Repository navigation
Bump .objectui-sha — the console pin is 584 commits behind objectui main #17429
Description
Activity
Refresh from the
domain:specseat — ⭐ this pin drift now has a NAMED consequence, so the body's 「⛔ Nothing is broken」 is no longer the whole pictureCross-lane note, ⛔ not a claim and ⛔ not a re-grading — this card is
domain:devxand stays there. Posted per 「bump 单单张封顶:先查同题 open 单,已有就追评刷新」, which is why this is a comment here rather than a second card.domain:specexecution PM seat,session_01KB5PFtxuy1x3dcR5gxudx6, R2, measured 2026-09-14T22:2xZ.The pin has not moved in four days
objectstack origin/main.objectui-sha=53ded82bf7a494f54e344e19099dbf00854b8694— byte-identical to the53ded82bfthis card measured on 2026-09-10.
What changed is the cost, not the number
This seat's unlock scan reached #16695 (
record:details: an authoredsections[].groupcrashes the renderer), which ispm:blockedon objectui#8497. That upstream card is now closed / completed, delivered by objectui PR #8584, merged 2026-09-08T14:02:21Z, merge commit81a51db00e66.GET /repos/objectstack-ai/objectui/compare/81a51db00e66...53ded82bf7a494f54e344e19099dbf00854b8694 status: behind · ahead_by 0 · behind_by 370⇒ the pinned commit is an ancestor of that fix: the pin predates it.
So the consequence is concrete and dated:
record:details: the #13855sections[].groupreference form crashes the renderer, and the enumeratedfieldsform renders raw field keys instead of declared labels #16695 could not be released this sweep. Its upstream half is merged and its issue is closed, but 「跨仓解锁判据是消费方可安装,⛔ 不是上游已合并」 — and by the pin reading it is not installable here. The card stayspm:blocked; the refusal and its measurement are recorded atrecord:details: the #13855sections[].groupreference form crashes the renderer, and the enumeratedfieldsform renders raw field keys instead of declared labels #166955671601791.- ⇒ the pin drift is now holding a graded
priority:p2Bug out of a dispatch queue, which is a different thing from 「routine drift, filed so it has an owner」.
⛔ This does not assert that #16695 is the only such card — the scan that found it only covered
domain:spec. Other lanes' cards blocked on merged objectui work would be in the same position, and this seat did not look.What this note is and is not
- ⛔ Not a request to re-prioritise; the four-axis call is this card's own lane's.
- ⛔ Not a pin bump by this seat — 「⛔ pin bump 不做 rider:走专用 bump 脚本连带 override 与 lockfile」.
- ✅ Just the reading, dated, with the downstream cost named, so whoever takes this card can price it against a real blocked card instead of a commit count.
⚠️ And one falsifiable premise for whoever acts: this note reads the pin atorigin/main1bdbf82cb5. Re-read it before acting — if it has moved since, re-run the compare above rather than trusting this number.
Generated by Claude Code
Claim: director seat under the maintainer's direct takeover instruction ·
session_012GcsUbuqFGBibkEDMRC1eE· 2026-09-20T15:14Z
Branch:claude/issue-17429-objectui-pin-bump(new, fromorigin/main)
Seat: director (maintainer-directed; ⛔ not a lane claim)
Clause-②: no (a pin bump plus regenerated artefacts and re-derived anchors; declared from the diff at delivery)
File surface:.objectui-sha,scripts/sdui-manifest.record.json,sdui.manifest.json, lockfile / overrides the bump script writes, the.objectui-shacitations underpackages/spec/src/**(re-derived percheck:objectui-pin-citations), and nothing underpackages/*/srcbeyond those anchorsWhy this card, now. The maintainer, in chat today, verbatim: 「1天之前的pr都帮我诊断修复继续处理完,并跟进到合并」 (objectstack only) and, on PR #18723: 「18723 os-try-charles 已下班,你也接手」. PR #18723 (#17645) is blocked on this card by the devx seat's own reading (5718211831): its port is sound, but
check:sdui-lockstepandcheck:objectui-pin-citationscannot both be green until the pin moves with the 26packages/spec/src/**citations re-derived — and the standing rule on this card says the pin bump is never a rider on another PR. ⇒ This bump lands first, as its own PR, with the manifest artefacts, lockfile/override state, the re-derived citations and post-landing smoke evidence (triage's condition, 5690507317); #17645 then re-merges and lands.Target revision: objectui
origin/mainat the time the dev reads it (≥cfcc17d9dd04, the revision #18723 ported from, so the two codes the lockstep record needs are in the pinned tree). Anos-devround does the work; this seat writes no code.
Generated by Claude Code
os-dev-report
{
"issue": 17429,
"status": "needs_decision",
"branch": "claude/issue-17429-objectui-pin-bump",
"pr": "#19398",
"session": "session_012GcsUbuqFGBibkEDMRC1eE",
"premise_still_valid": true,
"summary": ".objectui-shamoves 53ded82bf7a4 -> 87af769e9a3e (objectui origin/main tip; cfcc17d9dd04 is an ancestor, merge-base --is-ancestor exit 0), produced by scripts/bump-objectui.sh --no-commit, never by hand. It travels with the script's own 584-changeset console digest (@objectstack/console: minor) carrying a real ADR-0087 disposition, the regenerated sdui manifest record (the manifest itself is byte-identical - the pinned commit declares the same published @object-ui 17.6.0), the re-taken sdui-parser lockstep record, and all 27 .objectui-sha citations under packages/spec/src re-MEASURED at the new pin. scripts/build-console.sh is GREEN at the new pin (exit 0, 60228 KB dist, canary and spec-injection assertions both pass). ONE gate is red and it is not this diff's to fix: check:sdui-lockstep reports [code-drift] naming exactlyinert-quick-addandmember-type-mismatch- the two codes PR #18723 (card #17645) already ports. Measured: objectui carries 26 codes at the new pin, this repo 24, the 24 a strict subset, grammar region byte-identical, so pinning the current tip does NOT enlarge the port #18723 already carries; and reverting the record only swaps the red for [pin-moved]. The ordering of the two deliveries is the open question.",
"tests": "SMOKE (triage's condition):bash scripts/pm/os-verify-lock.sh -c 'bash scripts/build-console.sh'-> "os-verify-lock: VERDICT command-exit 0 - held the lock 653s (10m53s)"; "8786 modules transformed"; "Bundle canary 'import/jobs' present"; "Console bundle carries THIS tree's @objectstack/spec, and only it"; "@objectstack/console dist ready (60228 KB) from objectui@87af769e9a3e". dist = 51,702,779 bytes / 3,285 files, content hash sha256 259d2dfecaa12507fd3166a3f91c66c52581990e5ccac3899757027fa6ac108d, dist/.objectui-sha stamped 87af769e9a3e;pnpm check:console-shaexit 0. || GATES (exit code captured before any pipe): check:sdui-lockstep EXIT 1 ([code-drift] only objectui inert-quick-add, member-type-mismatch); check:objectui-pin-citations EXIT 0 ("27 asserting objectui pin citation(s) match .objectui-sha (87af769e9), 48 historical ... 7 anchor content assertion(s) verified"); --verify-anchors EXIT 0; check-sdui-manifest EXIT 0; check:objectui-bump EXIT 0; check:objectui-changeset EXIT 0; check:console-sha EXIT 0; spec check:generated EXIT 0 (15/15 artefacts current after gen:docs + gen:migration-registry); check:react-declaration-parity EXIT 0 with MANIFEST (reading unmoved - manifest byte-identical, sha256 49211fee7792); check-adr-0087-registration --base origin/main EXIT 0; check-changeset-no-major EXIT 0; check-empty-changeset EXIT 0;pnpm lintrepo-wide EXIT 0;pnpm --filter @objectstack/spec testEXIT 0 (503 files / 14710 tests, VERDICT command-exit 0);pnpm --filter @objectstack/spec --filter @objectstack/sdui-parser typecheckEXIT 0; plus 27 further derived families all EXIT 0 (nul-bytes, merge-driver, watch-hint-literal, comment-mask-adoption/-corpus, doc-frontmatter, docs-section-name, scripts-symbol-anchors, spec-docblock-symbol-anchors, ci-filter-parity, keyed-text-bounds, undeclared-dep-imports, section-landing-index, liveness, duration-unit-keys, llms-txt, spec-changes, upgrade-guide, strictness-ledger, export-origins, api-surface, authorable-surface, docs, doc-anchors, docs-spec-enumerations, quick-reference-counts, corpus-claim-drift, changeset-gate-self-tests, pm-widening-tells, published-files). || CITATION METHOD, not an ablation: each citation was re-derived by reading objectui at BOTH shas and comparing CONTENT (git show REV colon PATH), never by arithmetic on a line number. The corpus-absence records' counting method was calibrated first by reproducing the OLD pin's own published numbers exactly (6409 tracked files, useState 2304, objectstack 10171, @objectstack/spec 3479) before any new number was written. || NOT MEASURED: the 7 path-scheduled CI jobs and the ~95 remaining derived families (dispatch-gates derives 122 for this 19-path / 757-line change set); left to CI, which runs the whole farm. origin/main moved 7 commits during the round and touches NONE of this diff's 19 paths, so no merge was taken.",
"mcp_calls": "0 - no MCP GitHub tool was called, read or write",
"api_writes": "2 REST writes via the proxy: POST /repos/objectstack-ai/objectstack/pulls (draft PR #19398), POST /repos//issues/17429/comments (this report). No label write, no assignee write, no issue created - per the dispatch. Non-REST: 6git pushto claude/issue-17429-objectui-pin-bump.",
"open_questions": [
{
"question": "check:sdui-lockstep cannot be green on THIS PR at any pin this card is allowed to take. Measured: objectui carries 26 sdui-parser diagnostic codes at 87af769e9a3e, this repo 24, the 24 a strict subset, grammar region byte-identical - the whole divergence isinert-quick-add+member-type-mismatch, which is exactly the port PR #18723 (card #17645) already carries and which the devx seat already reviewed as sound (26 = 26 there). Pinning the current tip rather than cfcc17d9dd04 does not enlarge it. Reverting the lockstep record only swaps [code-drift] for [pin-moved].Lint & Repo Gatesis a required context, so neither PR can land alone. How should the two deliveries be ordered?",
"options": [
"A - Merge the two into ONE landing: cherry-pick PR #18723's two-code sdui-parser port (plus its tests and ablation) onto this branch, close #18723 as superseded. Cost: this PR absorbs another seat's reviewed in-flight work, and its Clause-2 becomes that PR'syesrather than this card'sno; benefit: one green PR, one landing, no ordering problem left.",
"B - Land #18723 FIRST, rebased onto this branch's content: i.e. move this PR's whole diff into #18723 instead. Cost: #17429's standing rule says a pin bump is never a rider on another PR, and this PR's 27 re-measured citations plus smoke evidence would become a rider on a parser card; benefit: #18723 keeps its reviewed port and its contract review.",
"C - Keep both PRs, land this one with check:sdui-lockstep red by having the maintainer merge it by hand (the required-context override). Cost: a red required gate lands on main and rides main's merge ref into every later PR until #18723 lands; that is exactly the class AGENTS.md calls 'an advisory red that lands rides main's merge ref into every later PR until stanched', here with a REQUIRED context. Not recommended.",
"D - Keep both PRs and add ONLY the two codes to this PR (not #18723's tests), letting #18723 keep its test and ablation halves. Cost: the two PRs then both touch packages/sdui-parser/src with a textual conflict guaranteed, and a code without its test is the half AGENTS.md refuses."
],
"recommendation": "A, because it is the only option where one PR is green on its own and nothing lands red. The dispatching seat owns the call: it means closing #18723 as superseded after copying its port verbatim (its contract-review record would have to be re-taken on this PR's head). ⛔ The four-axis escalation framework was NOT carried in this round's dispatch, so this recommendation is argued on landability and on the gates' own readings rather than on those axes - hand me the framework and I will re-argue it on them rather than invent a set."
}
],
"out_of_scope_findings": [
"noted, not filed:.changeset/18516-span-auto-wide-types.md(pending, unreleased) states "At this pin only the widest tier's class is emitted" about the FormField.span describe this PR corrects, so it ships a sentence about text that no longer exists. NOT filed and NOT fixed here because check:empty-changeset refuses any diff that modifies a changeset present at the merge base ("No changeset from the merge base modified or deleted by this diff (#17712)") - so the correction is structurally a separate landing. Carrier: the release that consumes it (the curated release-notes compilation reads these bodies), or a docs-only PR after this one lands.",
"noted, not filed: the logging semantic migration entry's lit controltimeout 702was WRONG at the pin it named - that corpus reads 832 (re-measured at 53ded82bf7a4 with the same git grep -F -o method that reproduces every other control on that record exactly). The #10274 class again: an error invisible to any sha refresh. Corrected in place in this PR (to the new pin's 1075) rather than filed, since re-measuring it is this card's own work. Carrier: none needed - it is fixed in this diff.",
"noted, not filed: theelement:buttonicon .describe() says the renderer resolves through lucide-react'siconsmap; at the new pin resolveIcon reaches the glyph one hop indirectly (recordIconName + lazyIconComponent). The accepted spellings and the no-fallback-glyph behaviour are unchanged, so the sentence is imprecise rather than false and was left rather than churned. Carrier: the next author of that record - the docblock above it now states the new mechanism."
]
}
Generated by Claude Code
Pointer from the director seat (
session_012GcsUbuqFGBibkEDMRC1eE), 2026-09-20T22:53Z — ruling batch #201 item 1, maintainer 「201 A」 (record on #17645): PR #19398 also carries #17645's two-code sdui-parser port (cherry-picked verbatim with its tests and changeset; lockstep regenerated at the new pin), so this card's delivering PR closes both cards. Clause-② correction for this card:yes(the save gate's diagnostic set grows with the port; the pin bump alone wasno, 5750664275) —Clause-②-correction: 5750664275 Clause-②: yes. Anos-devround is on the branch now; the at-tier contract review follows on the final head.
Generated by Claude Code
os-dev-report
{
"issue": 17429,
"status": "done",
"branch": "claude/issue-17429-objectui-pin-bump",
"pr": "#19398",
"session": "session_012GcsUbuqFGBibkEDMRC1eE",
"premise_still_valid": true,
"head": "6ef8344645a18028623d004bd7d106e6c81d56f7",
"summary": "Folded PR #18723's two-code sdui-parser port ontoclaude/issue-17429-objectui-pin-bumpas one commit (6ef8344645), eight files each byte-identical to the port branch's blob;.objectui-sha,scripts/sdui-manifest.record.jsonandpackages/sdui-parser/objectui-lockstep.jsonwere deliberately not brought.check:sdui-lockstepis now GREEN at the live pin: 26 = 26 diagnostic codes, grammar region byte-identical over 214 lines (blob 0131f27cf86d) against objectui@87af769e9a3e. The lockstep record needed NO re-recording (this branch already carries it taken at the live pin,recordedAgainstPin==.objectui-sha== 87af769e9a3e);pnpm gen:sdui-lockstepwas attempted, REFUSED with[head-off-pin](the sibling objectui checkout sits at c2f0f48329de) and wrote nothing (record blob e0a6988b1c9a before and after) — and its printed remedy was not taken, because parking that read-only shared checkout on the pin is a git write in someone else's tree. The type-check red #18723 carried wascheck:objectui-pin-citationsinside theType Check · source gatesjob, and the fold closes it without any edit underpackages/sdui-parser/**: this branch's 27 citations are already re-measured at 87af769e9a3e, so the gate exits 0; reverse-verified by rewriting.objectui-shaback to cfcc17d9dd04 throughscripts/ablation-replace.mjs, which turns it RED with five anchor-content mismatches, then restoring (blob back to 035b6937d641,git diff HEADempty). No@ts-ignoreand noanywere added.⚠️ objectui DID movepackages/sdui-parserbetween cfcc17d9 and the live pin — objectui#6771'sbody-dialect.ts(+99) withindex.ts+1 andvalidate.ts+31/-7 — so the ported files were re-verified against 87af769e9a3e rather than the revision #18723 read; that move touches neither ported code and stamps only codes already in the 26-set, which is why 26 = 26 holds with it unported. Its residue is filed below as a to-file finding, not absorbed. \n\n=== PR BODY TEXT FOR THE SEAT TO ADD TO #19398 (append after the existing pin-bump sections; the body already carriesFixes #17429,Fixes #17645andClause-②: yes, and the RED section it still contains aboutcheck:sdui-lockstepis now stale and should be replaced by this) ===\n\n## Fold-in: the two-code sdui-parser port (ruling batch #201 item 1, maintainer 「201 A」)\n\nCommit6ef8344645brings PR #18723's port onto this branch verbatim. Eight files, each byte-identical to the blob onclaude/issue-17645-sdui-lockstep-port(git hash-objecthere vsgit rev-parse claude/issue-17645-sdui-lockstep-port:PATHthere — all eight SAME, no hunk rewritten):\n\n| path | blob |\n|---|---|\n|packages/sdui-parser/src/kanban-quick-add.ts|20f9f74b9f05|\n|packages/sdui-parser/src/validate.ts|156a264073f7|\n|packages/sdui-parser/src/types.ts|567bc1259bea|\n|packages/sdui-parser/src/index.ts|2ea35795b9cf|\n|packages/sdui-parser/src/codegen.ts|76a7d654f1fd|\n|packages/sdui-parser/src/__tests__/inert-quick-add.test.ts|723cad21333a|\n|packages/sdui-parser/src/__tests__/member-type-mismatch.test.ts|4b87cadaa067|\n|.changeset/17645-sdui-parser-lockstep-port.md|7c171fddfc43|\n\nDeliberately NOT brought from that branch:.objectui-sha,scripts/sdui-manifest.record.json(both already this branch's, at87af769e9a3e) andpackages/sdui-parser/objectui-lockstep.json.\n\n### The lockstep record was neither text-merged nor re-recorded — and needed neither\n\nThis branch already carries the record taken at the live pin in the pin-bump round:objectui.rev=recordedAgainstPin=87af769e9a3ee28ace099fdd653d3ebd79fe82e2, 26 diagnostic codes includinginert-quick-addandmember-type-mismatch.--updatere-reads objectui's side only, so a re-record here would have rewritten the same bytes.\n\nAttempted anyway, and it REFUSED, which is the generator working:pnpm gen:sdui-lockstepexit 1,[head-off-pin] the objectui checkout is at c2f0f48329de, and .objectui-sha names 87af769e9a3e. Its record file is byte-unchanged across the attempt (e0a6988b1c9abefore and after). ⛔ The remedy it prints — parking the sibling checkout on the pin — was NOT taken: that is a git write inside a read-only checkout this container shares with other agents, and the record it would produce is the one already committed.\n\nResult —check:sdui-lockstepexit 0 on the final head:\n\n>check:sdui-lockstep: OK — this copy is byte-identical to objectui@87af769e9a3e (2026-09-20T15:03:40+00:00) over 214 grammar line(s) [blob 0131f27cf86d] and agrees on all 26 diagnostic code(s), across 8 non-test source(s).\n\n--self-testexit 0 alongside.check:sdui-manifestexit 0 (manifest intact, sha25649211fee7792, recorded at the live pin).\n\n### The type-check red PR #18723 carried is closed by the pin, not by a code change\n\nThe red wascheck:objectui-pin-citations, which runs inside theType Check · source gatesjob (.github/workflows/lint.yml, thetypecheck-source-gatesjob). Three readings:\n\n1. It reds exactly on the two heads of #18723 that claimed.objectui-sha=cfcc17d9dd04and is green on the head that dropped that claim (8a778067eb); #18723's own body declares the gate RED and deliberately left open.\n2. Reverse verification on THIS branch, throughscripts/ablation-replace.mjs: rewriting.objectui-shatocfcc17d9dd04(anchor 1 → 0, blob035b6937d641→2efe67506023, mutation proven on disk) turns the gate RED — five anchor-content mismatches, e.g.packages/spec/src/ui/view.zod.ts:1857 — the asserted first line does not match packages/plugin-list/src/ListView.tsx:146 at this pin. Restored and proven: blob back to035b6937d641,git diff HEADempty.\n3. With this branch's live pin the gate is exit 0:27 asserting objectui pin citation(s) match .objectui-sha (87af769e9), 48 historical citation(s) recorded and not checked, across 1497 spec source(s); 7 anchor content assertion(s) verified.\n\nSo the fix is the fold direction itself — the port lands on the branch whose 27 citations are already re-measured at the pin. No edit was made insidepackages/sdui-parser/**for it, and no@ts-ignoreoranywas added anywhere. The package's owntsc --noEmitis exit 0, and--listFilesconfirms the two new test files are inside that program (so the green is about them, not around them).check:type-check-coverageexit 0.\n\n### Ported fromcfcc17d9, and re-verified against the LIVE pin\n\nobjectui DID movepackages/sdui-parserbetweencfcc17d9dd04and87af769e9a3e, so the ported files were re-checked against the live pin rather than the revision #18723 read. The move is one unrelated feature — objectui#6771's retiredbodychild-list dialect: a newbody-dialect.ts(+99),index.ts(+1),validate.ts(+31/-7) and its own test. It touches neither ported code, and it stamps onlynot-a-containerandunknown-prop, both already in the 26-code set — which is why26 = 26holds at the live pin with it unported.\n\nAgainst objectui at87af769e9a3e, the whole non-comment divergence of the five ported sources is now four items, three of them pre-existing:\n\n-ValidationResulthere vsManifestValidationResultthere (this copy's own type name, unchanged by the port);\n-code: 'inert-quick-add'as an inline literal instead of theINERT_QUICK_ADDconstant — the divergence documented at the site, forced bycheck:dispatcher-error-vocabulary's literal grammar, pinned equal to the constant by a test row, and lockstep-neutral because the gate resolves an identifier at acode:position through the package's own constants (check:dispatcher-error-vocabularyexit 0 here);\n-provenance.tsandassertFullyLoaded/RegistryConfigLike.lazyabsent here (both stamp zero codes; already in #18723's acceptance notes);\n-body-dialect.tsabsent here — new sincecfcc17d9. See Acceptance notes.\n\nThe port strictly REDUCES divergence in every file it touches (non-comment plus comment diff lines vs objectui at the live pin):validate.ts146 → 73,index.ts114 → 92,codegen.ts30 → 3,types.ts96 → 85.\n\n### Ablations — re-run here, on this tree, not transcribed\n\nThree, each throughscripts/ablation-replace.mjs(anchor declared and required to hit exactly once, mutation proven on disk by anchor count and blob change, restore proven byblob == HEADAND an emptygit diff HEAD, EXIT/INT/TERM trap). The two suites import../index.js— a relative intra-package specifier, nodisthop — so no rebuild is in the resolution path.\n\n| # | mutation | file | blob | result |\n|---|---|---|---|---|\n| 1 |checkKanbanQuickAddstamps nothing |kanban-quick-add.ts|20f9f74b9f05→ca38271c7b0a| 5 failed / 8 passed (13) |\n| 2 | the validator's member check reads no declared arms |validate.ts|156a264073f7→a130ab30d201| 7 failed / 14 passed (21) |\n| 3 | the codegen narrows no member type |codegen.ts|76a7d654f1fd→4e779755e635| 3 failed / 18 passed (21) |\n\nIn each leg the failures are the SUBJECT rows and every control row stays green. Baselines before ablation: 13 passed and 21 passed. Ablation 1 reads 5 rather than #18723's 4 because the literal-vs-constant pin row landed after that reading and also fails when the diagnostic is disarmed.\n\n### Gates on the final head6ef8344645\n\nscripts/pm/dispatch-gates.mjs --ranreconciles at exit 0: 122 derived, 115 run, 7 NOT-MEASURED, 0 UNRUN, every accounted family carrying a captured exit code. The 7 NOT-MEASURED all exited 3 = PREREQUISITE NOT MET (they need a full workspace build):check:dual-build-cjs-loads,check:lean-entry-closure,check:type-check-debt,check:sourcemap-no-sources-content,check:docs-transcript-drift, and the@objectstack/lintpaircheck:doc-formula-expressions/check:doc-security-posture.\n\nOne more is NOT MEASURED in substance though it exits 1:check:skill-examplesrefuses onpackages/client-react/dist holds no .d.ts declarations — the package is not built, and building that package here fails on its own unbuilt@objectstack/clientdependency. Nothing in this diff reaches it.\n\nNamed readings:pnpm --filter @objectstack/sdui-parser test176 passed / 10 files ·pnpm --filter @objectstack/sdui-parser typecheckexit 0 ·pnpm --filter @objectstack/spec test14708 passed, 1 skipped ·pnpm --filter @objectstack/spec check:generatedexit 0 (all 15 artefacts up to date, after a realpnpm --filter @objectstack/spec build) ·check:react-declaration-parityexit 0 in the CI spelling (MANIFEST=$PWD/sdui.manifest.json … --baseline react-declaration-parity.baseline.json --strict; without--baselineit exits 1 by design and that reading is not a finding) ·check-empty-changeset,check-changeset-no-major,check-adr-0087-registration,check:changeset-gate-self-testsall exit 0 ·pnpm lintrepo-wide exit 0.\n\n### Changesets\n\nTwo, both already present:.changeset/console-87af769e9a3e.md(@objectstack/console: minor, the bump script's digest) and.changeset/17645-sdui-parser-lockstep-port.md(@objectstack/sdui-parser: minor, carried verbatim from #18723).check-empty-changesetexit 0 confirms no changeset from the merge base was modified or deleted.\n",
"tests": "LOCKSTEP (exit captured before any pipe):node scripts/check-sdui-lockstep.mjsEXIT 0 — "byte-identical to objectui@87af769e9a3e (2026-09-20T15:03:40+00:00) over 214 grammar line(s) [blob 0131f27cf86d] and agrees on all 26 diagnostic code(s), across 8 non-test source(s)";--self-testEXIT 0.pnpm gen:sdui-lockstepEXIT 1[head-off-pin], record blob e0a6988b1c9a unchanged.node scripts/check-sdui-manifest.mjsEXIT 0 +--self-testEXIT 0.pnpm --filter @objectstack/spec check:objectui-pin-citationsEXIT 0 (27 asserting citations match, 48 historical, 1497 spec sources, 7 anchor content assertions verified). || TESTS:pnpm --filter @objectstack/sdui-parser testvia os-verify-lock — VERDICT command-exit 0, 176 passed / 10 files.pnpm --filter @objectstack/sdui-parser typecheckEXIT 0, andtsc --listFilesshows BOTH new test files inside that program (so the green covers them).pnpm --filter @objectstack/spec testvia os-verify-lock — VERDICT command-exit 0, 14708 passed | 1 skipped.pnpm --filter @objectstack/spec build && check:generatedvia os-verify-lock — VERDICT command-exit 0, "All 15 generated artifacts are up to date".pnpm lintrepo-wide EXIT 0. || ABLATION (3 legs, eachscripts/ablation-replace.mjs: anchor declared x1 and required to hit, on-disk mutation proven by anchor 1->0 AND a changed git blob, restore proven by blob == HEAD AND emptygit diff HEAD, EXIT/INT/TERM trap; no rebuild in the resolution path — both suites import../index.js, a relative intra-package specifier, no dist hop): (1) kanban-quick-add.ts 20f9f74b9f05 -> ca38271c7b0a => 5 failed / 8 passed (13, baseline 13 passed); (2) validate.ts 156a264073f7 -> a130ab30d201 => 7 failed / 14 passed (21, baseline 21 passed); (3) codegen.ts 76a7d654f1fd -> 4e779755e635 => 3 failed / 18 passed (21). Failures were the SUBJECT rows only; every control row stayed green; tree clean after each leg. || REVERSE VERIFICATION of the type-check diagnosis:.objectui-sha035b6937d641 -> 2efe67506023 (87af769e9a3e -> cfcc17d9dd04) makescheck:objectui-pin-citationsEXIT 1 with 5 anchor-content mismatches; restored, blob back to 035b6937d641,git diff HEADempty. || GATE FAMILIES:node scripts/pm/dispatch-gates.mjs --ran RANFILE --repo objectstack-ai/objectstackEXIT 0 — 122 derived, 115 run, 7 NOT-MEASURED, 0 UNRUN, every accounted family carrying a captured exit code. NOT MEASURED (all exit 3 = PREREQUISITE NOT MET, all needing a full workspace build): check:dual-build-cjs-loads, check:lean-entry-closure, check:type-check-debt, check:sourcemap-no-sources-content, check:docs-transcript-drift, @objectstack/lint check:doc-formula-expressions and check:doc-security-posture. An eighth is NOT MEASURED in substance while exiting 1: check:skill-examples refuses on "packages/client-react/dist holds no .d.ts declarations", and building that package here fails on its own unbuilt @objectstack/client dependency — nothing in this diff reaches it. check:react-declaration-parity EXIT 0 in the CI spelling (MANIFEST + --baseline + --strict); without --baseline it exits 1 by design and that earlier reading of mine was an invocation error, not a finding. || NOT MEASURED, left to CI: the 7 path-scheduled CI jobs, the 45 artifact-roster families, the 11 declared WIDE-population families and the always-runs tail — each outside the derived 122. origin/main moved during the round and was NOT merged, per the dispatch.",
"mcp_calls": "0 — no MCP GitHub tool was called, read or write",
"api_writes": "1 REST write via the proxy: POST /repos//issues/17429/comments (this report). No PR body PATCH, no ready flip, no auto-merge, no label write, no assignee write, no issue created, PR #18723 untouched — per the dispatch. Non-REST: 2git pushto claude/issue-17429-objectui-pin-bump (one no-op write-routing probe, one carrying commit 6ef8344). Reads used REST only.",
"open_questions": [],
"out_of_scope_findings": [
"to file (class b — violates the declared lockstep contract; dedupe words: body-dialect, checkRetiredBodyDialect, RETIRED_CHILD_LIST_KEY, sdui-parser port, objectui#6771): this repo's packages/sdui-parser has no counterpart to objectui's body-dialect.ts, which landed in objectui between cfcc17d9dd04 and the live pin 87af769e9a3e (git -C ../objectui diff cfcc17d9..87af769e -- packages/sdui-parser: body-dialect.ts +99, index.ts +1, validate.ts +31/-7, plus its own test). #12719's contract, quoted in scripts/check-sdui-lockstep.mjs's header, is "both copies byte-agree on the accepted grammar and on diagnostic codes" — and for an authoredbodykey the two copies now stamp DIFFERENT codes: objectui's renderer answers not-a-container under a non-container and a named unknown-prop otherwise, while this copy's save gate answers the bare unknown-prop in both cases. check:sdui-lockstep is structurally blind to it: it compares code SETS, and body-dialect.ts stamps only not-a-container and unknown-prop, both already in the 26-set, so 26 = 26 holds. NOT fixed in place: it is a third code path the ruling does not name and it would add a new file, new tests and a new verification surface. Both severities arewarning, so compile().ok and the save verdict are unchanged — this is a message/code divergence, not a save-vs-render accept split.",
"noted, not filed: objectui's provenance.ts (HTML_TIER_NODE, isHtmlTierNode, markHtmlTierNode) and its assertFullyLoaded plus RegistryConfigLike.lazy are still absent from this copy. Both stamp zero diagnostic codes, so the lockstep code-set comparison is unaffected and the gate does not compare file lists. #18723's acceptance notes already record provenance.ts. Carrier: the next sdui-parser port round, which will have both trees in hand.",
"noted, not filed: check:skill-examples refuses its own prerequisite with exit 1, while seven other families in this same derivation spell that refusal exit 3 (PREREQUISITE NOT MET) — so a run record built from exit codes, which is exactly what dispatch-gates --ran classifies, reads that refusal as a finding. The gate's text is loud and correct; only the code is the odd one. Carrier: the next author of packages/spec/scripts/check-skill-examples.ts.",
"noted, not filed:pnpm gen:sdui-lockstepis unrunnable in this container because the shared sibling checkout /home/user/objectui sits at c2f0f48329de rather than the pin, and the remedy the generator prints is agit checkoutinside a read-only checkout other agents share. It cost nothing this round (the record was already taken at the live pin) but it will block the next pin bump. Carrier: the next pin bump, which must park its own checkout — scripts/bump-objectui.sh already requires one."
]
}
Generated by Claude Code
os-dev-report
{
"issue": 17429,
"status": "done",
"addendum_to": "os-dev-report 5753538552",
"branch": "claude/issue-17429-objectui-pin-bump",
"pr": "#19398",
"session": "session_012GcsUbuqFGBibkEDMRC1eE",
"premise_still_valid": true,
"head": "fc25a5925d7d6b55ed25b81d5f1fcef4be836109",
"previous_head": "6ef8344645a18028623d004bd7d106e6c81d56f7",
"summary": "Fix round for theLint & Repo Gatesred atpnpm check:issue-citations && node scripts/check-issue-citations.mjs. One commit,fc25a5925d, 2 files / +10 -2, both underpackages/spec/src/ui/. ⛔ The gate script was not touched. (1)view.zod.ts:2954—objectstack#17328respelled bare as#17328. The issue exists; the red was the gate's own blind spot, confirmed in its source:classifyCitationtreats the bare-repo qualifier as naming THIS repo and falls through to the board lookup, while the board is only asked for UNQUALIFIED citations (the probe guard reads!cite.qualifier), so a qualified-this-repo number is looked up on a board that was never asked for it. Bare is the grammar's spelling for this repository, and it now resolves — the run reports2 numbers resolvewhere the red run reported 1. (2)component.zod.ts:2135—--probe-causeclassified the number as DELETED, not transferred: "minted, gone from the board, and the web endpoint 404s too". Per the gate's rule no replacement number is guessed. The rewritten sentence now states in prose that the number does not resolve, how and when that was probed, that it is deleted rather than transferred, and what the live record is — THIS BLOCK pluscheck:objectui-pin-citations, whose header carries the same re-READ rule. The number itself is written in prose without the#sigil, because a#-sigil token on a line this diff adds is exactly what the gate judges and there is no target for it to resolve to; the block's later paragraph keeps the sigil form unchanged from main, per the dispatch, and the new prose is the tombstone that covers it.check:generatedconfirms the docblock edits moved no generated page: TSDoc above the key, not a.describe(), so theui/component,ui/viewandui/datasetreference pages are untouched — all 15 artefacts up to date with no regeneration.",
"tests": "All exits captured before any pipe, on the committed head fc25a59.node scripts/check-issue-citations.mjsEXIT 0 — "every citation this change adds resolves (or is a declared cross-repo reference)"; 31 judged across 15 files, 30 cross-repo-unjudged, 1 resolves, 0 findings (before the fix: EXIT 2, 32 judged, 2 allocated-but-absent).pnpm check:issue-citations(the self-test half CI runs first) EXIT 0 — 66 cases, 7 batteries.node scripts/check-issue-citations.mjs --probe-cause --use-env-proxyEXIT 2 on the OLD tree, and it is the measurement this round turns on: component.zod.ts:2135 reclassifiedallocated-but-absent->deleted, while view.zod.ts:2954 stayedallocated-but-absentbecause the probe never fires for a qualified citation.pnpm --filter @objectstack/spec check:objectui-pin-citationsEXIT 0 — 27 asserting citations match .objectui-sha (87af769e9), 48 historical, 1497 spec sources, 7 anchor content assertions verified.pnpm --filter @objectstack/spec build && typecheck && check:generatedvia os-verify-lock — VERDICT command-exit 0; typecheck is the three-part script (tsc --noEmit, check:scripts-typecheck, check:test-typecheck) and all three passed; check:generated reports "All 15 generated artifacts are up to date" with nothing regenerated, so the reference pages did not move. eslint on the two touched files, in the root script's own spelling (node --stack-size=4000 node_modules/eslint/bin/eslint.js --no-inline-config PATHS) EXIT 0. Control-byte sweep over both files: no match (grep exit 1). On-disk edit proof for both files: anchor count 1 -> 0 and a changedgit hash-objectblob (view.zod.ts c3c28356d705 -> 32e4b6b5a8dc; component.zod.ts a2eb32b2b727 -> 31fb9f595139). NOT MEASURED this round: everything outside the five commands the dispatch named — the rest of the derived family list was measured on 6ef8344 in the previous report and this diff touches two docblocks in two files already covered by it.",
"mcp_calls": "0 — no MCP GitHub tool was called, read or write",
"api_writes": "1 REST write via the proxy: POST /repos//issues/17429/comments (this addendum). No PR body PATCH, no ready flip, no auto-merge, no label write, no assignee write, no issue created. Non-REST: 1git pushto claude/issue-17429-objectui-pin-bump carrying commit fc25a59.",
"open_questions": [],
"acceptance_notes": [
"check-issue-citationsblind spot, recorded as an Acceptance-notes entry and NOT filed as a card, per the dispatch: a citation qualified with this repository's own name (objectstack#N, orobjectstack-ai/objectstack#N) is recognised byclassifyCitationas naming THIS repo, so it escapescross-repo-unjudgedand is resolved against the board — but the board is only asked for citations with NO qualifier, and--probe-cause's own guard is!cite.qualifiertoo. The result is a live issue reported asallocated-but-absentand, worse, reported as NOT MEASURED even under--probe-cause, so the run that exists to separate deleted from transferred cannot speak about it at all. Measured here on #17328 (closed completed 2026-09-11 by merged PR #17671): red while qualified,resolvesthe moment it is spelled bare, same tree, same run. ⛔ Fixed in the PR text only — the gate script was not touched."
],
"out_of_scope_findings": [
"unchanged from the previous report (5753538552): the to-file class-(b) body-dialect finding and three noted-not-filed entries stand as written; this round adds none."
]
}
Generated by Claude Code
Landing record — director seat, summon #25 (
session_012GcsUbuqFGBibkEDMRC1eE), 2026-09-21T01:00ZLanded. PR #19398 merged through the merge queue 2026-09-21T00:58Z, squash
fbc12be318de0713e82e1f38ab804b5b63478e64onorigin/main(tip at this read), 27 files, +1340 −219. Verified by content on the fetched ref, ⛔ not the working tree:.objectui-sha=87af769e9a3ee28ace099fdd653d3ebd79fe82e2; the 27packages/spec/src/**citations re-measured at that pin (check:objectui-pin-citationsand--verify-anchorsexit 0 in the at-tier review 5753767727); the two corrected.describe()sentences present inui/dataset.zod.ts(measureformaton datetime honoursrelative/short) andui/view.zod.ts(FormField.spanemits one clamped class per multi-column tier) with their regenerated reference pages;sdui.manifest.jsonbyte-identical (57 components,@object-ui17.6.0 at both shas) with the provenance record moved; changesetsconsole-87af769e9a3e.md(@objectstack/console: minor) and17429-pin-bump-describe-corrections.md(@objectstack/spec: patch) present. Console dist built at the new pin by the dev (VERDICT 0, 60228 KB) and rebuilt by the reviewer (VERDICT 0, stamp87af769e9a3e,check:console-sha0).Also carried by the same PR under ruling batch #201 item 1 (「201 A」): the two-code sdui-parser port from #17645 (
check:sdui-lockstep26 = 26 at the live pin) — its own landing record is on #17645. The card auto-closedcompletedon the merge (Fixes #17429);pm:dispatchedremoved in this stroke. At-tier review PASS 5753767727 on the merged headfc25a5925d; provenance 5753797054 on the PR.Carried, not closed here: the
check-issue-citationsblind spot (a citation qualified with this repository's own name is judged against a board never asked for it — Acceptance notes on the PR, for the next author ofscripts/check-issue-citations.mjs); objectui'sbody-dialect.tsdivergence in this copy ofsdui-parser(a warning-level code divergence invisible to the lockstep code-set comparison — recorded on #17645's landing record as the next port round's item);gen:sdui-lockstepunrunnable in a container whose sibling checkout is off the pin (the next pin bump parks its own checkout).
Generated by Claude Code
- added a commit that references this issue
on Sep 28, 2026
Routine pin drift, filed so it has an owner.
.objectui-sha(repo root) pins whichobjectstack-ai/objectuicommit this repo builds the console from; it is read byci.yml,cut-rc.yml,lint.yml,release.ymlandshowcase-smoke.yml.Measured 2026-09-10
53ded82bforigin/main:efead6c60Why now rather than at the next sweep
⛔ Nothing is broken.
scripts/build-console.shwas measured GREEN against the post-#17372 tree at this pin by the dev of that card, because the thing the chain removed from objectui (export type * as Cloud from '@objectstack/spec/cloud', objectui#8360) was type-only and erased at build time.The one chain-specific reason to prefer sooner: objectui#8360 is among those 584 commits, so until the pin moves, this repo's console build still compiles an objectui tree that re-exports a spec subpath which no longer exists. Type-only today; it stops being harmless the moment anything in that range reaches the subpath at runtime.
cut-rc.yml'spnpm objectui:build. ⛔ NOT theConsole Pin Gate— its path filter watches.objectui-shaand five console scripts, so it runs on neither an unrelated PR normain. A green PR is not evidence here.Doing it
Use the dedicated bump script rather than editing the file by hand — the pin travels with
pnpm sdui:manifestoutput and lockfile/override state, which is why ⛔ pin bumps are never riders on another PR. Expect the Playwrightsdui:manifeststep and 584 commits' worth of unrelated objectui change to come with it; that is the cost of the cadence, not of this card.Refs: #16325 (the chain), #17372 (step 3), objectui#8360 (step 2, the commit that motivates it).