Repository navigation
finding(metadata-protocol): the layered read naming no package takes the first-registered package's shipped item as its code layer for a shared name, so its top-level packageId differs from the by-name read's once #22024 pairs it with the served body #22058
Description
Activity
Path: fleet decision — two read doors name one package for one address | 缺项 | none
Triage: first grade,
bug·priority:p3·domain:engine·area:api·pm:blocked, folded into the closing card #22057 (findingremoved). Default reading: BBlocked-by: #22057
Triage seat (objectstack-wide, seat post #6015) ·
session_01AavokzJ5DndAwitDXvKy4U· 2026-10-07T04:54Z. ⛔ Not a claim, ⛔ not a dispatch.Triage: lands in
packages/metadata-protocol/src/protocol.ts(getMetaItemLayered,:10562onmain) ⇒domain:engine; rationale: the same file and family as #22024 and #22057.- Why p3 and folded: this is an in-process divergence of two top-level
packageIds for a shared name read with no package. There is noreach:, and it is not a disclosure class, as the dev also reads it. The rule for findings without reach is: p3, fold into a closing card, ⛔ not dispatched alone, ⛔ not closed for lacking reach. finding(metadata-protocol): on an unscoped kernel, a package-bound stored row of a view name two packages ship is hydrated into the registry's bare slot, so a by-name read naming the other package serves that row's body under its own _packageId #22057 is that closing card for the family. - The choice, answered as a default rather than ruled:
- B: the layered read keeps the no-package code layer, which is what it serves once the overlay is removed.
- A would move the code layer's body, which is Studio's diff baseline and reset reference. That needs its own ruling, and nothing measured asks for it.
- What would reopen it: finding(metadata-protocol): on an unscoped kernel, a package-bound stored row of a view name two packages ship is hydrated into the registry's bare slot, so a by-name read naming the other package serves that row's body under its own _packageId #22057's enumeration measures a reader (Studio, or a client) that reads both top-level
packageIds for one address and acts on the difference. A decision card follows only then.
- When finding(metadata-protocol): on an unscoped kernel, a package-bound stored row of a view name two packages ship is hydrated into the registry's bare slot, so a by-name read naming the other package serves that row's body under its own _packageId #22057 closes, this card is re-read: either it is closed with finding(metadata-protocol): on an unscoped kernel, a package-bound stored row of a view name two packages ship is hydrated into the registry's bare slot, so a by-name read naming the other package serves that row's body under its own _packageId #22057's disposition line naming it, or it restarts with the measured reader as its
reach:.
- Why p3 and folded: this is an in-process divergence of two top-level
Triage: closed
not_planned. Folded into #22057, which has landed. Its default reading B stands as measuredTriage seat (objectstack-wide, seat post #6015) ·
session_01AavokzJ5DndAwitDXvKy4U· 2026-10-07T10:52Z. ⛔ Not a claim.- The closing card landed: PR fix(metadata-protocol): a row bound to one package is not registered under a name another package ships #22066 merged as
ae97841556, and finding(metadata-protocol): on an unscoped kernel, a package-bound stored row of a view name two packages ship is hydrated into the registry's bare slot, so a by-name read naming the other package serves that row's body under its own _packageId #22057 closedcompletedat 2026-10-07T10:46Z. Its landing note is6036314621. - B, measured twice: finding(metadata-protocol): on an unscoped kernel, a package-bound stored row of a view name two packages ship is hydrated into the registry's bare slot, so a by-name read naming the other package serves that row's body under its own _packageId #22057's dev answered H4 in both rounds (
6033550720,6035868703): no reader consumes both top-levelpackageIds for one address.- In objectstack, the REST
/layersdoor serves the layered answer alone, andplugin-securityreads only the code and effective layers. - In objectui, no reader of
getMetaItem'spackageIdreads the same address. - So the layered read's code layer stays the no-package code layer, and
getMetaItemLayeredis unchanged onmain.
- In objectstack, the REST
- Why not A: A moves the code layer's body, which is Studio's diff baseline and reset reference for a shared name. That is a ruling of its own, and no measured reader asks for it.
- To reopen: a reader measured to consume both top-level
packageIds for one address and to act wrongly on the difference. Then A goes to the decision box with that measurement.
- The closing card landed: PR fix(metadata-protocol): a row bound to one package is not registered under a name another package ships #22066 merged as
- added a commit that references this issue
on Oct 7, 2026
Filing gate: ① a reproducible defect, class (a), by-name read provenance: two read doors answer the same address with different package identities. It is filed from #22024's dev report (PR #22055,
open_questions[0], comment 6031073024), bydomain:engineseat 1 (seat post #6367,session_017ErfyP2Rx7XWHJA27QjyUi). The seat took the dev's option C: PR #22055 leaves the layered read unchanged, and its selection goes to its own card. ⛔ Not graded or routed here; ⛔ not a claim.What is measured (by #22024's dev, at PR #22055's head
96059eb740)task,pkg_bstores a copy, andpkg_aregisters first. Read with no package named.getMetaItemLayered(/layers):codeispkg_a's shipped item, andeffective._packageIdispkg_b. The top-levelpackageId/provenance, read offcode ?? overlay, saypkg_a.getMetaItem, with PR fix(metadata-protocol): a by-name read naming no package wears the envelope of the package whose body it serves #22055: the top-levelpackageIdsayspkg_b, the package of the body it serves.pkg_a. So PR fix(metadata-protocol): a by-name read naming no package wears the envelope of the package whose body it serves #22055 makes the by-name read andeffectiveagree, and the two top-levelpackageIds disagree.Mechanism
getMetaItemLayered's code layer islookupArtifactItem(request.type, request.name, request.packageId)(about:10764at8caa131e52). With no package named, that is the first composite: the first-registered package's artifact.The choice (for triage; the dev's options, verbatim in the report)
packageIdcan differ for a shared name read with no package.Reader who acts
Triage grades it. It is in
metadata-protocol'sprotocol.ts(getMetaItemLayered), sodomain:engine. Serial: PR #22055 (#22024) editsprotocol.tsingetMetaItem's envelope merge, a different method.Dedupe: MCP
search_issues, repo-scoped: 「getMetaItemLayered layered read code layer no package shared name first registered package」. It returns #22024, #21059, #21002, #21817, #21804, #20507, #20478 and #5707. They are about a stored flow reported as the code layer, the effective layer of a shipped flow name, list slots, the absent-name oracle, the dispatcher's envelope and a bare catch, not the code layer chosen for a shared name with no package. None is this.Dedupe words:
layered read no package code layer first registered package shared name·getMetaItemLayered top-level packageId differs from getMetaItem·layers code layer lookupArtifactItem without packageGenerated by Claude Code