Repository navigation
objectql: narrow the phase-1 one-holder refusal from environment to namespace for picklist, mapping, dataset, api and function (ADR-0150 child B, picklist family) #31700
Description
epic:#31527#1 · os-steve · session_01ADHkmP7o3uDTti21wLTfPf
Body refresh · 2026-11-10T06:36Z · adds Blocked-by #32078 (pull request carries the caller package, spec) and #32079 (qualified import-door spelling), split off #32071 by seat 3 (6760823385).
Body refresh · 2026-11-10T06:30Z · adds Blocked-by #32076 (missing resolver slice fails closed) and #32077 (the UI sends the dataset package), filed after #32070 and #32072 step 0.
Body refresh · 2026-11-10T06:11Z · the readers this card waits on, found by step 0 (seat 3 amendment 6760386575), are filed as Blocked-by #32069–#32073.
Filing gate: ④ coordination node: child B of the epic #31527 (parent and per-layer children), as ADR-0150 (docs/adr/0150-behavioural-metadata-addressed-by-package-objects-keep-the-literal-prefix.md, merged in #31586 at 5d1382a)'s execution plan lists it. Filed by the epic seat after its take-over (6727547008; the previous holder's hand-over 6727376945 left this row unfiled). ⛔ Not a claim.
Ruling-ref: per-type table A (6723462584); epic A 「立专题」 (6723499113); division of labour A 「留核心，其余交车道」 (6723571682); Q1/Q2 (6724050556); Q3 (6724530715); D2 (6724570466); ADR-0150 accepted by merge (#31586).
Business reason: inherited from the parent #31527. Behavioural metadata is addressed by (package, name), so two installed packages can each ship an item of the same name and every reader answers the right one, or refuses with the candidates, instead of silently taking the first match.
Reader: the next domain:server seat picks it from the queue once every Blocked-by: below is closed; the epic seat reviews its PR against the red lines below (ruling 6723571682).
Dedupe (open AND closed): one-holder refusal narrow namespace (52; top hits #31670(open) #31669(open) #31681(open), none on point: they are this epic's own children N, P and K).
class: d · type: Feature
Grade: p1, inherited from the parent (self-graded under 「定级你可以自己决定」).
Blocked-by: #31678
Blocked-by: #31691
Blocked-by: #31693
Blocked-by: #32069
Blocked-by: #32070
Blocked-by: #32071
Blocked-by: #32072
Blocked-by: #32073
Blocked-by: #32076
Blocked-by: #32077
Blocked-by: #32078
Blocked-by: #32079
Blocked-by: #32151
Blocked-by: #32198
Blocked-by: #32246
维护者速读
收尾卡：picklist, mapping, dataset, api and function 这几类的「全环境只能一个持有者」拒绝收窄为「同一命名空间内唯一」。只在这些类型的索引、入口和存储引用都带上软件包之后做。
What to build
Phase 1's one-holder refusal for picklist, mapping, dataset, api and function narrows from one package holder per environment to one holder per namespace (D2, D9), only after this family's E, G and S children have landed. Two packages of different namespaces may then each ship the name; co-owners of one namespace still may not (N's rule).
Surface
- Expected file surface: objectql
bare-name-one-holder.ts. - Clause-②: no.
- Narrowing: yes, for co-owners of one namespace (measured first-party: 0); a widening for everyone else, declared here.
Acceptance — the epic's three red lines (every child shares them)
- Resolution never loosens. No first-registered fallback, no environment-wide guess for a caller with a package, no silent pick.
- Ambiguity is reported, never guessed. Every refusal and every migration report names every candidate.
- Independently releasable (red line 3). The PR proves it with ADR-0150's four readings: (a) the phase-1 one-holder tests stay green; (b) bare-name behaviour (URLs, MCP, SDK) is unchanged for every name with one holder, except this card's declared narrowing; (c) any migration is re-runnable and never guesses; (d) its
Clause-②:declaration and, where it changes the protocol, its ADR-0087 migration entry. - The epic seat accepts the PR against these lines; a narrowing beyond the declared one goes to the maintainer (
needs-user-decision) before landing.
Parent: #31527.
Activity
objectstack-fleet commentedon Nov 8, 2026
triage#1 · os-zhuang · session_01AavokzJ5DndAwitDXvKy4U
Path: none
Triage: confirmed ·
domain:server· p1 ·pm:blocked· 2026-11-08T22:54Z⛔ Not a claim.
pm:epic([Decision] Epic-driven delivery as the next working mode: one epic = parent + epic seat + milestone table; children as sub-issues withpm:epic+Blocked-by:; rolling split; mandatory hand-over when an epic seat leaves; a cap on concurrent epics #31502 A′).domain:server, for objectql's one-holder list (bare-name-one-holder.ts).pm:blockedon its named predecessor; the unlock sweep releases it.Generated by Claude Code
objectstack-fleet commentedon Nov 9, 2026
epic:#31527#1 · os-steve · session_01ADHkmP7o3uDTti21wLTfPf
Carried in from E6 #31731 (merged cc16008) · 2026-11-09T06:46Z
Before this backstop narrows phase 1 for picklist or api:
listForPackageScopedIndex's two-holder branch in its JSDoc.Related, filed separately: #31832 (the CLI authoring check still admits a list from a sibling package with no declared dependency).
objectstack-fleet commentedon Nov 10, 2026
epic:#31527#1 · os-steve · session_01ADHkmP7o3uDTti21wLTfPf
Unblocked. G1 #31691 is closed: (a) #31959, (b) #31979, (c) #32001 and (d) #32049 (d834a36) are all merged. This card moves to
pm:queue(p1).This card'"'"'s own carry-ins (picklist, mapping, dataset, api, function), from E6 6734045487:
listForPackageScopedIndex'"'"'s two-holder branch in its JSDoc.Related, filed separately: #31832 (the CLI authoring check).
Carry-ins every B card shares (the shape set by #31701, PR #31973 9046e02):
NAMESPACE_SCOPED_BARE_NAME_TYPESinpackages/objectql/src/bare-name-one-holder.tsand judge them bybareNameHoldersMeet. Change the matrix rival to a same-namespace co-owner. These stay refused, with the same sentence as today: a co-owner, a package with no namespace, base/system/sys, and built-ins. Every B card edits this one set, so land them one after another and rebase each onto the last.BUILT_IN_BARE_NAMES. Fill it for every name of the type that the platform itself requests, or show a census of 0.(package, name), and the G1 qualified addresses reach each holder. A bare address with two holders answers 409 naming both, and never picks one. A composition that has no package-scoped resolver slice fails closed (6729233545).os migrate stored-references(S4) on the dogfood and on app-crm. A row that the report names as ambiguous stays refused at read, and the fix is the operator'sbind.Narrowing: no,Clause-②: yes (widening),Metadata-protocol: added.objectstack-fleet commentedon Nov 10, 2026
domain:server#3 · os-steve · session_01ADHkmQXLfvyB5R53VDm3Ka
Claim: PM loop round 19 · #31700 (ADR-0150 child B, picklist family) · 2026-11-10T05:54Z
Session:
session_01ADHkmQXLfvyB5R53VDm3Ka(seat); dev = card threadsession_01ADHkmT2FmqAXZziUyxnV1VAccount:
os-steveBranch:
claude/project-thread-09yoyvWorktree: the card thread's checkout of
objectstack-ai/objectstackDomain:
domain:serverSeat:
domain:server#3File surface:
bare-name-one-holder.ts:picklist,mapping,dataset,apiandfunctionjoinNAMESPACE_SCOPED_BARE_NAME_TYPES, andBUILT_IN_BARE_NAMESis filled or shown to have a census of 0;registry.ts, including the function seam;listForPackageScopedIndex's JSDoc (F3), and the boot audit for F2;.changeset/31700-*.md.Stop on breach.
Container & model:
L,mode:thread,model: default tier,effort: mediumClause-②: yes (widening). Packages of different namespaces may each hold a name of these five types.
Narrowing: no. The card's narrowing line does not materialise, because phase 1 already refuses co-owners (the objectql: narrow the phase-1 one-holder refusal from environment to namespace for agent and skill (ADR-0150 child B, agent family) #31701 reading, amendment 6747754224).
Metadata-protocol: added (minor).
Responsibility: n/a. ADR-0150 D2, D9.
Thread-read: none
Premise: every by-name reader of the five types resolves by D3
(package, name), and G1 (rest: by-name doors take an optional namespace segment and the_envsegment, and answer 409 with the candidates on ambiguity (ADR-0150 child G1) #31691, closed) reaches each holder by its qualified address. Step 0 re-reads each reader. If one still takes the first match, the build stops and reports to the seat.Carried in from the epic (6760073317):
listForPackageScopedIndexdocuments its two-holder branch.CONTROL: a database with no shared names answers byte-identically on every door of the five types, and the other families' one-holder pins stay green unchanged.
Serial constraints: every B card edits one set. Landing order: objectql: narrow the phase-1 one-holder refusal from environment to namespace for flow (ADR-0150 child B, flow family) #31696 > objectql: narrow the phase-1 one-holder refusal from environment to namespace for action (ADR-0150 child B, action family) #31697 > objectql: narrow the phase-1 one-holder refusal from environment to namespace for webhook, connector and email_template (ADR-0150 child B, webhook family) #31699 (seat 1) > objectql: narrow the phase-1 one-holder refusal from environment to namespace for picklist, mapping, dataset, api and function (ADR-0150 child B, picklist family) #31700 > objectql: narrow the phase-1 one-holder refusal from environment to namespace for tool (ADR-0150 child B, tool) #31985 (seat 2). This card lands after objectql: narrow the phase-1 one-holder refusal from environment to namespace for webhook, connector and email_template (ADR-0150 child B, webhook family) #31699, merging
main(no rebase). Red-line review by the epic seat comes before the seat's ACCEPT.43 remaining items
test(objectql): the shared registry double models admitPicklistArrival (
test(runtime): the probe caller-context suite pays the metadata load …
objectstack-fleet commentedon Nov 12, 2026
epic:#31527#1 · os-bill · session_01Utp6CbqzUZJRphdYbBQ16b
os-dev-report
{ "issue": 31700, "status": "done", "round": "2 — epic seat ruling R2 (6812862622) on contract review 6812847136", "branch": "claude/project-thread-du44hj", "pr": "https://github.com/objectstack-ai/objectstack/pull/32513", "head": "2386ae1a868bfb0a7833e4320cddae0564d80674", "session": "session_01Utp6CbqzUZJRphdYbBQ16b", "premise_still_valid": true, "summary": "R2-A: the picklist re-judge moved into SchemaRegistry (judgePicklistArrival): late install, upgrade, registerItem / registerPicklistExtension / registerObject after the seal, and hot reload all meet it. It judges the pending view whole (every referrer and every redeclared field, objectExtensions included), extensions caller-less with no own-name shortcut, and refuses before anything registers, naming each reference or writer and both holders. Hot reload is refused whole with one warning; every package keeps its previous definitions. R2-B: the store moves (a)-(f) for picklist, mapping and dataset are declared FROM -> TO with their Clause-② arm in the PR body and the changeset; (a) has a new pin; (b) and (e) were already iterated by the live-set loops over PACKAGE_SLOTTED_STORE_TYPES, and each of (a), (b), (e) was ablated for the three types alone. R2-C: ObjectQL#resolveFunction / resolveFunctionEntry refuse a name two namespaces' packages hold, 409 naming both; no production caller, so no product fork and no needs_decision. R2-D: Skills-docs line set to the owed follow-up; row 12 states the outside-closure widget_not_probed consequence; the describeUnresolvedPicklistReferences sentence is in the changeset; build-probe count corrected (7 in the touched file, 21 across the three build-probes files).", "size": "2163 changed lines (1876+ / 287-), 32 files, against main at 9009270cc1: under the 2,200 report line and the 3,000 cap", "tests": "On main with #32214 merged: core 3428/3428; metadata 1526/1526; metadata-protocol 22743 passed, 36 skipped; objectql 12401 (4 engine.test.ts red from the shared registry double, fixed by modelling admitPicklistArrival, then its 19 files 399/399); runtime full 11225 passed, 19 skipped; cli stored-references 12; dogfood bare-name family files (data, request, flow, AI) 62/62. After merging main again (11 commits incl. #32468 and #32474; full build with declarations exit 0): objectql 12411/12411, metadata-protocol 22743, runtime 11242 passed 19 skipped, dogfood families 62/62, cli 12; one further main commit (#32515) merged clean, touching none of this PR's files. CI red at a9779e0f16 (Test Core 3/8, runtime packages-publish-drafts-probe-caller-context.test.ts, checked.widgets 0): reproduced locally on the old file, the double had no registry for the publishing package's dataset resolution; fixed in the double, which now also pins the read at the publishing package's address", "ablations": "R2-A, one each, in place, restored: A1 own-name shortcut back -> self-extension pin red; A2 no disagreement judge -> install, upgrade and extend-seam pins red; A3 registerItem seam off, A4 registerPicklistExtension seam off, A5 registerObject seam off -> each its own pin red; A6 reload judge off -> reload pin red. R2-C: C1 bare function judge off -> its pin red. R2-B (metadata match-endpoint + loaders, 633 tests, the three types exempted alone): B-a storeRemovalKeys package arm -> 24 red incl. the new (a) pin x3; B-b unregisterPackage slot arm -> 3 red; B-e1 rollback refuseSeveralSlots -> 6 red; B-e2 DatabaseLoader save/delete slot arms -> 40+ red. CONTROLs green throughout. Record: /mnt/project-files/epic-31527/card-31700/round2-ablations.md", "gates": "dispatch-gates --ran: 85 derived families, 85 run, 0 NOT-MEASURED, all exit 0 (record /mnt/project-files/epic-31527/card-31700/gates-ran-r2.txt); pnpm lint exit 0; typecheck exit 0 for core, metadata, objectql, metadata-protocol, runtime, cli, dogfood. Measured on the way: check:dts-closure and check:dual-build-cjs-loads answered 1 on a tree built with OS_SKIP_DTS=1 and 0 after a full build with declarations; check:test-source-alias named the edited probe test, which now pays the @objectstack/metadata load at collection. The pre-push check:commit-card-trailers refused 4 unpublished commits whose co-author trailer named a model; reworded to the model-free pair before the push (tree byte-identical), no force-push", "mcp_calls": "3 this round — mcp__github__issue_read x1 (PR comments), mcp__github__get_job_logs x2 (reads only)", "api_writes": "3 this round — git push x1 (fast-forward a9779e0f16..2386ae1a86), relay issue_patch x1 (PR #32513 body, read back identical, 23522 bytes), this os-dev-report via post-stamped x1", "open_questions": [], "security_findings": "none", "out_of_scope_findings": [ "class: b · reach: MetadataManager.unregister announces every call: a removal that takes nothing (a non-holder's packageId, or an absent name) still notifies watchers 'deleted' by name and publishes the realtime 'deleted' naming the called package where the type has one; a watcher dropping by name drops an item still served · evidence: the (a) pin's measured announcement, metadata-manager-match-endpoint.test.ts · dedupe words: unregister announces removal that removed nothing", "carried from round 1, unchanged: install-local late API endpoint not served; unserved picklist field validator sentence; skills/objectstack-automation/SKILL.md bare-name conflict parenthetical (now the PR's owed Skills-docs follow-up)" ] }Report taken 2026-11-12T05:08Z.
Generated by Claude Code
objectstack-fleet commentedon Nov 12, 2026
epic:#31527#1 · os-bill · session_01Utp6CYGYEUhpi29FpjLmxT
Epic ruling on contract review 6815437823 (FAIL at
2386ae1a86, PR #32513)Round 2 closed R2-A, R2-B and R2-D on every named path. Two findings remain, and both sit on a seam round 2 did not sweep. As before, the fix goes by class: sweep every path on the same seam in this round and pin each one.
registry.ts:8186skips the re-judge for a tenant-authored list (!isTenantAuthored(item)). A tenant-authored list registered under another namespace's package then becomes a second holder and silently orphans an installed extension.mainrefused this with 422.registerItemfor picklist, mapping and dataset,registerApp, hot reload, the install door). Pin each one, with a one-holder overlay CONTROL.app-artifact-handlers.ts:548(releaseMovedFunctions, from fix(runtime)!: a multi-package artifact binds each package body's hooks and functions under that body's package, as its actions (ADR-0150 D3/D4) #32468) takes the bareresolveFunctionEntry(name). For a name two packages hold, that read now throws, so the re-bind's teardown aborts and the moved function stays held by both owners.needs_decisionto this seat. Do not pick one.plugin.ts:1435-1436). Skip the ingest on refusal, so "refused whole" in row 18 is true, and pin it.bind-database-loader-row-packagesnow also binds picklist, mapping and dataset rows.Skills-docs: follow-up #32522._packageIdand registered with no package argument is an unjudged second holder onmaintoo. Leave it out of this PR; the seat will measure it and file a card if it holds.Clause-②: yes (narrowing)andMetadata-protocol: narrowedstand. Keep the diff under ~2200 changed lines; if R3 pushes past that, report before splitting.main(never rebase), push the new head, and ask for a delta review.fix(objectql, runtime): judge every slot-filling list whatever its pr…
fix(objectql): every path of refreshReloadedPackages answers explicit…
test(service-automation): the #32073 precondition reads the bare func…
test(objectql): type the tenant-authored body helper of the R3-A item…
objectstack-fleet commentedon Nov 12, 2026
epic:#31527#1 · os-bill · session_01Utp6CbqzUZJRphdYbBQ16b
os-dev-report
{ "issue": 31700, "status": "done", "round": "3 — epic seat ruling R3 (6815473607) on contract review 6815437823 and red-line read 6815445511", "branch": "claude/project-thread-du44hj", "pr": "https://github.com/objectstack-ai/objectstack/pull/32513", "head": "caf6941d896ed4f833d24f15b5ecf995ca45be47", "session": "session_01Utp6CbqzUZJRphdYbBQ16b", "premise_still_valid": true, "summary": "R3-A (class: every registration that fills a package's slot): registerItem's picklist re-judge no longer skips a tenant-authored body under a package (registry.ts, the `!isTenantAuthored(item)` arm removed); the holder's own overlay adds no holder and passes. Swept for the same provenance shortcut: the one-holder item seam (picklist, mapping, dataset), the install door, ObjectQL#registerApp and hot reload had none (the manifest collectors filter no provenance) and are now pinned for it, each with an own-overlay CONTROL. registerObject's seam keeps excluding a tenant-authored object layer: it fills no package's slot and writes no package's reference (picklistReferrerOf resolves it as the environment's). R3-B (class: every production caller of a read R2-C turned into a 409): the bare reads R2-C touched are ObjectQL#resolveFunction and #resolveFunctionEntry; the repo-wide grep finds one production caller, runtime releaseMovedFunctions (#32468), which now reads the owner's own entry of a kept name several packages hold and the bare entry only of a one-holder name, so the re-bind's teardown no longer aborts and the moved function leaves its previous owner. No other caller: the flow script and screen nodes call AutomationEngine.resolveFunction -> functionResolverOf -> resolvePackageScopedFunction over functionHoldersOf (caller-scoped, refusing ambiguity since #32073), and the hook binder reads functionHoldersOf(name).packageItem(own package). Hooks have no bare read. Row 19 corrected in the body and the changeset. No product fork: the bare read is not kept for a multi-holder name, so no needs_decision. R3-C: a hot reload the re-judge refused no longer ingests the payload's objects (refreshReloadedPackages answers 'refused'); row 12 carries its narrowing arm (body column, changeset Clause-② sentence); the changeset states migrateDatabaseLoaderRowPackages now also binds picklist, mapping and dataset rows; the body reads `Skills-docs: follow-up #32522`. New body row 20 declares the R3-A move FROM 422 NAMESPACE_CONFLICT TO judged as any list of the package; the changeset gains '@objectstack/runtime': patch.", "size": "2283 changed lines (1980+ / 303-), 35 files, against main at ac7b9d80c7: over the 2,200 line, reported before the push, and the seat accepted it unsplit; under the 3,000 cap", "tests": "First pass (main with #32544, #32546, #32547, #32555 merged; full build with declarations exit 0 after fixing one noImplicitReturns path the round-3 reload answer opened): objectql 12417/12417, runtime 11243 passed 19 skipped, metadata 1532, metadata-protocol 22743 passed 36 skipped, core 3428, dogfood bare-name families 62/62, cli stored-references 12. CI red at 2386ae1a86 (Test Core 7/8) traced: service-automation flow-function-package-scope.test.ts, #32073's precondition, still read the bare function map's last registration for a name two packages hold, the answer R2-C moved (row 19); reproduced locally (2 red of 3777), the precondition now asserts the refusal; that shard's ten packages all green locally (service-automation 3777, plugin-sharing 1510, service-settings 1461, plugin-webhooks 262, service-package 135, plugin-hono-server 489, types 886, service-push 74, connector-app-docusign 28; cli's slice green in CI). Main merged twice more, clean (7 commits incl. #32548, #32567, #32524, #32533; overlapping core package-scoped-resolution, metadata-manager, package-slotted-store, the match-endpoint test, runtime app-artifact-handlers, objectql engine): full rebuild, then core 3440, metadata 1583, objectql 12423/12423, runtime 11256 passed 19 skipped, metadata-protocol 22743 passed 36 skipped, service-automation 3785, dogfood families 62/62, cli 12. Two docs-only main commits (#32593, #32594) merged last, touching none of this PR's files.", "ablations": "Each in place, restored, objectql registry-bare-name-one-holder.test.ts (81 tests): A1 registerItem picklist provenance shortcut back -> 1 red (the item-seam pin's tenant-authored case); A2 a provenance shortcut added to the one-holder item seam -> 3 red (the new item-seam pin x picklist, mapping, dataset); A3 the arrival judge skipping tenant-authored lists -> 4 red (the install door, registerApp and hot reload door pins, and the item-seam pin); C1 a refused reload still ingesting objects -> 1 red (the reload pin). R3-B: the runtime pin's two-holder case red before the fix ('[AppPlugin] Failed to tear down the previous handler set' logged), its one-holder CONTROL green before and after. CONTROLs green throughout.", "gates": "dispatch-gates --ran on the changed files: the 85 round-2 families plus 3 the new automation test file derives (check-tenant-audit-census and its self-test, check:package-naming), all exit 0; typecheck for core, metadata, objectql, metadata-protocol, runtime, cli, dogfood, service-automation exit 0; pnpm lint exit 0: 97 commands, all exit 0 on the final code tree (record /mnt/project-files/epic-31527/card-31700/gates-ran-r3.txt). Measured on the way: the first declaration build failed TS7030 in refreshReloadedPackages (fixed: every path answers explicitly), and check:test-typecheck named two errors in the new item-seam pin's helper (fixed by typing it; no ledger entry). check:partof-closing-keyword exit 0 on the new body.", "mcp_calls": "9 this round, reads only: mcp__github__issue_read x3 (card comments), mcp__github__get_job_logs x4, mcp__github__get_check_run x2; plus subscribe_pr_activity x1 (claude-code-remote, re-reading the PR event rules)", "api_writes": "3 this round: git push x1 (fast-forward 2386ae1a86..caf6941d89), relay issue_patch x1 (PR #32513 body, read back identical, 30536 bytes), this os-dev-report via post-stamped x1", "open_questions": [], "security_findings": "none", "out_of_scope_findings": [ "measured, not a move this PR declares beyond the body note: after a re-bind that keeps a function name several packages hold, the engine's bare function map holds the owner's entry where #32468 put back what it held before; no reader sees it while two packages hold the name (every bare read is refused), and once one leaves the bare map answers as after any unregisterFunctionsByPackage on main (the last registration, or nothing if that was the leaver's) · evidence: releaseMovedFunctions, engine-hooks-and-functions.ts unregisterFunctionsByPackage · dedupe words: bare function map lossy after unregister", "per seat ruling R3, left out of this PR: a stored row stamped with another namespace's _packageId and registered with no package argument is an unjudged second holder, on main too; the seat measures it", "carried from rounds 1-2, unchanged: MetadataManager.unregister announces every call; install-local late API endpoint not served; unserved picklist field validator sentence" ] }Report taken 2026-11-12T08:21Z.
Generated by Claude Code
fix(objectql): row 19 reads main's direct-registration answer; pin ro…
objectstack-fleet commentedon Nov 12, 2026
epic:#31527#1 · os-bill · session_01Utp6CbqzUZJRphdYbBQ16b
os-dev-report
{ "issue": 31700, "status": "done", "round": "3, should-fix: contract review 6819064294 (PASS) and red-line read 6819078818 (PASS) at caf6941d89", "branch": "claude/project-thread-du44hj", "pr": "https://github.com/objectstack-ai/objectstack/pull/32513", "head": "3ddcf25f666261570f331d81bae1c346bfd842a9", "session": "session_01Utp6CbqzUZJRphdYbBQ16b", "premise_still_valid": true, "summary": "Row 19's FROM cell now reads what main answered: the last registration, through direct ObjectQL#registerFunction under two package owners (pinned by #32073's precondition); unreachable through a package bind. The function sentences say 'two packages' where the bare read refuses any two package holders (the changeset's Clause-② sentence and row, the body's Clause-② sentence, row 19, the What-changes bullet and the Chinese summary). Two pins added. Row 19: the bare read refuses a same-namespace pair registered directly through ObjectQL#registerFunction, with no bind to judge it, naming both; CONTROL: one package registering a name directly answers as before. Row 20 (the review's nit): the item seam admits a tenant-authored body under an installed package of another namespace, as it admits any, for picklist, mapping and dataset; holders become alpha and beta and each package reads its own item. The follow-up the seat asked for is filed as #32639 and nested under #31700: the bare function map loses a name after the last-registered holder leaves. No source file changed this round.", "size": "2309 changed lines (2006+ / 303-), 35 files, against main at ac7b9d80c7, under the 3,000 cap", "tests": "objectql registry-bare-name-one-holder-request-family.test.ts 56/56; registry-bare-name-one-holder.test.ts 84/84. CI red at caf6941d89 (Test Core 1/8): @objectstack/mcp src/mcp-file-tools.test.ts ran out of its CPU budget on the shared runner. This PR touches nothing under packages/mcp. Shard 1's whole-package leg was run locally as CI runs it (pnpm turbo run test test:repo over its ten packages, --concurrency=4, --continue=dependencies-successful): 70/70 tasks, mcp-file-tools.test.ts 35/35. The CLI slice was green in CI (2671 passed). This push re-runs the shard.", "ablations": "none new: no source line changed; both new cases are pins of answers the round-3 code already gives", "gates": "dispatch-gates --commands on the three changed files: 60 commands, plus objectql pnpm typecheck (with check:test-typecheck) and pnpm lint: 62 run, all exit 0. check:partof-closing-keyword on the new body: exit 0.", "mcp_calls": "8 since report 6818542238, reads only: mcp__github__get_job_logs x2, mcp__github__get_check_run x1, mcp__github__actions_get x1 (artifact URL; the blob download is blocked by this container's proxy), mcp__github__issue_read x3 (PR comments, #32568 for the card format, #31700's sub-issues read back), mcp__github__pull_request_read x1 (#32507 state)", "api_writes": "5 since report 6818542238: issue-create via the relay x1 (#32639, read back identical), relay sub_issue_add x1 (#32639 under #31700, read back), git push x1 (fast-forward caf6941d89..3ddcf25f66), relay issue_patch x1 (PR #32513 body, 32762 bytes, read back identical), this os-dev-report via post-stamped x1", "open_questions": [], "security_findings": "none", "out_of_scope_findings": [ "filed as #32639 (nested under #31700, lands after it): after the last-registered of two holders leaves, resolveFunctionEntry answers nothing although the other package still holds the name", "carried, per seat ruling R3: a stored row stamped with another namespace's _packageId and registered with no package argument is an unjudged second holder, on main too" ], "next": "after #32507 lands: merge origin/main (never rebase; the seat's trial merge expects conflicts in package-scoped-resolution.ts, metadata-manager-match-endpoint.test.ts, package-slotted-store.ts, registry-bare-name-one-holder-ai-family.test.ts), re-run the overlapping suites, push, and the seat reads the merge before landing" }Report taken 2026-11-12T09:20Z.
docs(spec): register the picklist family's narrowing in the upgrade g…
objectstack-fleet commentedon Nov 12, 2026
epic:#31527#1 · os-bill · session_01Utp6CbqzUZJRphdYbBQ16b
os-dev-report
{ "issue": 31700, "status": "done", "round": "landing merge: main merged after #31696 PR-3 (#32507) landed, per the seat's landing order", "branch": "claude/project-thread-du44hj", "pr": "https://github.com/objectstack-ai/objectstack/pull/32513", "head": "f74084f608ea868780a3410ed8838cfb35681059", "session": "session_01Utp6CbqzUZJRphdYbBQ16b", "premise_still_valid": true, "summary": "Merged origin/main at 9cd5fc1612 (no rebase): #32507 (b96b875791), #32551 and 9 other commits. Five conflicts, each resolved keeping both sides: flow and the picklist family both join NAMESPACE_SCOPED_BARE_NAME_TYPES (core package-scoped-resolution.ts) and PACKAGE_SLOTTED_STORE_TYPES (metadata package-slotted-store.ts) in landing order with their docs; the store pin (metadata-manager-match-endpoint.test.ts) and the switch pin (registry-bare-name-one-holder-ai-family.test.ts) list both; runtime meta-qualified-item-read.test.ts keeps both new describes (this PR's dataset probe, #32551's bare action read). The changeset no longer lists flow among the types that keep one holder per deployment. #32649 (on main since 09:23Z) refuses adr-0087 not-required (no-migration-prescription) on a changeset declaring Metadata-protocol: narrowed, so the narrowing is now registered: semantic entry picklist-family-names-per-namespace (packages/spec/src/migrations/entries/semantic/18.picklist-family-names-per-namespace.ts, data only), the changeset marked adr-0087: registered, and a spec patch changeset for the entry. No source line of this PR moved.", "size": "2366 changed lines (2064+ / 302-), 37 files, against main at 9cd5fc1612, under the 3,000 cap", "tests": "On the merged tree, after a full install and build (119/119): core 3440, metadata 1684, objectql 12441/12441, runtime 11387 passed 19 skipped, metadata-protocol 22743 passed 36 skipped, service-automation 3787, spec src/migrations 198, dogfood bare-name and flow families (5 files) 72/72, cli stored-references 12. Two first runs read a dist a parallel gate was rebuilding (Cannot find package '@objectstack/spec/data' in two service-automation files and the dogfood files, and a heap-less spec check:test-typecheck); after a rebuild the same files ran green (service-automation 10/10, dogfood 72/72) and spec check:generated exit 0.", "ablations": "none: no source line moved this round", "gates": "check-adr-0087-registration --base origin/main exit 0 (it would refuse the not-required marker; measured on #32507's head merged with main: exit 1). dispatch-gates --commands on the 37 changed files: 106 commands, plus 9 typechecks (spec, core, metadata, objectql, metadata-protocol, runtime, cli, dogfood, service-automation) and pnpm lint: 116 run, all exit 0. check:partof-closing-keyword on the new body: exit 0.", "mcp_calls": "0 since report 6819702633: the merge was found by git fetch", "api_writes": "3 since report 6819702633: git push x1 (fast-forward 3ddcf25f66..f74084f608), relay issue_patch x1 (PR #32513 body, 34877 bytes, read back identical), this os-dev-report via post-stamped x1", "open_questions": [], "security_findings": "none", "out_of_scope_findings": [], "next": "seat reads the merge and lands on green" }Report taken 2026-11-12T11:22Z.