A held have that carries a wait is a hold of the composition, not a failed backup; a backup ask whose have now carries a wait or is no longer held is withdrawn with its condition.
Builds: design 47 G A-453, ADR 0539 §16, §541
compose-candidate composes a whole revision in shadow on each node and
answered in about 34 s on g14 and shanks (2026-11-16 23:27), past the
SDK's 30 s call bound, so node-tools' composition step could not run. A
responder that is absent still answers at once. Quick fix; the thorough
one is a bound per verb from the seat's definition (after G).
main.981.node-tools passed build, lint, unit and integration; only its
composition step could not run (g14 and shanks composed in about 34 s,
past the assigner's 30 s, while shanks ran the pipeline's own tests).
The seat verb rerun asks one step again, but no tool reached it.
Builds: design 47 G §482
mesh-hand-acts 0.1.13 refuses a hand act with no cause; the pipeline
runner's record (cancel, rerun, accept, run, ask-again, review, start) and
the checker tools' requests drop gave none, so every one was refused and
nothing was done (runner_cancel, 2026-11-16 23:12: 96 of 96 refused). The
cause is the verb's own word, the one a second act for the same reason uses.
A batched take-in names only its newest revision's movers; the other modules it moved never got a pipeline.
Starter.TakenIn and the runner's first-half pickup now start one for each module of the branch's head with no pipeline at its fingerprint.
Builds: design 47 G §44, §54, §88, §476
#228 cut long hexadecimal runs from an assigner ask's words but checked the warrant against the assigner's whole words:
the operator's answer to mesh-hand-acts' backup ask on novox (20:19) was refused as no longer the ask answered.
The backup ask of mesh-hand-acts on novox (2026-11-16 19:53) carried the node-engine's whole fingerprints and was refused
by the channel; nothing reached the operator and the module stayed held.
Builds: design 47 G §675 (kept while named). A start read that ended with a stream empty, or with an event whose set was not
written, leaves the sets not ready; an entry is also kept when a heard want or running build names it though its set key is
missing; a removal says how many wants and running builds it read.
novox 2026-11-16: mesh-hand-acts' want stayed at 4eb3cf48 with no version because the pass returned at
no-passed-pipeline before it named a build. Now each held want without a version, at a fingerprint other than the
trunk's, takes the release whose parts that build made (and parts where its engine reads them); nothing else changes.
Builds: design 47 G §379, §526 (A-96, A-421, A-430a)
novox 2026-11-16 18:46: the mesh-sdk merge moved mesh-hand-acts' fingerprint to 9b3a5534; rule 2 holds the want at
4eb3cf48 (no pipeline passed at the new one), and namedBuild named a release only for the trunk's fingerprint, so the
held want never got version 0.1.13 though it was released at 18:04:46. Now a held build takes the release whose parts
are exactly what it made (A-421's match).
Quick fix = thorough fix here: namedBuild is the one place a want's release is looked up.
novox 2026-11-16: mesh-hand-acts 0.1.13 was released from source 962 (the builder's reservation), its catalog entry's
fp-src is 963, so the assigner found no release and the want on novox named no version and no parts. When no release of
the fp-src is heard, the release whose parts are exactly the digests this fingerprint's build made is this build's (A-421's
match, extended).
Quick fix (this): the assigner matches by parts, which are the evidence. Thorough fix (noted, not here): the builder's
reservation records the entry's fp-src as its source, so both name the same revision.
queue() returned on the first version whose catalog entry could not be read, before it started the drain of the
holders it had already queued at: on 2026-11-16 17:20 mesh-hand-acts 0.1.13 was queued at distribution, then
mesh-delivery-indexer 0.1.14's entry was not kept, and the queued publish was never sent, so no version was
released. Now an unreadable entry leaves only its own version unqueued, the error is still answered (heard again),
and the drain of every holder given a version is started in all cases (design 47 G section 588).
Quick fix and thorough fix are the same here: queue() is the one place that starts a holder's drain.
GrantWaits compared permissions by their text, so a composed subject or allow-responses counted as
missing though the loaded list's '>' already allowed it: all 182 grants waited on '1 permissions'
from 15:43 on 2026-11-16 and no want moved. A subject a loaded one covers is carried now, and
answering is carried where the loaded user may publish every subject.
A grant held (kept or waiting) or left out was read only from the pass object, which no tool opens,
and the assigner's sixth rule waited on it in silence: no grant for any module was written after
15:43 on 2026-11-16 and nothing said why. Each pass now says its grants: wanted, written, deleted,
and every kept, waiting and missing one with its reason.
The build seats' step-ended is a kept seat event on its own stream, never on EVENTS. The assigner
registered its handler on the hub but never bound the kept streams, so every composition step
waited for ever after its composer run passed (pipeline main.961.mesh-hand-acts, 2026-11-16).
Design 47 G §354, A-240.
The indexer's inputs-objects bucket filled twice: deleting an input's key (step-ended, the day sweep)
left its object, which only the person's drop tool removed. The input's key and object now go together,
each write first drops the objects no key names, and a write the store refuses names the store and its fill.
Builds: design 47 G §354, §712, A-237