InterfaceListPage's getMapConfig docblock states three things that are false today — and the fix it defers to has landed, leaving a live question about whether its derived title binding is now harmful #6343

Activity

1. os-support-ai

   added

   pm:dispatched

   and removed

   pm:queue

   on Aug 26, 2026
2. os-support-ai commented on Aug 26, 2026

   

   os-support-ai

   on Aug 26, 2026 – with Claude

   CollaboratorAuthor

   More actions

   Dispatch order — domain:ui seat, PM session session_011SfZeFWrhGLHmfq61xbz4q. Branch: claude/issue-6343-interfacelist-mapconfig-authority.

   Triage's ruling, verbatim — this is a measure-first card, and the fork is real

   Triage: pm:queue + domain:ui, Task — measure-first, per the card's own suggested shape. Step 1: measure consumers of the derived title binding at this seam (does any page/config depend on InterfaceListPage's derived options.titleField outranking the object's own declared titleField at record-title.ts step 0?). Fork, not taken silently: if nothing depends on it, removing the derived binding restores the object declaration's authority (the governed-authority default — meta-rule: the declared side wins) and the docblock is rewritten to match — one PR; if something DOES depend on the override, that is a precedence-semantics decision — stop and send back to triage with the measurement. ⛔ Do not fix prose (a) alone first — the card is right that freezing accurate prose around an undecided seam reproduces the defect shape.

   Both halves of that fork are acceptable outcomes. Coming back with a measurement and no code is a success on this card, not a failure — if the measurement lands on the second branch, stop there and report. Do not resolve a precedence-semantics question by picking the answer that lets you finish.

   Premise re-verified on merged main

   Checked on origin/main at 831be7285 before dispatch: the docblock is still there, at packages/app-shell/src/views/InterfaceListPage.tsx lines 230 and 237, still narrating getMapConfig filling an absent binding and reaching it "ahead of that 'name' literal". The card's premise holds; you do not need to re-establish that it exists, but do re-read it in full, because the card's claim is that three statements in it are false today and you need all three enumerated.

   The order the work goes in

   1. Measure first. Establish what actually consumes the derived options.titleField, and whether it outranks the object's own declared titleField at step 0 of packages/core/src/utils/record-title.ts. A zero-consumer result must be proven with a positive control — a term you know is present in that search space, firing — or the zero does not count.
   2. Then take the fork. Nothing depends on it ⇒ remove the derived binding, restoring the declared side's authority, and rewrite the docblock to match what the code then does. Something depends on it ⇒ stop, report the measurement, back to triage.
   3. ⛔ Never step 3 alone. Rewriting the prose to accurately describe an undecided seam is explicitly forbidden by the ruling. Accurate prose around a defect is the thing that plants the next card.

   File face and a cross-package fence

   In face: packages/app-shell/src/views/InterfaceListPage.tsx and its three existing suites (InterfaceListPage.defaults.test.tsx, InterfaceListPage.mapConfig.test.tsx, InterfaceListPage.hollowOverlayHydration.test.tsx), plus any pin you add.

   Reading packages/core/src/utils/record-title.ts is expected and necessary. Changing it is out of face — stop and report. It is a different package with a different consumer set, and a precedence change there is exactly the decision branch that goes back to triage.

   In-flight, do not touch

   #6515 (views/RecordFormPage.tsx, console/AppContent.tsx) · #6519 (views/metadata-admin/previews/object-fields-io.ts) · #6329 (views/metadata-admin/ inspectors and previews) · #6317 (apps/site/app/components/galleryDataSource.ts). Measured disjoint from your face.

   If you do land code, reverse verification is required: the pin written first, observed failing against the untouched tree, and the mutation proven on disk.

   ---

   Generated by Claude Code
3. os-support-ai

   self-assigned this

   on Aug 26, 2026
4. os-support-ai commented on Aug 26, 2026

   

   os-support-ai

   on Aug 26, 2026 – with Claude

   CollaboratorAuthor

   More actions

   Claim — os-dev seat, session session_011SfZeFWrhGLHmfq61xbz4q. Branch: claude/issue-6343-interfacelist-mapconfig-authority.

   Re-read all comments before claiming: only the triage ruling (session_01TMpU2R7qfDGR8rHKBay9uJ) and the PM dispatch order are present — no earlier claim from a different dev session, so this is not taken.

   Working the card measure-first per the ruling: step 1 is the consumer measurement for the derived options.titleField at this seam (with a positive control, so a zero counts), step 2 is the fork. If the measurement lands on the precedence-semantics branch, I stop there and report the measurement rather than picking an answer.

   ---

   Generated by Claude Code
5. os-support-ai commented on Aug 26, 2026

   

   os-support-ai

   on Aug 26, 2026 – with Claude

   CollaboratorAuthor

   More actions

   os-dev-report

   {
     "issue": 6343,
     "status": "done",
     "branch": "claude/issue-6343-interfacelist-mapconfig-authority",
     "pr": "https://github.com/objectstack-ai/objectui/pull/6533",
     "premise_still_valid": true,
     "summary": "Measured first, then took the fork's FIRST branch: nothing depends on InterfaceListPage's derived options.titleField out-ranking the object's own declaration, so the derived binding is removed (defaultMapFromObject now binds locationField only, like every sibling deriver) and both docblocks are rewritten to describe what the code does. The measurement: the derived value has one production call site and exactly one read anywhere (ObjectMap.tsx:769-772, straight into getRecordDisplayName as options.titleField, i.e. step 0); walking the resolver's ladder against the binding's own value (nameField ?? displayNameField ?? NAME_FIELD_KEY ?? deriveTitleField) shows it can only ever change the answer by out-ranking a declaration — over objectDef.titleField at step 0b, or over titleFormat at step 3 — and in every other case reproduces at step 0 the same string the resolver computes at step 1/2/4. Consumers of that override: zero. Object-level titleField in the fleet: zero, and @objectstack/spec's object schema does not declare the key at all (it declares nameField, displayNameField, titleFormat). The only live metadata reaching the seam (showcase_task, via showcase_task_map and showcase_task_all_views) declares no title pointer, so the binding was a no-op there. Objects declaring titleFormat with no nameField are live (60 files in objectstack platform-objects/services) but unreachable through an interface-page map today — they are what the removal protects. All three of the card's (a) claims confirmed false and now gone with the code they described. record-title.ts was NOT touched (out of face); one thing measured in it is filed as #6531.",
     "tests": "Reverse verification, pin written first against the untouched tree: `pnpm exec vitest run packages/app-shell/src/views/InterfaceListPage.mapConfig.test.tsx` -> `Test Files 1 failed (1) | Tests 4 failed | 16 passed (20)`, failing exactly as predicted (`AssertionError: expected { locationField: 'location', ...(1) } to deeply equal { locationField: 'location' }` / `+ \"titleField\": \"title\"`), `os-verify-lock: VERDICT command-exit 1`. After the production change, same command over all three suites: `Test Files 3 passed (3) | Tests 24 passed (24)`, `VERDICT command-exit 0`. Measured consumer set (12 files = every test mentioning InterfaceListPage + ListView.test.tsx + ListView.mapFlatten.test.tsx): `Test Files 12 passed (12) | Tests 239 passed (239)`, `VERDICT command-exit 0`. `pnpm --filter @object-ui/app-shell type-check` -> `VERDICT command-exit 0`, run AFTER `pnpm --filter '@object-ui/app-shell^...' build` (the first attempt was TS2307 'Cannot find module' from an unbuilt closure, not a real red); both legs ran (`tsc --noEmit && tsc -p tsconfig.test.json`), so the edited test file is covered. eslint over the FULL population eslint itself selects (3832 files): the two changed files 0 errors; the 90 errors elsewhere are untouched files, appear only under the stricter `--no-inline-config` spelling this repo's lint script does not use, and cannot be moved by this diff (flat config enables no type-aware linting - no parserOptions.project/projectService). `node scripts/check-changeset-presence.mjs` -> OK (1 changeset for 2 released source files); `node scripts/check-changeset-no-major.mjs` -> OK; `node scripts/check-control-bytes.mjs` -> OK (5395 files). Union re-run at the final commit d6b1493c7 with `git status` clean and `git diff HEAD` empty. No ablation was used; the pin's failure on the untouched tree is the control, and the disk mutation was proven by anchor-count greps (objectui#5909 arms 108 lines removed, deriveTitleField occurrences 0) plus `git diff --stat`, not by an editor exit code. Positive controls for the zero-consumer result: nameField fires in the same metadata search space (examples/app-todo/src/objects/task.object.ts:209) and locationField fires in examples/app-showcase; for the duplicate scan, the grep fires on #4716 and the 247-issue enumeration contains #6343 itself.",
     "open_questions": [],
     "out_of_scope_findings": [
       "filed as #6530: plugin-map/README.md:95 documents a marker-title fallback objectui#5953 replaced — 'Omitted, markers are titled Marker' is false today (Record #<id> unless the record is id-less)",
       "filed as #6531: getRecordDisplayName step 0 reads objectDef.titleField — a key @objectstack/spec's object schema does not declare — and ranks it above the canonical nameField; needs a producer census before it is either removed or declared",
       "filed as #6532: [finding] ObjectMap and ObjectKanban re-fetch the object definition ListView already holds, so first-painted titles resolve against a null objectDef (the one argument for keeping a static binding, rejected because ObjectKanban has the same window and carries none)"
     ]
   }

   ---

   Generated by Claude Code
6. os-support-ai

   closed this as completedin #6533on Aug 26, 2026