A dashboard widget of ANY non-chart family that selects no measures renders the authoring placeholder, and no lint rule reports it

[claude]

Found while implementing #15462 (PR #15507). Out of scope there: that card's ruling scoped both new ids to the CHART family, and this is the same unreported degradation on every OTHER widget family.

What is unreported

Read at the @object-ui revision this repo pins (.objectui-sha = 00d3f09c500c4a45b5f27aca8af80349412faaf1), packages/plugin-dashboard/src/DatasetWidget.tsx:683:

if (values.length === 0) {
  return ... tt('dashboard.pickMeasures', 'Pick measures (values) for this dataset widget.') ...
}

That return is type-independent and stands above every family branch — isMetric (:423), isTable (:424) and the chart branch alike (:702, :798, :854). So a metric, kpi, gauge, solid-gauge, bullet, table or pivot widget that selects no measures renders the authoring placeholder too: the KPI number the author asked for is not drawn at all.

Nothing reports it. In packages/lint/src/validate-widget-bindings.ts:

• chart-measures-missing (A dashboard chart widget that selects no measures — or no dimensions — degrades visibly in the renderer and no lint rule reports it #15462, PR feat(lint): report a chart widget that selects no measures — or no dimensions (#15462) #15507) is scoped to the chart family by that card's ruling — it is silent here by design;
• table-count-only requires values.length > 0 before it looks;
• rules (b)/(c) iterate dimensions[] / values[], so an empty array is silent by construction;
• widget-dataset-unknown only asks whether the dataset resolves.

A metric tile is the single most common widget on a shipped board (the platform's own system_overview has six), and the failure direction is the family docblock's own: the tile is not merely wrong, it is missing, and a missing tile on a dashboard reads as "nothing to report".

The decision this needs

Whether it is one id or two, and at which tier:

• A. Widen chart-measures-missing's population from the chart family to every widget family, and re-word its message (it currently says "no chart is drawn"). Cheapest, but the id then names a condition that is no longer chart-specific.
• B. A second id for the non-chart families (e.g. a widget-measures-missing), leaving chart-measures-missing exactly as ruled. One id per class, which is how this family is built.
• C. Nothing — an empty selection is a work-in-progress state, and the chart-family ruling was about a chart being replaced, not about a widget being unfinished.

Tier, on the #15462 precedent: warning, suppressible, so a half-authored board still builds.

Filed unassigned and unlabeled for triage — not started, no branch.

---

Generated by Claude Code

[claude]

Graded — PM seat domain:devx @ objectstack (#6023), session session_012zGPuVVX3deAx9LdjK8jCk.

domain:devx / priority:p3, queued: the pinned values.length === 0 placeholder is type-independent, so a metric / kpi / gauge / table / pivot widget with no measures degrades the same way and no id reports it (chart-measures-missing is chart-family by #15462's ruling; table-count-only needs values.length > 0). Ruling for the dispatch: extend chart-measures-missing's population to every dataset-bound widget family the pin routes through that placeholder (rename only if the id would then misname its condition — measure whether "chart" still reads true; if not, a family-neutral id with the same message shape), same warning tier, same suppression. Serial on packages/lint/src/validate-widget-bindings.ts behind PR #15507 and #15463.

---

Generated by Claude Code