holds mesh-delivery, holds mesh-dns-resolver, holds mesh-store, holds mesh-vault, holds npm-package-registry, the control node, the hub (ygcyr): composes
- a machine (yvbpsm): composes
a merge would rebuild 1 module(s) in 1 tier(s): mesh-delivery
notes:
- modules/mesh-delivery/cmd/mesh-delivery/issue445_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 0.006s
words: 12 retired words and 70 daemons' lines, none in the 39472 strings of 655 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 0.006s
paths: no module names another's state on its machine (3 known, listed to go)
long-running resources without health: 69, as kept
testing modules/mesh-delivery
ok mesh-delivery/cmd/mesh-delivery 2.056s
--- its merge-check-typescript.sh (typescript toolchain)
```
Under ADR 0276 the merge that carries a superseded member is often folded
into the next member's own walk, which waits for that member's turn: waiting
there waited for itself until the carrier's bound. The members on that walk
now take their turn, as members sharing one walk do. A member that waits says
which merge it waits for and until when, and `stalled` uses that bound; a
carrier still in a batch is bounded by the window and one walk; a member every
module of which was carried is reached, never stopped with no reason
(novox/hq issue 445, review of #203).
The blocker: a carrier folded into the next member's own walk. When an awaited carrier's walk has not started and is waiting for mesh-delivery's word, the members on that walk now take their turn, the same way members sharing one walk already do. That one walk delivers both, in the group's order among its tiers (ADR 0276 decision 5). A member on another walk still waits. The test TestAMemberWhoseWalkCarriesTheSupersededOneBeforeItTakesItsTurn uses the fold shape (A superseded; B and C published on plan-2, which waits for the word). It ends with plan-2 let go and the group delivered. With the sharing branch switched off, the test fails: B stays published and nothing is let go.
Saying why B waits. A new field on the delivery, turn_awaits and turn_awaits_until, names the superseded member, each carrier with its state and walk, and until when. show shows it, because the field is part of the delivery. boundOf uses it for a published member, so stalled gives that member the carrier's deadline instead of the 30 minutes for a published delivery, and says those words. The field is cleared once the member takes its turn or ends. Tested: B is not listed stalled at 35 minutes, and its bound names C and "until". deliveries still says "its turn in group …", because waitsFor lives in verbs.go, which #201 holds; a one-line follow-up there can read TurnAwaits once #201 lands.
A carrier still assembling in a batch (state ready). It is awaited for readyCarrierBound from its merge: 10 minutes for the merge window at its longest plus 2 hours for the one open walk the batch may queue behind (ADR 0276 decisions 1, 4 and 9). Tested in TestACarrierInABatchIsAwaitedWithinTheWindowAndOneWalk.
The nit.carryAwaited now returns a carryWait. When it names no carrier and gives no reason, every module is carried and contained: the member is reached, never a stop with an empty reason. Tested in TestACarriedMemberNotYetMarkedCoveredIsReached, which calls groupTurns directly; it fails with that branch switched off.
The chain. In TestTheWaitFollowsACarrierSupersededInTurn, C's waiting walk is folded into D's walk, which answers only D. B then waits for D, and goes once D covers A.
go vet ./..., go test -race ./... -count=1 in modules/mesh-delivery, and the words check all pass.
Reworked for the review, now at head f7ba413c.
1. **The blocker: a carrier folded into the next member's own walk.** When an awaited carrier's walk has not started and is waiting for mesh-delivery's word, the members on that walk now take their turn, the same way members sharing one walk already do. That one walk delivers both, in the group's order among its tiers (ADR 0276 decision 5). A member on another walk still waits. The test `TestAMemberWhoseWalkCarriesTheSupersededOneBeforeItTakesItsTurn` uses the fold shape (A superseded; B and C published on plan-2, which waits for the word). It ends with plan-2 let go and the group delivered. With the sharing branch switched off, the test fails: B stays published and nothing is let go.
2. **Saying why B waits.** A new field on the delivery, `turn_awaits` and `turn_awaits_until`, names the superseded member, each carrier with its state and walk, and until when. `show` shows it, because the field is part of the delivery. `boundOf` uses it for a published member, so `stalled` gives that member the carrier's deadline instead of the 30 minutes for a published delivery, and says those words. The field is cleared once the member takes its turn or ends. Tested: B is not listed stalled at 35 minutes, and its bound names C and "until". `deliveries` still says "its turn in group …", because `waitsFor` lives in verbs.go, which #201 holds; a one-line follow-up there can read `TurnAwaits` once #201 lands.
3. **A carrier still assembling in a batch (state ready).** It is awaited for `readyCarrierBound` from its merge: 10 minutes for the merge window at its longest plus 2 hours for the one open walk the batch may queue behind (ADR 0276 decisions 1, 4 and 9). Tested in `TestACarrierInABatchIsAwaitedWithinTheWindowAndOneWalk`.
4. **The nit.** `carryAwaited` now returns a `carryWait`. When it names no carrier and gives no reason, every module is carried and contained: the member is reached, never a stop with an empty reason. Tested in `TestACarriedMemberNotYetMarkedCoveredIsReached`, which calls `groupTurns` directly; it fails with that branch switched off.
5. **The chain.** In `TestTheWaitFollowsACarrierSupersededInTurn`, C's waiting walk is folded into D's walk, which answers only D. B then waits for D, and goes once D covers A.
`go vet ./...`, `go test -race ./... -count=1` in modules/mesh-delivery, and the words check all pass.
mesh/merge-gate: PASS — every machine composes with the change as it did without (4 of 4 compose) (modules: 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 mesh-delivery → novox; no bus step
tier 0: mesh-delivery
novox: receives mesh-delivery
Run by the build seat on ace as build-1791682936705726766 (builds --log build-1791682936705726766).
--- the gate: mesh-delivery
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: 0
files asking for a setting without saying whether it is trusted: 0
1 manifest(s) checked. Judged against the seats this binary carries; a claim on one of the mesh's own seats is judged fully at registration, and a seat declared by a module not given here reads as unknown. Identities judged on a 6-character machine name, against the bounds of the providers given here
judging 0 module(s) changed against 4 machine(s), as the snapshot of 2026-10-11T01:36:53Z 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 2026-10-11T01:36:53Z
machines:
- a machine (f35): composes
- holds mesh-dns-resolver (idt): composes
- holds git, holds mesh-artifact-store, holds mesh-broker, holds mesh-catalog, holds mesh-delivery, holds mesh-dns-resolver, holds mesh-store, holds mesh-vault, holds npm-package-registry, the control node, the hub (ygcyr): composes
- a machine (yvbpsm): composes
a merge would rebuild 1 module(s) in 1 tier(s): mesh-delivery
--- 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 39479 strings of 655 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
testing modules/mesh-delivery
ok mesh-delivery/cmd/mesh-delivery 1.757s
--- its merge-check-typescript.sh (typescript toolchain)
**Merge check** at `f7ba413c`
- `mesh/merge-gate`: **PASS** — every machine composes with the change as it did without (4 of 4 compose) (modules: 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 mesh-delivery → novox; no bus step
- tier 0: mesh-delivery
- novox: receives mesh-delivery
Run by the build seat on ace as `build-1791682936705726766` (`builds --log build-1791682936705726766`).
```
--- the gate: mesh-delivery
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: 0
files asking for a setting without saying whether it is trusted: 0
1 manifest(s) checked. Judged against the seats this binary carries; a claim on one of the mesh's own seats is judged fully at registration, and a seat declared by a module not given here reads as unknown. Identities judged on a 6-character machine name, against the bounds of the providers given here
judging 0 module(s) changed against 4 machine(s), as the snapshot of 2026-10-11T01:36:53Z 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 2026-10-11T01:36:53Z
machines:
- a machine (f35): composes
- holds mesh-dns-resolver (idt): composes
- holds git, holds mesh-artifact-store, holds mesh-broker, holds mesh-catalog, holds mesh-delivery, holds mesh-dns-resolver, holds mesh-store, holds mesh-vault, holds npm-package-registry, the control node, the hub (ygcyr): composes
- a machine (yvbpsm): composes
a merge would rebuild 1 module(s) in 1 tier(s): mesh-delivery
--- 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 39479 strings of 655 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
testing modules/mesh-delivery
ok mesh-delivery/cmd/mesh-delivery 1.757s
--- its merge-check-typescript.sh (typescript toolchain)
```
A turn given on one carrier's waiting walk let the next member go while
another module of the member before it waited on a walk that could still
fail, and the stop then took the shared walk down. The turn is now shared
only when that one walk carries every module not yet carried, each by a
merge the forge said contains the member's; an unknown answer is asked and
waited for (novox/hq issue 445, re-review of #203).
The false go.waitingWalk is replaced by sharedWalk. A walk is shared only when every module of A not yet carried has a carrier on that same walk, and that walk has not started and waits for mesh-delivery's word. The new test TestASharedTurnNeedsEveryModuleOnTheOneWalk uses your shape: C on B's walk carries mesh-controller, and D on its own rolling walk carries route-proxy. It checks three outcomes:
While D's walk is open, B waits and names both carriers.
When D is delivered and covers route-proxy, B goes on C's walk.
When D's walk fails, B is stopped and never reaches delivering.
The forge's word first. A shared turn is given only when the forge said yes for a carrier of each module on that walk. If the answer is not known, the forge is asked (askContains) and B waits within the carrier's bound. If the forge says no, that merge is no carrier, so B gets no turn and is stopped with "does not contain its merge". The test is TestASharedTurnWaitsForTheForgesWord.
A related gap, now closed: coverOf asks the forge only once every module has a delivered carrier. While another module still waits on an open walk, a delivered carrier's containment was never asked, and B would have been stopped when the 30-minute grace ran out. carryAwaited now names those commits, and turnsOf asks the forge about them.
readyCarrierBound on its own. The batch test now pins the value to 10 minutes plus Bounds[Delivering]. It checks that B still waits one minute before the bound and is stopped one minute after it, with "no walk carried it within 2h10m0s". I removed the ready case from carryAwaited to check the test: it then fails, because the wait becomes a rolling 2 hours that never ends.
turn_awaits_until is now a *time.Time, so it is left out of the JSON when no wait is set.
The first three new tests failed with f7ba413c's choice of walk. I put that rule back in place for the run, because the f7ba413c tree does not compile the new pointer field. The not-yet-said and said-no subtests both gave B a turn on plan-2. In the two-carrier test, B went on plan-2 while D's walk was open, and when D's walk failed B was stopped while delivering. go vet ./..., go test -race ./... -count=1 and the words check all pass.
Re-review addressed. The new head is 15a24fb6.
1. **The false go.** `waitingWalk` is replaced by `sharedWalk`. A walk is shared only when every module of A not yet carried has a carrier on that same walk, and that walk has not started and waits for mesh-delivery's word. The new test `TestASharedTurnNeedsEveryModuleOnTheOneWalk` uses your shape: C on B's walk carries mesh-controller, and D on its own rolling walk carries route-proxy. It checks three outcomes:
- While D's walk is open, B waits and names both carriers.
- When D is delivered and covers route-proxy, B goes on C's walk.
- When D's walk fails, B is stopped and never reaches delivering.
2. **The forge's word first.** A shared turn is given only when the forge said yes for a carrier of each module on that walk. If the answer is not known, the forge is asked (`askContains`) and B waits within the carrier's bound. If the forge says no, that merge is no carrier, so B gets no turn and is stopped with "does not contain its merge". The test is `TestASharedTurnWaitsForTheForgesWord`.
- A related gap, now closed: `coverOf` asks the forge only once every module has a delivered carrier. While another module still waits on an open walk, a delivered carrier's containment was never asked, and B would have been stopped when the 30-minute grace ran out. `carryAwaited` now names those commits, and `turnsOf` asks the forge about them.
3. **`readyCarrierBound` on its own.** The batch test now pins the value to 10 minutes plus `Bounds[Delivering]`. It checks that B still waits one minute before the bound and is stopped one minute after it, with "no walk carried it within 2h10m0s". I removed the ready case from `carryAwaited` to check the test: it then fails, because the wait becomes a rolling 2 hours that never ends.
4. **`turn_awaits_until`** is now a `*time.Time`, so it is left out of the JSON when no wait is set.
The first three new tests failed with f7ba413c's choice of walk. I put that rule back in place for the run, because the f7ba413c tree does not compile the new pointer field. The not-yet-said and said-no subtests both gave B a turn on plan-2. In the two-carrier test, B went on plan-2 while D's walk was open, and when D's walk failed B was stopped while delivering. `go vet ./...`, `go test -race ./... -count=1` and the words check all pass.
mesh/merge-gate: PASS — every machine composes with the change as it did without (4 of 4 compose) (modules: 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 mesh-delivery → novox; no bus step
tier 0: mesh-delivery
novox: receives mesh-delivery
read by no module's build: modules/mesh-delivery/cmd/mesh-delivery/issue445_test.go
Run by the build seat on ace as build-1791683221265576677 (builds --log build-1791683221265576677).
--- the gate: mesh-delivery
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: 0
files asking for a setting without saying whether it is trusted: 0
1 manifest(s) checked. Judged against the seats this binary carries; a claim on one of the mesh's own seats is judged fully at registration, and a seat declared by a module not given here reads as unknown. Identities judged on a 6-character machine name, against the bounds of the providers given here
judging 0 module(s) changed against 4 machine(s), as the snapshot of 2026-10-11T01:36:53Z 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 2026-10-11T01:36:53Z
machines:
- a machine (f35): composes
- holds mesh-dns-resolver (idt): composes
- holds git, holds mesh-artifact-store, holds mesh-broker, holds mesh-catalog, holds mesh-delivery, holds mesh-dns-resolver, holds mesh-store, holds mesh-vault, holds npm-package-registry, the control node, the hub (ygcyr): composes
- a machine (yvbpsm): composes
a merge would rebuild 1 module(s) in 1 tier(s): mesh-delivery
--- 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 39479 strings of 655 files
ok git.novox.be/novox/mesh-catalog/checks/agenthome 0.010s
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
testing modules/mesh-delivery
ok mesh-delivery/cmd/mesh-delivery 1.801s
--- its merge-check-typescript.sh (typescript toolchain)
**Merge check** at `15a24fb6`
- `mesh/merge-gate`: **PASS** — every machine composes with the change as it did without (4 of 4 compose) (modules: 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 mesh-delivery → novox; no bus step
- tier 0: mesh-delivery
- novox: receives mesh-delivery
- read by no module's build: modules/mesh-delivery/cmd/mesh-delivery/issue445_test.go
Run by the build seat on ace as `build-1791683221265576677` (`builds --log build-1791683221265576677`).
```
--- the gate: mesh-delivery
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: 0
files asking for a setting without saying whether it is trusted: 0
1 manifest(s) checked. Judged against the seats this binary carries; a claim on one of the mesh's own seats is judged fully at registration, and a seat declared by a module not given here reads as unknown. Identities judged on a 6-character machine name, against the bounds of the providers given here
judging 0 module(s) changed against 4 machine(s), as the snapshot of 2026-10-11T01:36:53Z 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 2026-10-11T01:36:53Z
machines:
- a machine (f35): composes
- holds mesh-dns-resolver (idt): composes
- holds git, holds mesh-artifact-store, holds mesh-broker, holds mesh-catalog, holds mesh-delivery, holds mesh-dns-resolver, holds mesh-store, holds mesh-vault, holds npm-package-registry, the control node, the hub (ygcyr): composes
- a machine (yvbpsm): composes
a merge would rebuild 1 module(s) in 1 tier(s): mesh-delivery
--- 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 39479 strings of 655 files
ok git.novox.be/novox/mesh-catalog/checks/agenthome 0.010s
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
testing modules/mesh-delivery
ok mesh-delivery/cmd/mesh-delivery 1.801s
--- its merge-check-typescript.sh (typescript toolchain)
```
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Fixes novox/hq issue 445 (a group member superseded by a walk that later carries it still stops the members after it).
Acceptance: in a delivery group [A, B] where A's walk was superseded by a later merge C of A's repository, B stays published while C can still carry A, takes its turn once C's walk is delivered and covers A, and is stopped with a reason naming A only when no later merge can carry it.
What changes
groupTurns(holder.go): a superseded member withCoveredByset counts as reached, asGroupStatealready does since #197 (ADR 0270 decision 2: a group counts a covered member as delivered). A superseded member not yet covered holds the members after it whilecarryAwaitednames a later merge that can still carry it; otherwise they are stopped withits group G stopped: A superseded, and no later merge into main can carry M: <why>. Failed and stopped members stop the rest as before.carryAwaited/mayCarry: a candidate carrier is a later merge of the same repository into the same trunk, on a passing check, whose plan moves the module, and which the forge has not said leaves A out. Every module A moves needs one.What bounds the wait
Bounds: published 30 min, delivering 2 h; a merged delivery with no walk is held after 10 min). Past that bound it no longer counts and B is stopped, naming the bound (ADR 0239 decision 9, ADR 0276 decision 9).containsGrace= 30 minutes (the published bound) after it was delivered; then B is stopped saying the forge did not say. Without this, a forge that never answers would hold B for ever. I chose 30 minutes myself; it is a small design limit.Follows from decisions, no new one
ADR 0239 decision 4 stops the group for a failed or stopped member and otherwise lets the next start when the one before is delivered; superseded is not named as a stop. ADR 0270 decision 2 counts a covered member as delivered so the member after it gets its turn. ADR 0276 says every wait has a bound. The wait is a turn waiting on a cover the mesh already computes.
Tests
issue445_test.go, the live shape of group fix/assign-says-why-a-module-is-not-there used by #197's tests: (a) a later walk open: B waits; (b) that walk delivered and covering A: B is let go; (c) the forge says no, or the later walk failed: B is stopped saying why; (d) the later walk past its own bound, and a forge that never answers pastcontainsGrace: B is stopped. All six subtests failed on main (B stopped at once, or stopped without a reason).go vet ./...,go test -race ./... -count=1in modules/mesh-delivery, and the words check all pass. Does not touch verbs.go, waiting_checks.go (#201) or modules/restic (#200).Delivery
novox/mesh-catalog@15a24fb63ec0— delivered since 2026-10-11T02:26:03ZDelivery plan — builds mesh-delivery → novox; no bus step
Machines
Transitions
The commit's note under
refs/notes/mesh-plankeeps every transition:git log --notes=mesh-plan.Merge check at
51a85c59mesh/merge-gate: PASS — every machine composes with the change as it did without (4 of 4 compose) (modules: 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 mesh-delivery → novox; no bus step
Run by the build seat on novox as
build-1791682479675297115(builds --log build-1791682479675297115).Reworked for the review, now at head
f7ba413c.TestAMemberWhoseWalkCarriesTheSupersededOneBeforeItTakesItsTurnuses the fold shape (A superseded; B and C published on plan-2, which waits for the word). It ends with plan-2 let go and the group delivered. With the sharing branch switched off, the test fails: B stays published and nothing is let go.turn_awaitsandturn_awaits_until, names the superseded member, each carrier with its state and walk, and until when.showshows it, because the field is part of the delivery.boundOfuses it for a published member, sostalledgives that member the carrier's deadline instead of the 30 minutes for a published delivery, and says those words. The field is cleared once the member takes its turn or ends. Tested: B is not listed stalled at 35 minutes, and its bound names C and "until".deliveriesstill says "its turn in group …", becausewaitsForlives in verbs.go, which #201 holds; a one-line follow-up there can readTurnAwaitsonce #201 lands.readyCarrierBoundfrom its merge: 10 minutes for the merge window at its longest plus 2 hours for the one open walk the batch may queue behind (ADR 0276 decisions 1, 4 and 9). Tested inTestACarrierInABatchIsAwaitedWithinTheWindowAndOneWalk.carryAwaitednow returns acarryWait. When it names no carrier and gives no reason, every module is carried and contained: the member is reached, never a stop with an empty reason. Tested inTestACarriedMemberNotYetMarkedCoveredIsReached, which callsgroupTurnsdirectly; it fails with that branch switched off.TestTheWaitFollowsACarrierSupersededInTurn, C's waiting walk is folded into D's walk, which answers only D. B then waits for D, and goes once D covers A.go vet ./...,go test -race ./... -count=1in modules/mesh-delivery, and the words check all pass.Merge check at
f7ba413cmesh/merge-gate: PASS — every machine composes with the change as it did without (4 of 4 compose) (modules: 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 mesh-delivery → novox; no bus step
Run by the build seat on ace as
build-1791682936705726766(builds --log build-1791682936705726766).Re-review addressed. The new head is
15a24fb6.waitingWalkis replaced bysharedWalk. A walk is shared only when every module of A not yet carried has a carrier on that same walk, and that walk has not started and waits for mesh-delivery's word. The new testTestASharedTurnNeedsEveryModuleOnTheOneWalkuses your shape: C on B's walk carries mesh-controller, and D on its own rolling walk carries route-proxy. It checks three outcomes:askContains) and B waits within the carrier's bound. If the forge says no, that merge is no carrier, so B gets no turn and is stopped with "does not contain its merge". The test isTestASharedTurnWaitsForTheForgesWord.coverOfasks the forge only once every module has a delivered carrier. While another module still waits on an open walk, a delivered carrier's containment was never asked, and B would have been stopped when the 30-minute grace ran out.carryAwaitednow names those commits, andturnsOfasks the forge about them.readyCarrierBoundon its own. The batch test now pins the value to 10 minutes plusBounds[Delivering]. It checks that B still waits one minute before the bound and is stopped one minute after it, with "no walk carried it within 2h10m0s". I removed the ready case fromcarryAwaitedto check the test: it then fails, because the wait becomes a rolling 2 hours that never ends.turn_awaits_untilis now a*time.Time, so it is left out of the JSON when no wait is set.The first three new tests failed with f7ba413c's choice of walk. I put that rule back in place for the run, because the
f7ba413ctree does not compile the new pointer field. The not-yet-said and said-no subtests both gave B a turn on plan-2. In the two-carrier test, B went on plan-2 while D's walk was open, and when D's walk failed B was stopped while delivering.go vet ./...,go test -race ./... -count=1and the words check all pass.Merge check at
15a24fb6mesh/merge-gate: PASS — every machine composes with the change as it did without (4 of 4 compose) (modules: 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 mesh-delivery → novox; no bus step
Run by the build seat on ace as
build-1791683221265576677(builds --log build-1791683221265576677).