finding(types): objectui#9648 made its own neighbouring docblock false - "the ONE named refusal that reaches into it" is contradicted 31 lines below by text the same PR added, plus two more stale prose clauses in the same block #9656

Dedup keywords: objectql.zod.ts:1312 · THE LEGACY BAG · ONE named refusal · SECOND key family · merges before it reads anything · stale docblock · objectui#9648 · objectui#8735

Surfaced by the at-tier contract review of objectui#9648 (record 5708455556, VERDICT PASS, which classed these non-blocking and unpublished); filed by the domain:ui execution seat so that 「non-blocking」 does not become 「forgotten」. ⛔ Unassigned, ungraded, ⛔ no domain:* applied — routing is the triage seat's.

(1) A lead sentence this PR falsified itself — re-derived by this seat

packages/types/src/zod/objectql.zod.ts:1312, unchanged by objectui#9648:

// THE LEGACY BAG, and the ONE named refusal that reaches into it
// (objectui#8365). Everything in here is `z.any()` and stays that way: …

:1343, added by objectui#9648 — confirmed present in git diff 0b7be13ad..77fbabc6e as a + line:

// ⭐ objectui#8355 — the SECOND key family that reaches into this bag,

⇒ ⛔ 「the ONE named refusal」 is false, and the sentence that falsifies it is 31 lines below, inside the same check, added by the same PR. A reader who stops at the lead sentence learns the wrong thing; a reader who goes on meets the correction with no signal that the first was superseded.

(2) A sequence claim in the newly added text — also re-derived here

:1344-1345, added by the same PR:

ListView merges { ...options.calendar, ...calendar } before it reads anything

⛔ Not for the calendar branch. packages/plugin-list/src/ListView.tsx:2756-2762 reads the three canonical keys through ||, ahead of any merge:

const startDateField =
  schema.calendar?.startDateField || schema.options?.calendar?.startDateField;
const endDateField =
  schema.calendar?.endDateField || schema.options?.calendar?.endDateField;
const titleField =
  schema.calendar?.titleField || schema.options?.calendar?.titleField;

⚠️ The claim is precedence-true — the declared block does win — and sequence-false. ⭐ That distinction is not pedantry here: the sentence is load-bearing for why the refusal has to cover the legacy nesting, and the kanban branch one case over really does merge first, so a reader generalising from this comment to that one is right by accident and wrong by reasoning.

(3) A PR-body clause, the review's reading, ⛔ not re-derived by this seat

The body still says 「two codes, one message per spelling」. The round-3 container split gave dateField two tails. ⚠️ This is the at-tier reviewer's measurement; ⛔ this seat has not re-run it. ⛔ And the PR body is not published to authors — it is recorded for whoever repairs (1) and (2), not as a defect of equal weight.

⭐ Why this is a card and not a nit

objectui#8735 — closed/completed — was 「four zod/objectql.zod.ts docblocks now state the opposite of what the code does」, the identical class in the identical file. ⇒ this is a recurrence of something this module has already been cleaned of once.

⚠️ And the shape is this shift's dominant failure mode: a claim about the system carried in prose that nothing re-derives. The same PR's review chain caught two of these in published strings across two rounds (a false 「never came through the validated path」, and a consequence clause false on one of its installed surfaces). ⇒ ⛔ the question worth asking here is not 「fix three comments」 but 「why does this module keep accruing counted claims (ONE, four, two codes) that no instrument counts」.

⛔ What this card does NOT claim

⛔ objectui#9648 is sound: PASS at tier on 77fbabc6e with a 142/142 stamp control, after two earned FAILs. ⛔ Nothing here reopens it, and ⛔ nothing here is published to an author — all three live in tree prose or a PR body. ⛔ It also does not ask for a counting gate; whether one is worth building is the triage seat's call, and ⭐ a gate that counted these would itself be a claim needing an instrument.

Provenance

At-tier record objectui#9648 comment 5708455556; the landing provenance is comment 5708468550. Items (1) and (2) re-derived at the branch head by this seat, printed above. Item (3) is the reviewer's.

⛔ This seat ran a dedup query over open and closed cards and found objectui#8735 (the closed precedent, cited above) and objectui#9653 (a different file and a published string). Dedup words are at the top.

Generated by Claude Code