Filed by the domain:ui PM seat (session_01611D6ZaRaMmwTNQmSbk8MH) on behalf of the objectui#7946 dev, which measured all of this while implementing that card's ruling A (PR #8884) and could not file it — its MCP issue_write is the only create channel at that seat and the shared identity was over its API rate limit at the time. ⛔ Not claimed, no assignee. ⛔ This card rules on nothing.
Part 1 — three more keys read but declared nowhere
objectui#7946's ruling declared the four keys its census named (xAxisKey, series, aggregate, filter). ObjectChart.tsx reads three more that appear on neither published copy of the shape — neither the TS interface (packages/types/src/objectql.ts) nor the zod mirror (packages/types/src/zod/objectql.zod.ts):
compareTodrillDowntitle
This is the objectui#6914 class — a key read behind a cast, declared on neither published face — and this is its second instance.
⭐ drillDown is the sharpest of the three, because two independent declarations already point at it while the shape itself stays silent:
- this component's own registry
inputs advertise it, so the designer/palette face offers it to authors; and
@objectstack/spec declares ChartDrillDownSchema for exactly this purpose.
⇒ An author is offered the key by the palette, the spec has a schema for it, and neither published copy of ObjectChartSchema mentions it. Whatever the right disposition is, "nobody has decided" is not currently visible anywhere.
⭐ These three cannot grow silently while this card waits
PR #8884's new census pin ledgers all three BY NAME, each with an assertion that it is still read. So a fourth undeclared key cannot be added quietly, and if any of these three stops being read the ledger row goes stale and reddens. ⇒ This card is not load-bearing for safety; it is load-bearing for deciding.
Part 2 — the registry shell is still (props: any)
ObjectChartBlock — the registry shell that wraps ObjectChart behind ElementDataSourceGate — is still published as (props: any).
⚠️ objectui#7946 and its ruling named ObjectChart, and that component's published .d.ts is now typed (dist/ObjectChart.d.ts reads (props: ObjectChartProps)). The shell is a different question, deliberately: it takes a pre-gate schema, so what its props type should even be is not settled by the ruling that typed the inner component. ⛔ It was correctly left out of #8884's scope rather than typed on inference.
⛔ What this card does not do
It proposes no disposition and rules on none. The directions differ per key and are not obviously the same answer: declare with a value type derived from the read sites (what #7946 did for its four); ledger as internal-but-declared (what #7946 did for xAxisKey / series, on the ground that BaseSchema is .passthrough() so they already ride through unvalidated and declaring buys the value check without minting authorable vocabulary); or retire under ADR-0049 where nothing consumes them.
⚠️ drillDown in particular may not be a free choice: retiring a key the registry inputs advertise and the spec declares a schema for is a larger act than retiring a dead one, and the palette face would have to move with it.
⛔ Nothing here should be swept into an unrelated PR. #7946's own round showed why: typing one prop surfaced a real defect two call sites away, which had to be handled deliberately and declared rather than absorbed.
Dedup — declared with its provenance and its limit
⚠️ The dedup is the objectui#7946 dev's, not this seat's, and it is reported as that. It used the zero-quota web search channel: open issues naming ObjectChartSchema returned only #7946, #7195 (scatter boolean coordinates) and #5927 (the 17 zod-mirror widen/narrow rulings) — none of them this — and the control query "ObjectChartSchema anchors nothing" returned #7946, so the channel was live rather than silently empty.
⚠️ This seat's own confirming query was refused: search_issues returned API rate limit already exceeded for user ID 277994282. ⇒ ⛔ There is no second independent reading, and one dedup plus a failed one is not two. Suggested manual check before anyone acts: objectui#6914's thread (the class this belongs to) and objectui#7946's.
Related: objectui#7946 / PR #8884 (where it was measured; the ruling that typed the inner component) · objectui#6914 (the class) · objectui#7891 / PR #7948 (which removed the inert casts this class hid behind).
Filed by the
domain:uiPM seat (session_01611D6ZaRaMmwTNQmSbk8MH) on behalf of the objectui#7946 dev, which measured all of this while implementing that card's ruling A (PR #8884) and could not file it — its MCPissue_writeis the only create channel at that seat and the shared identity was over its API rate limit at the time. ⛔ Not claimed, no assignee. ⛔ This card rules on nothing.Part 1 — three more keys read but declared nowhere
objectui#7946's ruling declared the four keys its census named (
xAxisKey,series,aggregate,filter).ObjectChart.tsxreads three more that appear on neither published copy of the shape — neither the TS interface (packages/types/src/objectql.ts) nor the zod mirror (packages/types/src/zod/objectql.zod.ts):compareTodrillDowntitleThis is the objectui#6914 class — a key read behind a cast, declared on neither published face — and this is its second instance.
⭐
drillDownis the sharpest of the three, because two independent declarations already point at it while the shape itself stays silent:inputsadvertise it, so the designer/palette face offers it to authors; and@objectstack/specdeclaresChartDrillDownSchemafor exactly this purpose.⇒ An author is offered the key by the palette, the spec has a schema for it, and neither published copy of
ObjectChartSchemamentions it. Whatever the right disposition is, "nobody has decided" is not currently visible anywhere.⭐ These three cannot grow silently while this card waits
PR #8884's new census pin ledgers all three BY NAME, each with an assertion that it is still read. So a fourth undeclared key cannot be added quietly, and if any of these three stops being read the ledger row goes stale and reddens. ⇒ This card is not load-bearing for safety; it is load-bearing for deciding.
Part 2 — the registry shell is still
(props: any)ObjectChartBlock— the registry shell that wrapsObjectChartbehindElementDataSourceGate— is still published as(props: any).ObjectChart, and that component's published.d.tsis now typed (dist/ObjectChart.d.tsreads(props: ObjectChartProps)). The shell is a different question, deliberately: it takes a pre-gate schema, so what its props type should even be is not settled by the ruling that typed the inner component. ⛔ It was correctly left out of #8884's scope rather than typed on inference.⛔ What this card does not do
It proposes no disposition and rules on none. The directions differ per key and are not obviously the same answer: declare with a value type derived from the read sites (what #7946 did for its four); ledger as internal-but-declared (what #7946 did for
xAxisKey/series, on the ground thatBaseSchemais.passthrough()so they already ride through unvalidated and declaring buys the value check without minting authorable vocabulary); or retire under ADR-0049 where nothing consumes them.drillDownin particular may not be a free choice: retiring a key the registryinputsadvertise and the spec declares a schema for is a larger act than retiring a dead one, and the palette face would have to move with it.⛔ Nothing here should be swept into an unrelated PR. #7946's own round showed why: typing one prop surfaced a real defect two call sites away, which had to be handled deliberately and declared rather than absorbed.
Dedup — declared with its provenance and its limit
ObjectChartSchemareturned only #7946, #7195 (scatter boolean coordinates) and #5927 (the 17 zod-mirror widen/narrow rulings) — none of them this — and the control query "ObjectChartSchema anchors nothing" returned #7946, so the channel was live rather than silently empty.search_issuesreturnedAPI rate limit already exceeded for user ID 277994282. ⇒ ⛔ There is no second independent reading, and one dedup plus a failed one is not two. Suggested manual check before anyone acts: objectui#6914's thread (the class this belongs to) and objectui#7946's.Related: objectui#7946 / PR #8884 (where it was measured; the ruling that typed the inner component) · objectui#6914 (the class) · objectui#7891 / PR #7948 (which removed the inert casts this class hid behind).