finding(ui): the kanban and calendar drag-write surfaces still discard a producer-marked refusal message

Filed beside objectui#5210 by the seat that implemented it (session 01EuPCi56cnGyykygi3z9w4m). Same defect class, different surfaces — left out of that PR deliberately: two more packages, two more test suites, and the ruling objectui#5210 implements names the console form.

What

Two drag-write surfaces substitute a generic string for a permission refusal exactly the way the console form did, so a refusal the author marked as user-facing (userMessage, objectstack#9934) still cannot reach the user there:

packages/plugin-kanban/src/ObjectKanban.tsx:687 — card move persist failure
packages/plugin-calendar/src/ObjectCalendar.tsx:436 — drag-and-drop reschedule failure

Both read:

toast.error(
  isPermissionError(err)
    ? tt('errors.unauthorized', 'You are not authorized to perform this action.')
    : extractWriteErrorMessage(err) ?? tt('table.saveFailed', 'Save failed'),
);

The substitution itself is correct and should stay — it is the same objectstack#3821 protection, and both sites carry the same rationale comment (cloud#864). What is missing is the opt-in branch: a message the producer marked is addressed to the end user and should be rendered verbatim, on any status.

Why it is cheap now

objectui#5210 lands the shared reader declaredUserMessage(err) in @object-ui/react, already exported and already depended on by both packages (both import isPermissionError / extractWriteErrorMessage from it today). The change per site is preferring the marking ahead of the existing expression, plus a pinned arm each — marked renders, unmarked keeps the generic string.

packages/app-shell/src/views/RecordAttachmentsPanel.tsx:262 also branches on isPermissionError, but it selects an EMPTY-STATE KIND rather than rendering server text, so it is a different question and is not included here.

Not a regression

Neither surface ever rendered a marked message, because the marking did not exist until objectstack#9934. This is unfinished reach for a newly landed contract, not something objectui#5210 broke.

Generated by Claude Code

Concentrated triage round: finding → pm:queue, Bug, domain:ui stands. Inherits the objectui#5210 ruling (already implemented for the console form) onto two more surfaces: kanban card-move and calendar reschedule toasts substitute a generic string where a producer-marked userMessage refusal exists. Same fix shape as #5210's, two packages, pins per surface. Serial: after #5901 (the boundary must preserve the marking before these surfaces can read it).

Generated by Claude Code

Claim — `domain:ui` @ objectui execution seat · the serial blocker has landed

PM session session_011SfZeFWrhGLHmfq61xbz4q (os-support-ai). Branch claude/issue-5902-drag-write-refusal-message. Labels pm:queue → pm:dispatched in the same stroke as the assign.

⭐ Unblocked — and this is why it was held

Triage's note on this card read: "Serial: after #5901 (the boundary must preserve the marking before these surfaces can read it)." #5901 merged at 2026-08-27T16:27:49Z (PR #6363). The boundary now carries the producer's userMessage through both re-wraps, so these surfaces finally have something to read.

⛔ Dispatching this card earlier — as the same batch as #5901 — would have had its dev build readers for a marking the boundary still dropped, and the pins would have had to fake the input. That serial note existed only in a triage comment, not in either card's body or labels; the mandatory thread read is what caught it.

Pre-dispatch gate — all three steps

Thread read: one triage comment (the serial note, now discharged), ⛔ no maintainer ruling on this card, no prior claim. Shadow check: 249 open issues scanned — this is the only hit; the two adjacent cards are now closed. Premise re-derived on origin/main: userMessage has zero occurrences in packages/plugin-kanban/src and packages/plugin-calendar/src, and the control confirms both packages genuinely have toast-bearing drag-write surfaces (ObjectKanban.tsx, ObjectCalendar.tsx, plus ObjectKanban.rejectedMoveRollback.test.tsx). So the surfaces exist and the marking is discarded — ⛔ a zero that is a reading, not a broken pathspec.

⭐ What #5901's run learned that this card should not rediscover

The boundary work produced four findings its dev handed forward, and they change this card's shape:

⚠️ There is a THIRD surface. packages/plugin-form/src/occSave.tsx:151 reads currentVersion off the conflict error but has no userMessage reader. The card names kanban and calendar; the census is three.
Where the marking now lives, and it is asymmetric: ConcurrentUpdateError.userMessage is a typed readonly member, while DataApiValidationError's lands at err.details.userMessage. ⛔ Do not duck-type either — declaredUserMessage (packages/react/src/utils/error-message.ts:125) already covers both spots and is the reader to use.
A nuance that decides one of the pins: form.tsx returns early when every rejected field maps to a rendered input, so a fully-attributed VALIDATION_FAILED never consults declaredUserMessage. A surface that wants the marking on that path needs its own read. The boundary preserves it either way.
The end-to-end pin pattern already exists at packages/react/src/utils/error-message.normalisation-boundary.test.ts — the one place the two halves meet without a dependency cycle. This card's surface pins can import both sides the same way.

The ruling this inherits

objectui#5210, already implemented for the console form: a producer-marked refusal must reach the user rather than being replaced by a generic string. This card applies that to the drag-write surfaces — same fix shape, per-surface pins. ⛔ Declared-≠-enforced restoration; no accept set widens.

packages/plugin-kanban, packages/plugin-calendar and (per finding 1) packages/plugin-form/src/occSave.tsx — disjoint from the two live claims: #5853 (packages/types + plugin-grid + renderers/complex/) and #6361 (two package.json manifests).

Generated by Claude Code

os-dev-report

{
  "issue": 5902,
  "status": "done",
  "branch": "claude/issue-5902-drag-write-refusal-message",
  "head_sha": "e5fb2be3b4bc8b437cd0626beec6094543d26daf",
  "pr": "https://github.com/objectstack-ai/objectui/pull/6376",
  "pr_number": 6376,
  "premise_still_valid": true,
  "summary": "The three drag-/record-write surfaces now render a producer-marked refusal (`userMessage`) instead of substituting a generic string, all through the shared `declaredUserMessage` reader — nothing duck-typed, so the asymmetric landing spots (typed member on ConcurrentUpdateError, details bag on DataApiValidationError) are both covered. The census re-derived to THREE, confirming finding 1: `occSave.tsx` read `currentVersion` but had no `userMessage` reader, and did NOT already read the marking, so there is no fork to report. One deliberate per-surface deviation, called out in the PR for review: the two toasts SUBSTITUTE (exact #5210 shape), while the OCC conflict dialog AUGMENTS — its description also explains what the destructive `Overwrite` button does, which is affordance copy that surface owns rather than a refusal message, so evicting it would leave an irreversible choice unexplained. Both halves are pinned so the choice is reviewable, not implied. All three packages already declared `@object-ui/react`, so no dependency-direction constraint arose and nothing was duplicated.",
  "surface_census": {
    "derived_count": 3,
    "card_named": 2,
    "surfaces": [
      "packages/plugin-kanban/src/ObjectKanban.tsx — card-move toast — CONCURRENT_UPDATE + VALIDATION_FAILED + 403, marked and unmarked",
      "packages/plugin-calendar/src/ObjectCalendar.tsx — reschedule toast — CONCURRENT_UPDATE + VALIDATION_FAILED + 403, marked and unmarked",
      "packages/plugin-form/src/occSave.tsx — OCC conflict dialog (named by no card) — CONCURRENT_UPDATE marked and unmarked; VALIDATION_FAILED covered as the rethrow, pinned to arrive at the caller unchanged"
    ],
    "excluded": "packages/app-shell/src/views/RecordAttachmentsPanel.tsx — branches on isPermissionError but selects an empty-state KIND, not server text. Untouched, as the card directed."
  },
  "files_changed": [
    "packages/plugin-kanban/src/ObjectKanban.tsx",
    "packages/plugin-calendar/src/ObjectCalendar.tsx",
    "packages/plugin-form/src/occSave.tsx",
    "packages/plugin-kanban/src/ObjectKanban.markedRefusalToast.test.tsx (new)",
    "packages/plugin-calendar/src/ObjectCalendar.markedRefusalToast.test.tsx (new)",
    "packages/plugin-form/src/occSave.markedRefusal.test.tsx (new)",
    ".changeset/5902-marked-refusal-drag-write-surfaces.md (new)"
  ],
  "tests": "All at head e5fb2be3b. RED BEFORE (base 9180fd139 + pins only): predicted 7 of 15 new tests failing, being the 3 marked arms per toast surface plus the marked-409 dialog arm — actual `Test Files 3 failed | 1 passed (4)` / `Tests 7 failed | 13 passed (20)`, matching in BOTH count and identity; no prediction wrong. GREEN AFTER: `Test Files 5 passed (5)` / `Tests 28 passed (28)`. Package suites: `vitest run packages/plugin-{kanban,calendar,form}/` → `Test Files 97 passed (97)` / `Tests 862 passed (862)`, exit 0. Repo-state family (the one that reads .changeset/, re-run after the changeset commit so the reported green is on the pushed head): `vitest run scripts/__tests__/` → `Test Files 79 passed (79)` / `Tests 2279 passed (2279)`, exit 0. Type-check: `turbo run type-check` 3 pkgs → `Tasks: 17 successful, 17 total`, exit 0 — and NOT the 'compiled but excluded your tests' reading: `tsc -p tsconfig.test.json --listFiles` reports each new test file present in its package's program (1 hit each). Dependency closure built first (`turbo run build --filter=^...`, 13/13) because tsconfig.test.json sets `paths: {}` and resolves @object-ui/* through built .d.ts — an unbuilt worktree would have given a false red. Every heavy run went through the shared verify lock; all verdicts read from the gate's own line, exit codes captured before any pipe.",
  "ablation": {
    "question": "are the UNMARKED arms load-bearing, or decorative?",
    "mutation": "ObjectKanban.tsx — keep the declaredUserMessage read, drop the isPermissionError substitution (the plausible wrong fix that leaks raw server text)",
    "predicted": "exactly 1 failure, specifically 'keeps the localized substitution for an UNMARKED 403'; all 3 marked arms and the other 2 unmarked arms stay green",
    "actual": "exactly that — `Tests 1 failed | 5 passed (6)`, the failure being that named test",
    "mutation_confirmed_on_disk": "injected marker present x1, removed anchor `errors.unauthorized` present x0, mutated blob hash differs from HEAD blob — not a no-op reading",
    "restore_proven": "by bytes, not exit code: git hash-object == HEAD blob 2c99db839a55bcbbb25e6599fb8140f441967ab6, git diff HEAD empty, errors.unauthorized back to x1; script carried trap ... EXIT INT TERM with absolute paths",
    "note": "The red-before run is itself the stronger ablation for the marked arms — measured on the genuinely unmodified tree, before any source edit, so it has no restore leg to get wrong."
  },
  "rejectedMoveRollback_T5": "Nothing moved. The file is UNTOUCHED and absent from the diff; zero assertions changed. Its fixture (invalidTransition — a 400 invalid_transition) carries no userMessage, so declaredUserMessage answers null and the toast text is unchanged; its `toHaveBeenCalledWith('Invalid status transition')` still pins the extractWriteErrorMessage arm and the rollback behaviour is not weakened. Green 5/5 in both the red and the green runs.",
  "gate_results": {
    "green_exit_0": [
      "check-changeset-presence.mjs — '6 source file(s) of 3 released package(s) changed, and this change declares 1 changeset(s)'",
      "check-changeset-no-major.mjs — 'No changeset declares a major bump.' (patch bump, never major)",
      "check:control-bytes — 'OK (scanned 5248 tracked text file(s); skipped 85 binary)'",
      "check:vi-mock-specifiers — 'OK' (the kanban pin reuses the existing @dnd-kit/core mock pattern)",
      "check:phantom-deps — 'Every in-scope import is declared by the package that publishes it.'",
      "check:self-import — 'No package names itself inside its own src/.'",
      "check:i18n-keys — 'Every in-scope call-site key resolves against the en pack (2824 keys)' (no new key added on any surface)",
      "check:i18n-drift — 'No en value changed in this range.'",
      "type-check:coverage — '45/46' + 'test type-check coverage: 41/41 packages compile their tests'",
      "lint:coverage — '46/46 packages linted, 0 with outstanding errors (0 total)'",
      "check:entry-guard, check:spec-symbols, check:pre-install-import-graph, check:shell-escape-residue — all OK",
      "check:esm-specifiers — exit 0, specifier leg only ('the load leg (which builds) was not run')",
      "turbo run type-check (3 pkgs) — 'Tasks: 17 successful, 17 total'",
      "turbo run lint (3 pkgs) — 'Tasks: 4 successful, 4 total'",
      "vitest packages/plugin-{kanban,calendar,form}/ — 97 files / 862 tests",
      "vitest scripts/__tests__/ — 79 files / 2279 tests"
    ],
    "not_measured": [
      "check:eager-closure (exit 2) — 'No eager-closure report at apps/console/dist/eager-closure.json ... This is a broken gauge, not a passing budget.' Needs a console vite build; CI builds it.",
      "check:readme-exports (exit 1) — 'the population COLLAPSED -- this run proves nothing'; 23 of 40 packages unbuilt in this worktree. CI installs and builds.",
      "check:published-dist (exit 143) — killed by the container's 10-minute foreground cap mid-build. Not implicated: no package.json or build config changed."
    ],
    "not_applicable": "half-state-patrol is a scheduled board sweep triggered by scripts/pm/check-half-states.mjs (untouched). Docs / shadcn / spec-floors / skills-paths workflows match no path in this diff.",
    "lint_narrowing_declared": "Repo-wide `pnpm lint` narrowed to CI's own command scoped to the three packages (each package in full, not only changed files). Three pieces of evidence it excluded nothing: (1) population from eslint's own config resolution, not a guess; (2) `--format json` reports 'files linted: 6 | errors: 0 | warnings: 104'; (3) linting is NOT type-aware — eslint.config.js declares no parserOptions.project and no projectService — so this diff cannot move any untouched file's verdict. The 104 warnings are pre-existing no-explicit-any / react-hooks findings on the two large existing components; all three NEW test files are 0/0.",
    "note": "skip-changeset is N/A twice over: this PR ships a real changeset, and that label does not exist in objectui."
  },
  "false_assumptions": [
    "The card's census of TWO was incomplete — finding 1 confirmed independently: three surfaces. occSave.tsx had no userMessage reader, so no fork arose.",
    "MY OWN near-miss, worth passing on: I first named the changeset `.changeset/olive-donkeys-shave.md`, which ALREADY EXISTED as a tracked pending changeset for objectui#5846, and the heredoc clobbered it in the working tree. Caught because `git status` showed ` M` rather than `??`; restored byte-exact with `git checkout HEAD --` and nothing was lost (it was committed). Renamed to the repo's own issue-scoped convention `<issue>-<slug>.md`. `.changeset/` holds hundreds of pending files, so `pnpm changeset`-style random names are a live collision surface for every parallel agent — the issue-scoped convention is the fix, and checking `git status` for ` M` instead of `??` is the detector."
  ],
  "open_questions": [],
  "out_of_scope_findings": [
    "filed as #6375: normaliseClientError's first CONCURRENT_UPDATE guard is subsumed by the line under it, so its `httpStatus === 409` half decides nothing — dead code advertising an acceptance path that does not exist, plus a related asymmetry with the exported isConcurrentUpdateError. Observation-class, `finding` label, unassigned, no pm:queue. Recorded not edited: packages/data-objectstack is scoped out of this card and this PR does not touch it."
  ],
  "stopped_on": null
}

Generated by Claude Code

Navigation Menu

Repository navigation

finding(ui): the kanban and calendar drag-write surfaces still discard a producer-marked refusal message #5902

What

Why it is cheap now

Not a regression

Claim — `domain:ui` @ objectui execution seat · the serial blocker has landed

⭐ Unblocked — and this is why it was held

Pre-dispatch gate — all three steps

⭐ What #5901's run learned that this card should not rediscover

The ruling this inherits

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

finding(ui): the kanban and calendar drag-write surfaces still discard a producer-marked refusal message #5902

Description

What

Why it is cheap now

Not a regression

Activity

claude commented on Aug 25, 2026

os-support-ai commented on Aug 25, 2026

Claim — domain:ui @ objectui execution seat · the serial blocker has landed

⭐ Unblocked — and this is why it was held

Pre-dispatch gate — all three steps

⭐ What #5901's run learned that this card should not rediscover

The ruling this inherits

os-support-ai commented on Aug 25, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

Claim — `domain:ui` @ objectui execution seat · the serial blocker has landed