The pending value holds the stream sequence and the small fields the waits read; the take-in reads the message from the
forge bridge's stream by sequence. A message the stream no longer holds is said (take-in.pending lost, indexer_pending)
and never taken in silently. indexer_take_in_forge takes one forge message in by its sequence.
Claude-Session: https://claude.ai/code/session_01JG73KG7AfSdh6FP8VrBqCn
mesh/check being carried to c78941e80b59 on 4596fbe1b7ac | from ae31b66a9e33, mesh-catalog fd4931a75261 | novox/mesh-delivery#153 in group feat/take-in
The group's check failed at 7b7096f3: the group's own manifest validator refuses the indexer's stream "take-in", outside
the bus's predefined forms. The lead's ruling: the indexer says take-in.pending and take-in.taken as its ordinary events
(its emits, mesh.mod.<module>.event.take-in.<kind>), as manifest-accepted is said, and keeps no stream. The wait's key is
in the body only: an emitted event names one event (no wildcard), and a key such as g.<group> is no event name. The
watchman consumes the two events by name on the shared events stream; it needs no replay, since it seeds its waits from
indexer_pending at its start. indexer_pending says the last event of each wait as take-in.<kind>.
Run locally, as the check runs it: cmd/mesh-judge over every manifest of mesh-catalog, mesh-delivery, mesh-host,
mesh-tools, mesh-networking and mesh-media-catalog, with this branch's and mesh-catalog's feat/take-in manifests in the
change case and origin/main's in the trunk case: 171 of 171 accepted in both. The same run with the indexer's manifest at
7b7096f3 refuses it with the check's words.
Claude-Session: https://claude.ai/code/session_01JG73KG7AfSdh6FP8VrBqCn
mesh/check failed (2): mesh-build-go: the indexer's judge at the group's mesh-delivery head 7b7096f37ea76e1…; and 1 more | 7b7096f37ea7 on ae31b66a9e33
- A take-in releases its claims (compare-and-set, only while the claim is still its own) on a lost claim, on a claim it
cannot read back, and on a merge it fails to take in: a claim left behind refused every later take-in until a
restart. T5 row with a store that fails once, both ways.
- The start step waits for nothing: with the build it runs unknown it says so and the forge consumer starts; a merged
message left unacknowledged asks the build again (clause 1). T10 row: the start answers at once and takes nothing in.
- A toolchain report evaluates pending again (ToolchainReported), since the build a revision names moves with the
toolchain. T7 row: the revision is taken in once the report makes it name the running build.
- The group's end is told by its whole subject (process.ModuleStreamSubject), so a repository named forge/group is
never taken for one; row in the indexer's handle test.
- The watchman asks the indexer's waits through indexer_pending at its start (invokes), before its durable consumer,
which does not hear again what it acknowledged before a restart; the comment that said the whole stream is heard is
corrected. W2 row: a wait seeded within its bound is raised past it.
- bridge_groups names each group's last end ask, its who, why and when, and the bridge's answer (state end-asks).
- The check's words for the indexer's rules: the trunk's manifest validator.
- The bridge's end drops the merged-state test its merge commit already implies.
- Test budget: groupask_test folded into groups_test, and the K1 mesh/check test into the owed status one: 804 unit
tests (was 806).
Claude-Session: https://claude.ai/code/session_01JG73KG7AfSdh6FP8VrBqCn
The operator, 2026-11-04: logic one module owns lives inside that module as a package of its own, so the take-in step
changes no mesh-sdk. TakeIn, its keys and event names, PendingWait and PendingAnswer move to
modules/mesh-delivery-indexer/takein (its merges a takein.Merge of its own); Forge whole, with its group fields,
ForgeGroupEnd and GroupMerge, moves to modules/mesh-delivery-forge-bridge/forgemsg. Both are nested Go modules that their
readers take by replace, as buildrun is. Every importer moves in this commit: the indexer, the bridge, the watchman and
the view. Forge's private checks (the repository name, the version) are copied small; mesh-sdk main still holds an unused
Forge until its own fold removes it.
The indexer, bridge, watchman, view and checker pin mesh-sdk 5a38a12 (origin/main), no feat/take-in commit.
The bridge's integration bus now grants the state end-asks that bridge_group_end_again added: without it every test of
the package failed on a permissions violation. Test budget: two new test functions (TestForge, TestTakeInEvents) and the
two the join added are paid by merging four pairs in internal/bridge; the repository stays at 806 unit tests.
Claude-Session: https://claude.ai/code/session_01JG73KG7AfSdh6FP8VrBqCn
A revision changing the indexer writes only what the named build needs and is
taken in when that build starts; a gathered group stays across the restart and
its end makes one take-in.
Claude-Session: https://claude.ai/code/session_01JG73KG7AfSdh6FP8VrBqCn
The consumer tells a group's end apart by its name before the forge decode,
a merged message goes through the pending state, and the pending revisions
naming this build are taken in before the consumer starts; take-in events go
on the indexer's own stream. Until the pending state is on the bus every
merge is taken in at once, as before.
Claude-Session: https://claude.ai/code/session_01JG73KG7AfSdh6FP8VrBqCn
A merged message is kept in the indexer's own state pending and the merges
there are computed together on overlays of the indexer's states: taken in
whole when the indexer's fingerprint is the running build, otherwise only
the indexer's own entry, graph key and verdict are written and the rest
waits for the named indexer. A group's members are gathered until its end.
Waits are said on the stream take-in, read by indexer_pending and steered
by indexer_take_in_pending and indexer_drop_pending.
Claude-Session: https://claude.ai/code/session_01JG73KG7AfSdh6FP8VrBqCn
docker's daemon file names the image seat's reach while its manifest stays, so
the hold read from the manifest alone let an engine before the reading lose the
file. The forge bridge reads each manifest's templates at the merge and carries
the module's needs whole; the indexer takes them, keeps them past the judge, and
holds any changed entry, not only a new manifest digest.
A hold standing past 30 minutes is the finding manifest-held, its bound a key
of the new state hold-bounds that expires; its expiry and the start weigh the
holds again, so a restart loses no bound. A hold moved by hand names the nodes
it strands, in its answer and its hand act. The hold test is a case of the
manifests test, not a function of its own.
Claude-Session: https://claude.ai/code/session_01JG73KG7AfSdh6FP8VrBqCn
Every node-engine reads the store holder's manifest to find the store, and an
engine before form:seat-reach finds it only through the retired provision, so
writing the new entry while one counted engine lacks the reading would leave
that node without the store. The entry keeps its old manifest, fingerprint and
inputs; the new one waits in the state holds and moves once every counted
node-engine (ADR 0624 §1, read as the builder reads them) states it.
indexer_holds reads, asks again or moves a hold by hand; indexer_engine_reads
reads the engines' statements.
Claude-Session: https://claude.ai/code/session_01JG73KG7AfSdh6FP8VrBqCn
mesh-sdk's views.Emitter writes the owner's finding of a change not said past its bound only when Findings is
set; both owners of views in mesh-delivery now set it to their findings state.
Claude-Session: https://claude.ai/code/session_01JG73KJNrH5EgFXMKKYSrig
The indexer's, the assigner's, the builder's and the watchman's consumers that were not there or could not be
heard only said so in the log, so the part stayed healthy while it heard nothing (ADR 0615 §5). sdk/heard is
the one shared helper: the finding family-not-heard names the observer and the family, written when the
consumer does not bind and deleted when it does.
Claude-Session: https://claude.ai/code/session_01JG73KJNrH5EgFXMKKYSrig
views.IndexerStream and views.AssignerStream are gone from mesh-sdk (J3 lead's (a)): the indexer names its own views
family "indexer-views" and the assigner's "assigner-views", as each manifest's `streams` names them. Its first-view
ask-again records the hand act through seat:mesh-hand-acts.hand-act (views.StatusTools takes a handacts.Recorder;
the manifest already invokes the verb). judge_test.go gofmt'd (c6c6687 left it unformatted).
Claude-Session: https://claude.ai/code/session_01JG73KJNrH5EgFXMKKYSrig
At each trunk merge the indexer reads every package artifact's name from the merge commit (a Go package's go.mod module
path, a TypeScript package's package.json name) and writes it into the series key as name, beside the module and artifact
it is built from, so the builder numbers it under that name before its first version. Series keys, accepted series and
indexer_package_users go by the name (the lead's design limit for design 32, replacing the builder-name fallback).
Claude-Session: https://claude.ai/code/session_01JG73KJNrH5EgFXMKKYSrig
Cases in TestAMeshSDKMergeMovesOnlyTheModulesImportingAChangedPackage: the checker's verdict event keeps a pull request's
files, a trunk build end raises generated-types-out-of-step and a clean one clears it, as indexer_out_of_step reads them;
indexer_seat_definitions_again, asked through takeins, reads a pull request's head and indexer_seat_definition answers it.
Claude-Session: https://claude.ai/code/session_01JG73KJNrH5EgFXMKKYSrig
A second repository's trunk defining a seat another trunk holds is refused and kept nowhere, with the finding
seat-defined-twice naming both repositories; indexer_seat_definitions_again for a pull request reads the head's own
generated-from, so a pull request changing that field is read right by hand too (the lead's rulings on A2i 3 and 4).
Claude-Session: https://claude.ai/code/session_01JG73KJNrH5EgFXMKKYSrig
A case of the indexer's existing whole-module test: the assigner's first view is held until the
indexer has started, and only the placed node's toolchain counts once it answers (ADR 0621 §3).
Claude-Session: https://claude.ai/code/session_01JG73KJNrH5EgFXMKKYSrig
Design 26 and ADR 0608 §2-§4: the definitions are read at each trunk merge and pull request head
through the forge's get-file tool, parsed by mesh-sdk seatdef and kept in the state seat-definitions
by seat and source; seat-definition-changed is said, kept seven days, only when a digest moved. A
trunk definition that does not parse leaves the trunk's entry and writes seat-definition-unparsed;
a pull request's entries go when it closes or merges. A trunk build end naming out-of-step files
writes generated-types-out-of-step, cleared only by a clean build of the same module; the checker's
verdict keeps a pull request's. Tools indexer_seat_definition, indexer_seat_definitions_again (a hand
act) and indexer_out_of_step.
Claude-Session: https://claude.ai/code/session_01JG73KJNrH5EgFXMKKYSrig
The bus module's orphans need to hear a module that asks nothing now (design 25 § 12.2): every newly accepted manifest
is said, its entries an empty list when it asks for none, and a module whose directory leaves the trunk gets an empty
one; a manifest set aside to wait gets none. A failed publish is published once more; one failing twice writes the
finding accepted-unsaid naming the module, said again at the next judging pass and cleared once said.
indexer_say_accepted says it again by hand from the held verdict, a hand act through mesh-hand-acts; it refuses a
module with no accepted verdict.
Claude-Session: https://claude.ai/code/session_01JG73KJNrH5EgFXMKKYSrig
So a user left on an older series is seen (design 32): per package its series with each one's newest published
version, and its trunk users with the series they accept, that series' newest and the version they are pinned to.
Claude-Session: https://claude.ai/code/session_01JG73KJNrH5EgFXMKKYSrig
A series' newest published Go package keeps its module's digests read at the version's commit; each module of another
repository whose series the user's manifest accepts enters the import form and the version line by them, pinned to the
version (Go's form). A series with no published version keeps the commit form; a group member keeps its head.
A publish that moves a series' newest recomputes at once.
Claude-Session: https://claude.ai/code/session_01JG73KJNrH5EgFXMKKYSrig
A user takes the newest of the series it accepts (ADR 0609 §1), so the
indexer learns versions from the builder's numbered and the registry seats'
published, one key <package>.<series> beside the per-module digests.
Claude-Session: https://claude.ai/code/session_01JG73KJNrH5EgFXMKKYSrig
Filled at each merge from the go.sum the indexer reads (the lead's ruling on I7), read once per consumer per process
even where its hash is unchanged; the builder's floor asks {package} and reads {requirements: [...]}, highest last.
Claude-Session: https://claude.ai/code/session_01JG73KJNrH5EgFXMKKYSrig
Answers {revision, modules:[{module, digest, revision, bus-state}]} from the indexer's own states, at the verdicts
state's mark, as A5's keeper reads it.
Claude-Session: https://claude.ai/code/session_01JG73KJNrH5EgFXMKKYSrig
indexer_config_readers {module} answers {readers: [{module, key, kind,
readers}]} for every value key and own secret any kept manifest of the
module ever declared: the module where its trunk's manifest declares it,
each open pull request whose head declares it, none for a key no longer
read (design 32 *A value no longer read*). It reads one watched copy of
the modules, manifests and config-candidates states. Nodes is left out:
the indexer cannot say which nodes' builds read a key. The indexer's
findings are said on its views family and read by indexer_read_findings,
for the condition keeper (the lead's ruling).
Issue 1966.
Claude-Session: https://claude.ai/code/session_01JG73KJNrH5EgFXMKKYSrig
The judge's foundation holds exactly the nine parts of ADR 0619 §1 and §3,
left out of the old sections' count, and refuses a foundation manifest
asking bus state in the new form. The indexer also counts, across the
accepted manifests, the reads entries still declared on the indexer's and
the assigner's state, by owner and by entry; indexer_old_sections answers
both, so design 25 § 12.7's live row (0 on each owner) has its tool.
Issue 1966.
Claude-Session: https://claude.ai/code/session_01JG73KJNrH5EgFXMKKYSrig
The indexer no longer reads the assigner's placement bucket (ADR 0615 §1):
it asks mesh-delivery-assigner.placement at its start, waiting at most the
two-minute bound so a normal start computes the fingerprints once, then
follows assigner-views on its durable consumer, asking once more on a gap.
Past the bound it raises its finding naming the assigner and placement,
cleared by the answer. Its status is kept in its state first-views, read
by indexer_first_views; indexer_first_view_again asks again by hand.
sdk/views gains the event's write time, the status hook, and the status
state and tools a reader's bundle serves.
Issue 1966.
Claude-Session: https://claude.ai/code/session_01JG73KJNrH5EgFXMKKYSrig
The forge bridge now says a branch's end as branch-deleted (design 51); the
indexer terminated it as unreadable. It acknowledges it, writes it to heard
as ignored, and the two handle tests are one table (mesh-delivery is over
its unit budget).
Issue 1966.
Claude-Session: https://claude.ai/code/session_01JG73KJNrH5EgFXMKKYSrig
The indexer's tools bundle runs in a runtime of its own on its own account
(runs-as) and serves the first-view tools manifests, modules and verdicts,
each answering only the callers its manifest writes (ADR 0621 §1-§2, ADR
0622 §1), and the nine read tools of design 25 § 12.2. The process says each
change of its ten read states on the family indexer-views in its old streams
section (7 days, 256 MiB), with a mark per state the first view answers its
revision from, so a reader that takes the answer misses no event after it
(ADR 0615 §4-§5). The shared shape lives in sdk/views: event, answer, mark,
the emitter, the tools and the reader's follower.
Issue 1966.
Claude-Session: https://claude.ai/code/session_01JG73KJNrH5EgFXMKKYSrig
mesh/check checking 70ade15c36eb on dd90a26251b9 | novox/mesh-delivery#149 in group feat/l3-final
The indexer judged only the manifests its modules state holds, so a digest refused by an older judge (node-tools
48c729f6 and mesh-provisioner-client a382b9df, refused for words then unknown) kept that refusal once it was back on
the trunk or in a candidate. Each such digest is now judged alone, in the catalogue the trunk holds, writing its
verdict and nothing else; an accepted verdict stays accepted under any later judge (ADR 0579 §1), so only a refused or
waiting verdict by another judge, or none, is judged again.
Claude-Session: https://claude.ai/code/session_01JG73KG7AfSdh6FP8VrBqCn
dropRemoved hands the module over from the other trunk, read from the forge at the head the source names, so no wait in memory is needed. indexer_take_in with a repository takes in a head with no module.json as a merge that leaves none, and answers by name for a head with modules.
The check of 2026-11-02 22:35 on a loaded shanks failed two buildrun tests on wall-clock bounds of 60 ms, the
second failure of this kind tonight (the first: mesh-sdk's busretry tests, f02f142).
Quick fix: give the two buildrun asker tests bounds of seconds.
Thorough fix (chosen, because the kind recurs across every module of the repository): every test that raced a
wall clock now waits on an event, steps a clock it is given, or has a bound only a failing test reaches.
- buildrun asker tests: the steps' own times stand for the clock (bound an hour, starts and phases written in the
past), no goroutine paces phases with sleeps.
- builder: the decline bound reads Builder.Now; the decline test steps it; the stream-wait test waits for five
check reads instead of 100 ms.
- checker: the lost-ask try bounds an untaken ask at 200 ms only where that is what ends it; the program says when a
merge turn changes, and the integration test waits on that line instead of 300 ms.
- teller: the bed makes the teller's own ends bucket (a 10 s stall at start the tests raced by 0.5 s), waits for its
warrant subscription instead of 500 ms pauses, and the package-level now is restored with t.Cleanup.
- watchman: the recount, run in its own goroutine, is waited for instead of 10 ms.
- polls of 2-5 s become 10 s (unit) and 20 s (integration).
Under a throttled cgroup (half a core shared with two busy loops) beside the node-engine's suite, the buildrun
tests failed 5 of 20 runs before and passed 30 of 30 after; no test added.
Claude-Session: https://claude.ai/code/session_01NcbgfJ9Gum47zkFKWRaovL
Review finding 2 (ADR 0593 §2, §3): after the release no build has reported imports, and a module that
replaces a mesh Go module of its own repository by a directory had no go.sum line and no version for it,
so a change of buildstep or buildrun alone moved no importer. The indexer now reads go.mod's directory
replaces, publishes those modules, and puts their content digest on the version line; builds are pinned
to the other repositories' versions alone.
Claude-Session: https://claude.ai/code/session_01NcbgfJ9Gum47zkFKWRaovL
ADR 0593 §7: compose and frombus left mesh-sdk for the node-engine, so how a composer holds a refused
module is the engine's to test; the indexer's test keeps proving the verdicts and refusals it writes.
Claude-Session: https://claude.ai/code/session_01NcbgfJ9Gum47zkFKWRaovL
mesh-delivery is over its budget (810 unit, 73 integration on main), so this
branch adds none net: its new cases are subtests, the three work-queue process
tests and the consumes test become one, the history stream bounds join the
stream values table, and the go/-only mesh-sdk merge check joins the one new
integration test. buildrun's 20 unit and 1 integration tests moved in from
mesh-sdk with it.
Claude-Session: https://claude.ai/code/session_01NcbgfJ9Gum47zkFKWRaovL
The indexer whole: two catalogue modules report their imports at their builds'
end, mesh-sdk merges a change of buildstep alone, the packages key names the
merge with a digest per package, the buildstep importer moves and the other
keeps its fingerprint though the version line would have moved it (ADR 0593
§2-§4; D11).
Claude-Session: https://claude.ai/code/session_01NcbgfJ9Gum47zkFKWRaovL
jschoubben