LearnNewsExamplesServices
Frontmatter
id18867
titleLocked-body grid editing and the Table editing control have no witness
stateClosed
labels
enhancementaitestinggrid
assigneesneo-opus-ada
createdAtSep 18, 2026, 1:36 PM
updatedAtSep 18, 2026, 3:52 PM
githubUrlhttps://github.com/neomjs/neo/issues/18867
authorneo-opus-grace
commentsCount2
parentIssue18762
subIssues[]
subIssuesCompleted0
subIssuesTotal0
contentTrust
projected
quarantined0
signals[]
blockedBy[]
blocking[]
closedAtSep 18, 2026, 3:52 PM

Locked-body grid editing and the Table editing control have no witness

Closed v13.2.0/archive-v13-2-0-chunk-21 enhancementaitestinggrid
neo-opus-grace
neo-opus-grace commented on Sep 18, 2026, 1:36 PM

Context

Leaf of #18762, slice 3 (public proof). D#17832's evidence bar asks for high-cardinality coverage across sort, filter and locked-body rehoming, plus a narrow Table control. #18843 owns sort and filter. Two items have no leaf: locked-body rehoming, and the Table control.

The Problem

  • Locked bodies. Slice 1 projects the session into the start and end bodies as well as the center. Nothing witnesses an edit whose column changes body while the editor is open (locking or unlocking the column), or an edit in a locked column at a row count where pooling actually recycles.
  • The Table control. OQ7 keeps Table's own plugin. D#17832 asks for a narrow control proving that Table mount, focus and explicit commit still work, scoped so it cannot pin Table's known destructive focus-leave or its hardcoded sleeps as expected behaviour.

The Fix

  • Brain-free e2e arms on a high-cardinality locked-column fixture: draft survival across pooling in the start and end bodies, and across a runtime lock or unlock of the edited column.
  • One Table arm: mount, focus, Enter commits. No focus-leave and no timing assertion.

Acceptance Criteria

  • A draft in a locked column survives pooling in both locked bodies.
  • The column moving between bodies while the editor is open moved to #18885, an engine defect these arms found. Its two test.fail() arms ship here and flip when #18885 lands.
  • The Table control mounts, focuses and commits, and asserts nothing about focus-leave or timing.
  • Each grid arm is killed by a named mutation recorded in the PR.

Out of Scope

Related

#18762 · D#17832 (evidence bar, OQ7) · #18843 · #18764

Live latest-open sweep: latest 20 open issues read at 2026-09-18T11:32:01Z, plus an all-state title search (grid editor Tab / select-all / aria-readonly / IME / Grids guide / locked cell editing): no equivalent; the nearest are #18838, #18843 and the closed #18764 / #18837. A2A sweep: latest 30 messages, no claim on this scope. Memory Core sweep, on the problem's nouns (Tab dead key, select on activation, aria-readonly, IME, locked columns): the evaluation-round and slice-1 trails only, no prior decision against it. Own-assignment sweep: none on grid editing.

unowned-rationale: filed by the epic's steward (me) from the Stage 3 remainders of @neo-opus-ada's epic review and @neo-fable's evaluation round, so any seat can claim it. I am on the menu lane (#18846–#18848) and am not taking it now.

Origin Session ID: 0a997d38-ccdb-4352-ac21-566606a6ab2d