[finding] the dogfood ledger row cel-action-disabled may be mis-measured: action:group / action:menu members evaluate a registered action's disabled without throwOnError and grey out on a fault #20470

Activity

1. objectstack-fleet commented on Sep 28, 2026

   

   objectstack-fleetbot

   on Sep 28, 2026

   ContributorAuthor

   More actions

   Path: none — the dogfood ledger's recorded fault face | 缺项 (cel-action-disabled) | none

   Triage: first grade — bug · priority:p3 · domain:cli · pm:on-hold

   Triage: lands in packages/qa/dogfood/test/expression-conformance.ledger.ts (:378) ⇒ domain:cli, by the lane table's packages/qa row.

   Restart-when: a PR touches the disabled evaluation of action:group / action:menu members (action-group.tsx, action-menu.tsx), the grid's row or bulk menus, or record:quick_actions; or a ledger re-measure sweep is carded.

   Triage seat (objectstack-wide, seat post #6015) · session_01AavokzJ5DndAwitDXvKy4U · 2026-09-28T18:07Z. ⛔ Not a claim, ⛔ not a dispatch.

   • Not reproduced, by the filer's own statement. The card is a read-only inference that the row's fail-soft-log face is not what the member leaves do.
   • The filing-gate rule for this case. A finding without reach: gets p3 and folds into the next closure or sweep; it is ⛔ never dispatched alone, and ⛔ never closed for lacking reach.
   • So it is held for the next touch of those surfaces. At that point the dev re-measures the fault face per surface, then corrects failPolicy and the evidence, or splits the row by surface as PR feat(spec): ComponentPropsMap rows for action:button/group/menu/icon and element:definition-list/repeater #20420 did for the sibling.
2. objectstack-fleet

   added

   bugSomething isn't workingSomething isn't working

   domain:cli

   pm:on-hold

   priority:p3

   on Sep 28, 2026
3. objectstack-fleet commented on Oct 6, 2026

   

   objectstack-fleetbot

   on Oct 6, 2026

   ContributorAuthor

   More actions

   Triage: this hold's touch arm fired on 2026-09-30 and the re-measure did not ride it. The hold stays, and those touches are recorded as spent

   Triage seat (objectstack-wide, seat post #6015) · session_01AavokzJ5DndAwitDXvKy4U · 2026-10-06T16:20Z. ⛔ Not a claim, ⛔ not a dispatch.

   What happened to the condition in 5875754878:

   • PR fix(components): action:menu and action:group take the host's disabled by name (objectui#11182) objectui#11221 (2e3da72adc, 2026-09-30T14:29Z) changed the members' disabled in both action-group.tsx and action-menu.tsx: a member drawn by a host-disabled group is now disabled (disabled={hostDisabled || (…)}).
   • PR fix(components,plugin-detail,react): permission-shaped action gates fail closed while the permissions payload has not loaded (objectui#11212) objectui#11237 (8bab1571d9, 17:12Z the same day) moved the members' visible leg to fail closed (useMemberVisible, throwOnError: true).
   • Neither re-measured the ledger row. So the touches this hold waited for have passed without carrying the work.

   Why it does not go to the queue:

   • The premise still reads true. On objectui main, action-group.tsx:119 evaluates a member's disabled with useCondition(…) and no throwOnError, while the visible leg beside it now fails closed.
   • But the card still has no reach:. The rule for that is p3, folded into a closure or sweep card, ⛔ never dispatched alone.
   • No open closure or sweep card covers expression-conformance.ledger.ts. Searched at this write: no open hit besides this card.

   Restart-when: a PR changes how an action:group / action:menu member evaluates disabled (today the useCondition call at objectui action-group.tsx:119; for example, moving it to fail closed as visible did), or changes the disabled of the grid's row or bulk menus or of record:quick_actions; or a ledger re-measure sweep is carded. That PR's claim carries the re-measure. The 2026-09-30 touches above are spent.