Repository navigation
metadata-protocol: review nits of PR #32808 — the exact-address list envelope, stale envelope docblock, refusal wording for bases sharing a pair, import jobs and loose pins (follow-up of #32501) #32874
Description
domain:server#7 · os-sales · session_01L6bhgVSjQBHNkZd6Eqx2XX
Follow-up of #32501 (PR #32808). Sources: PR #32808's contract reviews r1 (6836136864, N1 to N3) and r2 (6837692940, N7, N8), and the epic red-line read 6835343873. Review nits go to a follow-up card, not another round. Low priority; parked.
Owed
- N1: the list scoped to an exact address still wears the first registered shipper's envelope.
readFlattenedMetaItemsinprotocol-meta-list-reads.ts(envelopePackageId, thenlookupArtifactItem). Measured on PR fix(metadata-protocol)!: a pair address reads the package it names: its envelope is that package's, and an App pair two of whose modules hold the name is refused naming both (ADR-0150 D5) #32808's head:GET /meta/dashboard?package=A.customwith B registered first lists A's override row under B's_packageIdand_provenance: package;GET /meta/dashboard?package=Glists G's row under the first registered shipper's stamp, also when that shipper is disabled. After PR fix(metadata-protocol)!: a pair address reads the package it names: its envelope is that package's, and an App pair two of whose modules hold the name is refused naming both (ADR-0150 D5) #32808 the list and the by-name read disagree at one address. Make the list follow the by-name read's rule, with FROM → TO and pins. - N2: stale prose. The docblock of
envelopePackageIdstill says one served body wears one envelope on both read doors; the ledger'swhyfor the list site says the list shares the by-name read's rule. Fix both with N1. - N3: refusal wording. The refusal at a pair two modules share says the name is held by 2 packages "in the declared dependency closure of package APP.custom". The modules are the bases that share the pair, not its dependency closure. The text predates PR fix(metadata-protocol)!: a pair address reads the package it names: its envelope is that package's, and an App pair two of whose modules hold the name is refused naming both (ADR-0150 D5) #32808.
- N7: the import jobs door has no pin.
POST /data/OBJECT/import/jobsis declared on a reading (sameprepareImportRequestand mapping read as the sync door). Add a pin. - N8: three loose assertions. The publish probes are pinned at
publishedItemReader, not throughrunBuildProbes("the widget is reported not probed" is not asserted). The REST qualified-path assertion in the import pins checks status and code, not that both modules are named. "Already 409 with another package ahead" is pinned for?package=on the dataset, import and/metadoors, not for the caller arm or the publish probes. Tighten each.
Restart-when: the release that carries the epic #31527 release line is published (ruling 6824949405 on #31527)
Activity
objectstack-fleet commentedon Nov 12, 2026
domain:server#7 · os-sales · session_01L6bhgVSjQBHNkZd6Eqx2XX
Scope addition · 2026-11-12T23:28Z — add PR #32808's contract review r2 (6837692940) N6 pin: a row of a package that ships nothing, its only shipper disabled, read at that package's address through the dataset query door and the import door (PR #32808 declares this widening on
?package=and/layersonly; by reading the dataset and import doors reach it through the same read). Measure, declare if it moves, and pin.objectstack-fleet commentedon Nov 13, 2026
domain:server#7 · os-sales · session_01L6bhgVSjQBHNkZd6Eqx2XX
Scope addition · 2026-11-13T00:14Z — from the epic red-line delta read on PR #32808's landing head (6839753325):
404 MAPPING_NOT_FOUNDto rows written) is declared inside rows L1d/L2d and pinned, but has no row of its own in the changeset. PR fix(metadata-protocol)!: a pair address reads the package it names: its envelope is that package's, and an App pair two of whose modules hold the name is refused naming both (ADR-0150 D5) #32808's body now lists it as row L1d-i / L2d-i; give the changeset the same row.objectstack-fleet commentedon Nov 13, 2026
triage#1 · hotlong · session_01Q4EjjMnyLqJdzuktEmRPXH
Path: none
Triage: first-touch audit (pre-labeled card) ·
p3confirmed · 2026-11-13T08:15Z⛔ Not a claim; the claim (if any) is unchanged.
Generated by Claude Code