The takeins durable delivers from the moment it is made, so a take-in said
while the runner was down reached only the main durable, which acknowledged
it with no work: that take-in started no pipeline. Until the takeins durable
is bound, the main durable now works each take-in.taken; from the bind, it
acknowledges with no work those the takeins durable delivers, told apart by
the stream sequence the durable delivers after, so a take-in still waiting on
the main durable is worked whatever the moment it is heard.
Builds: design 47 H2 H-A-500 fourth addendum
The runner heard every event on one serial durable, so a take-in waited behind
hundreds of late step events and its pipeline started ten minutes late. The
runner now makes its own durable <node>_<module>_takeins on EVENTS, filtered
to take-in.taken, and acts on a take-in there alone; the main durable
acknowledges a take-in at once with no work. The test-runs integration test
keeps the runner's findings on their own stream, as the manifest declares:
on EVENTS each reached the main durable unhandled and wrote event-unhandled
again, a flood the take-in now overtakes.
Builds: design 47 H2 H-A-500 (b) and its addenda (2), third addendum (Q2)
Every ask-taken and step-progress listed and read every pipeline to find its
test run, so after a cancel hundreds of late events of ended runs cost one to
two seconds each and held the runner's one event consumer for minutes. The
state test-runs now keeps an ask key ask.<ask> holding the run's key, written
with each ask a run is given and deleted with the run; an event of an ended
run's ask is judged from that key and the run alone and acknowledged at once,
one journal line per hundred.
Builds: design 47 H2 H-A-500 (a) and its addenda (1)
main.291 held the one pipelines-at-once turn 50 minutes waiting for kept
revision 292, which the catalog had removed: nothing pinned it and the wait
had no end. pipeline-started now carries every revision the pipeline's movers
are kept at, and a re-pin says it again with the full list, so the indexer
pins them. A catalog read answered "no such object" is the revision gone;
the placing then ends the pipeline failed "kept revision <r> gone", its
movers are left unfinished and rejoin the next pipeline, and its turn is
freed.
Builds: design 47 H2 H-A-496, H-A-496 addendum
The indexer pins a pipeline's revision from pipeline-started to
pipeline-ended, so a rerun of a step in an ended pipeline (main.258's
composition at 14:41, after its end at 14:15) asked a step against a
revision the catalog had already removed. A rerun of an ended delivering
pipeline now reads its revision fresh through the catalog seat's verb
revision first and is refused by name, nothing asked, when it is gone
("its revision <r> is gone; start the mover again at the head"); once
reopened it says its pipeline-started again, so the indexer pins the
revision until its next end.
Builds: design 47 H2 H-A-479, H-A-479 addendum (2)
A module whose pipeline failed at an unchanged entry was never carried again
until its fingerprint moved, since start refused the head whose pipeline
existed. The start verb now takes modules: the named modules whose newest
entry has no passed pipeline are started at the head's revision, at the
take-in's tiers, keyed <branch>.<revision>.again.<n> (n from 1, written on
absence); a module whose newest entry passed, or that the head holds no
entry of, is refused by name, nothing started. Exactly one of module or
modules is given. The new key form is read by every key-form list (the
sweep's, a release's, the mover keys'). The runner seat's form is 8; the
contract is regenerated.
Builds: design 47 H2 H-A-479, H-A-479 addendum (1)
A cold grants-for (one composing a revision's grants the grants part did
not keep yet) took 3 to 4 s against the 5 s bound and was cut three times
on 2026-11-22; the bound becomes 30 s in the manifest and the code.
Builds: design 47 H2 H-A-478
The holder of mesh-wants is away while a re-pin asks the composition again:
the put meets no responders, is put again at the kept-call pace (a hundredth
of its default, so 40 s is 400 ms) and lands once the holder is back; the step
then passes with no hand act.
Builds: design 47 H2 §775 (H-A-453 (3): the whole criterion proven at its level)
A kept mover's re-pin was declared in the runner's seat but never said, so a
wait on mesh-pipeline-runner.re-pinned was never answered. The runner's whole
test now waits on pipeline-started, step-ended, pipeline-ended and re-pinned by
seat and name, a same-source re-pin revision being one of its steps.
Builds: design 47 H2 §821 (round-1 finding: EventRePinned unused outside the contracts)
Eight unit tests and two integration tests become rows of the test they
sit beside (t.Run of the same body), so the runner ends below its start
of 105 unit and 4 integration tests counted by top-level function.
Builds: design 47 H2 §819
list answers by default the newest 20 short rows (key, call, start, end,
outcome, movers' count and the first failed or could-not-run step with its
module and why), and the whole pipelines only with full and a limit. show
answers one mover's steps when it names one, the steps its filters hold,
or else each mover's row, never every step of every mover. The tools
program describes each seat verb's tool from its contract, which seatgen
now generates a VerbDescription for; runner_cancel reads the whole
pipelines with full. The runner seat's form is 6.
Builds: design 47 H2 H-A-376, H-A-449, §789
A holder whose node lacks what a step's packages need (the artifact's
check-needs, now named on the integration step's ask) says the step's end
declined with needs and lacks, in place of could not run. The runner puts
it again under a new ask id naming in not-on the nodes that declined it;
a holder on a listed node hands it back at once. The nodes that can take it
are the holders of the step's seat and of the need's seat less those that
declined; when none is left the step ends could not run naming the need
and the nodes asked. The build-step part's form is 3, the runner seat's 5.
Builds: design 47 H2 H-A-375, H-A-445 (ADR 0552 §2, §4)
The findings stream is 8 MiB, kept 7 days, its oldest dropped when full
(no findings-max-age setting: nothing reads one). Every state names its
cap: 8 MiB, bounds 16 MiB above four times its live fill (2,188,123 bytes
on novox, 2026-11-22 00:16 UTC; queue 50,604, findings 285,807, the rest
under 3 KB). ask-silent-bound, its file and its environment go. The kept
call's pace is declared, kept-call-pace-first 1s and kept-call-pace-max
30s, read at runtime. The runner consumes mesh-builds' and mesh-wants'
ask-taken and invokes their drop. The whole test writes 10,000 findings
past the manifest's cap and none is refused.
Builds: design 47 H2 §783, §775, §777, §778, §820; H-A-411 (1), (2), H-A-437
A run whose one member is a source-only registration's key
source.<owner>.<name> runs that registration's check: one step of its
kind on the build seat it names, at the repository's root, asked on the
seat's check stream as every step, its status the registration's policy's.
Its mover is the registration's key, its repository the origin and its
fingerprint the commit's root tree hash, read through the seat git's trees
as a kept call; so its step-ended names the key, escaped into one subject
part, and the tree. A candidate's revision places nothing of it; at its
turn in the queue its step is placed from its registration. A source member
beside another, or a key with no source-only registration, is refused. The
registrations follower keys a source-only registration by its key.
Builds: design 47 H2 §815, §775; H-A-408, H-A-438
runner_step_output (pipeline, mover, step) reads the step from the state
pipelines and asks the verb output of the build seat's holder on the node
its step key names, by the step's ask, once per artifact the step ran for;
a step that reads a shared run is answered by that run's ask. The runner
keeps no output of its own: the objects step-outputs go, with the step's
and the log entry's output and show's outputs; the runner seat's form
rises to 4. The manifest invokes the three build seats' output.
Builds: design 47 H2 §791; H-A-412 (1)
A pipeline ending failed or cancelled keeps each mover whose own required
steps passed or never ran in the runner's state unfinished and names it in
pipeline-ended's unfinished; the branch's next kept catalog revision carries
it, or starts a pipeline for it alone. runner_unfinished reads the state,
runner_unfinished_drop drops one key, runner_unfinished_place asks the new
seat verb place-unfinished (contracts regenerated), both hand acts.
Builds: design 47 H2 §786, §820 (unfinished), H-A-398
The bound key ask.<seat>.<ask> is armed at the put, again at the holder's
ask-taken (heard now for mesh-builds and mesh-wants too) with what is left
since taken-at, again from now at each step-progress, and at a start for every
open ask whose key is gone; it is deleted at the step's end. Past it the step
ends could-not-run, naming an untaken ask's seat holder as taking none, and its
ask is dropped through the seat's drop. The silent seats' bound, the second put
under AgainID, the ask2. keys and the put again at a holder's running go; a put
refused by a full work queue is said and put again at the kept-call pace.
Builds: design 47 H2 §777, §778 (H-A-349 DQ-4, H-A-395)
A put that met a transport fault is put again under the ask's own id at the
kept-call pace instead of failing the step at the first fault.
Builds: design 47 H2 §775
The seat generator gives every value, event and answer the wire field form
(left out while 1) and a tolerant decoder answering the form it read; each
contract carries its forms and a digest of its fields, and the compare fails
a definition whose fields changed while its form did not. Every delivery seat
is regenerated, and the readers take the form.
Builds: design 47 H2 §780; H-A-352
main.137 stood twenty minutes with nats and mesh-vault waiting on
mesh-grants packages published at 20:00, before the wait was written,
so no publish event came. The runner now asks the build seat's
published when it writes the wait and places the mover when the
package is published, saying so in the journal apart from an event.
Builds: design 47 H H-A-326 (1), H-A-326 (3), H-A-328
main.137 stood seventeen minutes with nothing open while runner_health
said healthy. The tools program says each such pipeline in its journal
once per key (H-A-313).
Builds: design 47 H H-A-311 (3), H-A-313
main.117 judged its end at 17:45:19 and never wrote it: every mover's widens
was read inside the end's compare-and-set loop, 126 movers each cut on
grants-for, so each try took about eleven minutes and every write was refused
as changed until the restart, whose start pass judged no end.
The reads an end needs are made once, before the compare-and-set, shared by
every mover and every retry; the runner's start pass ends each running
pipeline with nothing open before the queue is placed; every end says its
outcome in the journal.
Builds: design 47 H H-A-288, H-A-289, H-A-223, H-A-243
Review round 1: a kept mover's ask ids still came from the pipeline's start
time, so a pin-held ask or a composition took the id of an ask the runner
had just dropped for the old fingerprint, and a restart never matched the
steps placed again; and a kept mover placed later read the branch's held
view, so a lagging view gave the new fingerprint the old pins. Every ask id
of a kept mover's steps (build, tests, validate, pin-held asks,
composition, rerun, restart) now mixes its kept revision into the id's
sum, and every placing of a kept mover (at once, later, at a restart, a
rerun) reads its entries at its kept revision by number, waiting while
that revision cannot be read. The test's starter gets a clock that moves.
Builds: design 47 H H-A-286 (amending H-A-284; with H-A-283, H-A-285)
A module importing packages its own pipeline releases was re-pinned at
each publish; each re-pin revision of the same source took it out of the
running pipeline, and each newer build withdrew the older reservation, so
nothing ever published. A revision of the same source now moves no mover
out of the open pipeline of that source: the pipeline keeps it at its new
fingerprint, ends every step of it superseded (said at the old
fingerprint, its asks dropped, its runs left) and places them again from
the re-pin revision's entry under new asks, the build first, still naming
the pipeline's revision. The revision the mover is kept at rides in the
record's kept-revisions, beside the movers, where the pin check reads its
entry. A revision whose every mover is kept starts no pipeline; one of a
new source moves movers as before.
Builds: design 47 H H-A-283, H-A-284, H-A-285 (with H-A-223, H-A-244, ADR 0682)
Review round 1. A pipeline that ended (failed, cancelled) with a mover
still waiting left the finding for ever. It is now judged at every
pipeline's end and in Waiting: cleared when the registration is held, or
when no running pipeline still has an unplaced mover of the module.
Builds: design 47 H H-A-262, H-A-265
A mover with no registration had its wait said only in the journal. The
runner now raises the finding registration-missing (scope module, id the
module) through its finding writer as it places a pipeline whose mover has
no registration, and clears it when the module's registration arrives or
the registrations held name it.
Builds: design 47 H H-A-262, G A-327, ADR 0666 §2
The start read the second bound with a get, which says not found both
for a key never armed and for one the holder's ask-taken deleted, so
every later start asked an answered step once more. The first read
keeps each key's last entry; a first bound's expiry is taken only when
its second bound left no entry at all, and with both passed only the
second is taken, ending the step without asking it again.
Builds: design 47 H H-A-238 (5)
The bounds were watched for updates only, so an expiry marker written
while the runner was down was never heard and that ask had no bound.
At the start each ask key's last entry is read; an expiry marker is
taken then. A running heard before the seat-holders view is read is
heard again after a delay instead of at once.
Builds: design 47 H H-A-238 (5), H-A-243 (1)
mesh-builds and mesh-wants say nothing before a step's end, so asking
them again at the short bound would double a running build. Their ask
has one bound, ask-silent-bound, and is put again when the node-engine
says the holder running, since the holder takes asks on a plain
subscription and loses those put while it is down. An ask asked again
on a build seat carries a second id derived in the accepted form.
Builds: design 47 H H-A-235, H-A-238, H-A-241, H-A-243
An ask whose holder lost it (the seat mesh-wants' holder restarted) held
its pipeline's turn for nine hours. Every ask but a test run's now arms
a key of the state bounds before it is put; its expiry asks once more,
said in the journal, and the second ends the step could not run with
the finding call-unanswered, so the pipeline ends and its turn frees.
Builds: design 47 H H-A-235, H-A-238 (1)-(5)
A release re-pinned node-tools while main.54 still built it at its old
fingerprint; the build failed and failed the pipeline for its other 119
movers, because supersession took movers out of waiting pipelines only.
Every start of a newer revision's pipeline now also walks the older running
pipelines of its branch: each mover whose fingerprint it moved leaves, its
open steps ending superseded with a why naming the newer pipeline, a step
reading a test run leaving it as a cancel's does, and the mover kept in the
pipeline's superseded. The end judgement counts a left mover's steps as
neither holding nor failed, so the pipeline ends by its other movers alone;
one every mover left ends superseded. A run on request's pipeline is never
touched.
Builds: design 47 H H-A-223, H-A-163, H-A-213
The runner crashed at every start on a live bus with "concurrent map
writes" in MoverStore.Get: the waiting pipelines of each branch are placed
from goroutines of their own, and the store's last reads and inline marks
were written under each pipeline's own lock alone. The store's mutex now
guards the lock table, the last reads and the inline marks; behaviour is
unchanged. The run test starts the runner over six kept pipelines in the
old form, one of 437 steps, so the start path runs under the race detector.
Builds: design 47 H H-A-187, H §774
The queue's and the test runs' integration tests took several sources in at one
catalog revision, which now are one key. Each take-in has its own revision; the
test runs' test also checks that a test ask names revision and source. A replaced
reader of another pipeline's run stays the unit tests' (one revision is one pipeline).
Builds: design 47 H H-A-213 (4), H-A-219
A re-pin on a module's release, a take-in again and a recompute that moves a
fingerprint each keep a catalog revision of their own. The runner hears a trunk
catalog-revision of those kinds, reads that revision by its number through the catalog
seat's revision verb, and starts one pipeline keyed <branch>.<revision> with the
revision's movers. A package-published no longer starts per-artifact pipelines: it
only places the movers waiting for it. A branch's newest pipeline for a mover is the
one of the highest revision number.
Builds: design 47 H H-A-212, H-A-213 (2) and (4), H-A-217
The runner built the asks of the steps that waited for their pins from the
branch's newest catalog half, so a newer revision that moved the module
changed what a passed build's tests fetched. The entry is now read at the
pipeline's own revision through the runner's revision reader, which keeps
what it read; a revision that cannot be read, or whose entry is of another
fingerprint than the mover's, ends the held steps could not run, naming
why.
Builds: design 47 H H-A-205, H-A-207, H-A-203
When the build of an entry with an empty pin passed, its held lint, unit
and integration steps had no pin to run with and nothing ended them, so a
pipeline stood running and, one pipeline at a time, the queue stopped. The
runner reads the pins the build's end names and, when the build passed,
asks each held step with them, its fetch the entry's as placed; a passed
build whose end names no pins ends the held steps could not run, naming
why, as does an artifact the branch's half no longer names.
Builds: design 47 H H-A-207, H-A-206, H-A-203
The runner's lint, unit and integration asks carried no pins, only the
build ask did, so the Go build module, which refuses a requirement of
another repository with no pin, could not run twelve test steps of the
contracts mover (main.45). stepAsk now puts the entry's packages into the
fetch in the form the entry keeps and the builder sends, so a test runs
against what the build was built against; a package the entry names with
no pin yet is left out.
Builds: design 47 H H-A-203, H-A-192, H-A-183
The runner's state pipelines held a pipeline's every step in its record, and
a record of 112 or 129 movers exceeded the bus's 1 MiB message limit (ADR
0486), so main.26 and main.30 placed no step. One MoverStore per process now
reads and writes the state: the record keeps the movers, tiers, call, source
and outcome with its own steps empty, each mover's steps are written under
steps.<pipeline>.<mover> by compare-and-set as the record is, and nothing is
written when the record or a mover key changed since the read. Every read
and judgement of steps goes through it, so the placing, the step ends, the
end judgement, the hand acts and the ageing see one composed pipeline; show,
list, stands and why-carried compose from the mover keys, and show naming a
mover answers its steps alone. A record still carrying its steps inline is
read whole, said once in the journal, and split at its next write, its
mover keys written before the record.
Builds: design 47 H H-A-187
Built against mesh-sdk at its pin (H-A-166, H-A-169), the runner counts every
event no handler takes as a finding; with no stream for its finding events each
write waited its 3 s for the bus, holding the event loop, so the queue test ran
into its 90 s bound. The tests now make the stream the manifest asks for.
Builds: design 47 H H-A-176, H-A-175 (the tests of the runner's placing), H-A-166, H-A-169
The end judgement took every mover with no step as waiting on its
registration; one may wait for the publish of a mover it builds on. The
registration lookup tells them apart, and the why names each with its own
reason.
Builds: design 47 H §746, H-A-163, H-A-171, G A-327
A pipeline the end judgement before H-A-163 left running had no way to end
but a cancel. The runner's process answers end-again, so the end takes its
turn in the queue; runner_end_again asks it. The seat's definition changes:
its Go and TypeScript contracts are generated again.
Builds: design 47 H §746, H-A-163, H-A-164
The runner started whole with pipelines-at-once 1 and pipelines-waiting-cap 1: a second take-in
waits, saying wait-held, and is placed with wait-ended on the first's end; a newer waiting take-in
takes its shared movers out of an older waiting one, which ends superseded; a start over the cap
waits and raises queue-full, cleared on the count; a cancel places the next; a pipeline on its
approval alone gives up its place and, answered while it still runs, is keyed running at once; a
placing the bus refuses raises pipelines-stalled, cleared when the next start places the waiting
from the runner's own state, which starts no pipeline of its own. The tools runner_rerun,
runner_accept and runner_ask_again are shown to take the mover's module.
Builds: design 47 H §748, §774; acceptance test 112 (the runner's half); lead answers H-A-14, H-A-36, H-A-52, H-A-53, H-A-96, H-A-101, H-A-106, H-A-107
Claude-Session: https://claude.ai/code/session_01JG73KHUnBze48rdua8u2sy
The seat verb place places a waiting pipeline ahead of its turn, written in the hand-act log first:
its running key now, over the bound, its waiting key deleted, its steps placed as at its turn with
wait-ended; a pipeline not waiting is refused. The tool runner_place asks it with the pipeline and
why; runner_waiting lists the waiting pipelines, lowest sequence first, from the state queue. Both
are in the manifest's tools.
Builds: design 47 H §748; lead answer H-A-53
Claude-Session: https://claude.ai/code/session_01JG73KHUnBze48rdua8u2sy
A pipeline whose every step not ended is an approval asked of the operator leaves the bound: its
running key goes and the next waiting is placed; answered and still running, it is keyed running
at once, over the bound. After every placing, pipelines waiting with none running raise the finding
pipelines-stalled naming the first waiting, cleared once one runs. A start beyond
pipelines-waiting-cap is an ordinary waiting pipeline and raises queue-full naming it, cleared when
an end brings the waiting back to the cap. Both findings are the scope delivery's, the kind as id,
warnings with no owner.
Builds: design 47 H §748; lead answers H-A-14, H-A-36, H-A-52, H-A-106
Claude-Session: https://claude.ai/code/session_01JG73KHUnBze48rdua8u2sy
The runner's settings pipelines-at-once (default 1, from 1 to 8) and pipelines-waiting-cap (default
64) are written to files by the manifest and read at the start; its state queue holds queue.next,
waiting.<sequence> and running.<pipeline>. A start (a take-in's, a publish's, a hand start's, a run
on request's) judged against the running keys alone either runs, its running key written after its
pipeline, or is written waiting under the next sequence, its waiting key first, with no step asked,
saying pipeline-started and wait-held step pipeline on pipelines-at-once. Every end, a cancel's too,
deletes the running key and places the lowest sequence, which says wait-ended and has its steps
placed (a run's from its candidate's newest revision). At the start a running key whose pipeline
runs no more is deleted and the queue placed. The test runs' whole test sets pipelines-at-once 8, as
its candidates run side by side.
Builds: design 47 H §748; lead answers H-A-14, H-A-36, H-A-52, H-A-53, H-A-60, H-A-72, H-A-96
Claude-Session: https://claude.ai/code/session_01JG73KHUnBze48rdua8u2sy
rerun, accept and ask-again take the mover's module beside the pipeline and step (C-R4): a call
naming none is taken as the pipeline's one mover and refused on a pipeline of several, and a module
the pipeline does not move is refused; the tools runner_rerun, runner_accept and runner_ask_again
pass it. A step of a mover whose dependency did not build, placed not asked, now says its own
step-ended naming the dependency, so nothing ends silently.
Builds: design 47 H §746, §748; lead answer H-A-101
Claude-Session: https://claude.ai/code/session_01JG73KHUnBze48rdua8u2sy