Commit Graph
10 Commits
Author SHA1 Message Date
jschoubben 2271b805bb D5 (sdk, indexer): the answer store is the steps state's own steps-objects; a state template takes objects-max-mib
Builds: design 47 G A-313, A-314

Claude-Session: https://claude.ai/code/session_01NcbgfGKaxw984XyAPfLm4h
2026-11-12 22:30:36 +01:00
jschoubben 43f6b57ab2 Main's subject takes the checker's check and recheck again, as A-184 says until it goes
The asker rename (d29e1426) let main's subject, which names no asker, take only a build and a run, so a holder refused the
checker's check asks ("the kind check is not taken on main's subject"), and seven of the checker's eight integration tests
failed. Until main's subject goes it also takes the checker's check and recheck; a step there is still refused.

Builds: design 47 G A-184, §543

Claude-Session: https://claude.ai/code/session_01NcbgfGKaxw984XyAPfLm4h
2026-11-12 20:30:13 +01:00
jschoubben d29e1426d0 Asker rename R2: every asker of a build seat asks under its seat's own name, and its step is keyed so
A-163: the asker in `accept.<queue>.<asker>` is the asking seat's name (mesh-pipeline-runner,
mesh-catalog, mesh-wants, mesh-builds), never a nickname: the builder asks as mesh-builds (was
"builder"), the indexer's validation as mesh-catalog (was "indexer"), the assigner's runs as
mesh-wants (was "assigner"), the runner's steps on `accept.check.mesh-pipeline-runner` and
`accept.delivery.mesh-pipeline-runner` (were `accept.check.step` and `accept.delivery.step`, where
the holder read `step` as the asker). An asker naming none (the checker) asks on main's subject
(A-184). A build seat's queues are delivery, check and recheck alone (A-170, A-184): buildstep's
AskKinds holds no step, a holder takes a step from check or delivery, and the grants part makes a
worker per queue, never per kind. A holder keys a named ask's step `<asker>.<node>.<id>` and main's
`<node>.<id>` (§481, §483); readers watch their own name's keys. The composition step's step-progress
names the one caller mesh-wants' definition lists for step, read from the seat list (A-294).

Builds: design 47 G §481, §483, §543, §600, A-163, A-170, A-184, A-248, A-294

Claude-Session: https://claude.ai/code/session_01NcbgfGKaxw984XyAPfLm4h
2026-11-12 18:37:29 +01:00
jschoubben 386095bd53 A-290 B1: a step answer over the steps state's bound is kept as the object its key names
A step's key keeps its answer only within what the steps state holds per value (kvlimit.ModuleStateValueBytes less
the headers' room). Over it, finish keeps the answer with the holder's KeepOutput as the object output.<ask> in its
outputs store and the key names it in output-object: a check's whole answer, its result emptied; a run's output, the
key keeping the run answer without it. An answer that cannot be kept ends the key failed, saying why. The Puller says
step-ended from the end finish wrote, so the key and the event say one outcome: a run whose program exited other than
0 ends its key failed as well, its answer kept; step-ended names the object finish kept instead of putting it twice.
Step.Whole reads a key's object back through the seat's verb output on the step's node. The part build-step's Step
gains output-object, the verb output says it answers a step's answer; fixtures and the generated contracts follow.

Builds: design 47 G A-290, A-254, §467

Claude-Session: https://claude.ai/code/session_01NcbgfGKaxw984XyAPfLm4h
2026-11-12 16:44:07 +01:00
jschoubben e313248b5f U2b: the reserved version reaches each artifact's build, and a versioned ask names each artifact's registry seat
The ask carried the reserved version (§573), but the build modules' shared run gave each
artifact's build only the artifact's own version, which no asker sets, so no build got
MESH_VERSION and a Go package was packed as a candidate and never published. The run now
gives every build the ask's version, and the builder names the registry seat each artifact
of a versioned build is published through, so a Go package is published under its
reserved version on go-package-registry.

Builds: design 47 G §370, §573, §458; design 51 *Publishing*

Claude-Session: https://claude.ai/code/session_01NcbgfGKaxw984XyAPfLm4h
2026-11-12 16:13:13 +01:00
jschoubben 5fbc366f1b A-286: a member whose fetch merged is previewed and judged whatever its later steps ended with
A check step that errors after its fetch merged ends failed with its answer (the fetch and the levels so far)
beside its error (steprun askrun.go), and the checker previews the merge from it (judge.go), so a failed level
no longer hides the manifest validator and the composition (try 2: catalog, contracts and delivery were
unpreviewed by a seatgen bootstrap failure, not a conflict). Only a head that does not merge or a fetch that
did not answer stays unpreviewed, and the check says which.

Builds: design 47 G A-286 (hq 41dd30fe)
Claude-Session: https://claude.ai/code/session_01NcbgfGKaxw984XyAPfLm4h
2026-11-12 15:19:14 +01:00
jschoubben 7125a9f92f C1c-2a: the step runner says a run's answer in step-ended, inline or as an object, and serves it
The puller's end of a run carries the run's program's last line of standard output as `output` up to StepEndedInlineMost, else keeps
it with KeepOutput and names the object in `output-object`; a program that exits other than 0 ends failed. OutputTool serves the
seat's verb `output` from the kept object, as the whole file the tool runner sends in chunks; OutputsStore names the store.

Builds: design 47 G A-254, A-257, §602

Claude-Session: https://claude.ai/code/session_01NcbgfGKaxw984XyAPfLm4h
2026-11-12 09:37:51 +01:00
jschoubben 7033d5e80d BG2, BG3: the puller refuses a kind its asker may not ask and pulls the named worker
An ask's asker is the last token of its accept subject; the holder checks the kind against the callers the build seat's
definition lists (Puller.Callers, CallersFromDefinitions), refuses an ask of another kind answering error naming the kind
and the asker and ending it failed, takes only a build and a run on main's subject, and gives back an ask the seat list
does not judge yet. A second worker per queue (AskNamedWorker, filtered to accept.<queue>.*) is pulled after main's; the
holder says `named worker not bound` while its membership lists none.

Builds: design 47 G §543, §544, §481, A-163, A-170

Claude-Session: https://claude.ai/code/session_01NcbgfGKaxw984XyAPfLm4h
2026-11-12 05:44:38 +01:00
jschoubben 00b52e3c1f BG4, BN4, BI4: the puller says ask-taken, step-progress and the end of any ask
A holder says ask-taken when an ask starts, step-progress every minute while it runs (phase fetch, build or run) and
step-ended for an ask outside a pipeline, each on the seat's subject of its node (SayOn); the asker is the last token of
the subject the ask came on (Pulled.Subject).

Builds: design 47 G §496, §602, §603, §481

Claude-Session: https://claude.ai/code/session_01NcbgfGKaxw984XyAPfLm4h
2026-11-12 05:33:26 +01:00
jschoubben 3e61b70517 SD5, BG11, BLD10: the build modules' shared code moves into the delivery SDK
All of mesh-build-go's buildstep Go module leaves mesh-catalog for
modules/mesh-delivery-sdk: the step run, fetch and release and the
regenerate-and-compare step as the package steprun, the CPU share as
steprun/buildcpu, the conformance kit as steprun/conformance, the seat
generator as seatgen with cmd/seatgen, and cmd/mesh-pack. The three
build modules and the builder's and checker's tests import the SDK and
keep no copy; the SDK requires nothing of mesh-catalog. Code the build
modules share is the delivery domain's SDK, not one module's package.
It carries the build-module round's two new files of that package
(dropdetected.go 247c422f, storedbuilds.go 904b8aed).

Builds: design 47 G §464, §680, §327, §420, §495

Claude-Session: https://claude.ai/code/session_01NcbgfGKaxw984XyAPfLm4h
2026-11-12 05:18:34 +01:00