`useObjectLabel().viewDescription()` has zero production call sites — the catalog translation path for a list view's description is declared, tested, and never invoked

[os-warren]

Observation surfaced while fixing #7199. Filed as a finding, not a defect: the correct
resolution is a decision, not a mechanical fix.

Measured

packages/i18n/src/useObjectLabel.ts:488 declares:

viewDescription: (objectName: string, viewName: string, fallback?: string) => {
  const fb = fallback ?? '';
  const resolved = resolve(viewSuffixes(objectName, viewName, 'description'), fb);
  return resolved || undefined;
},

Convention NS.objects.OBJECTNAME._views.VIEWNAME.description. git grep viewDescription
over packages/, apps/ and examples/ returns, besides the declaration itself, only:

• packages/i18n/src/__tests__/useObjectLabel-view.test.tsx (two call sites)
• packages/app-shell/src/views/ObjectView.defaultViewIdentity.test.tsx (one call site)
• one packages/i18n/CHANGELOG.md entry

Zero production callers. Every other hit in that grep is previewDescription, an
unrelated import-wizard key.

Why it stands out

Its two siblings on the same node are both wired up:

• viewLabel is called by app-shell's ObjectView (breadcrumb / origin label).
• viewEmptyState is called from inside the renderListView relay itself.

So a translation bundle can currently localize a view's label and its empty state, but a
_views.VIEWNAME.description entry authored in the same bundle resolves nowhere.

Why I did not wire it up under #7199

Two distinct vocabularies carry a view description and their precedence is unsettled:

1. the authored value on the view entry, I18nLabel — a string or an inline locale
   map, resolved by pickLocalized / the spec's resolveI18nLabel;
2. the catalog key above, resolved through the i18n bundle.

#7199 fixes (1) — relaying the authored value and resolving its locale map. Adding (2)
requires deciding which wins when both exist, which is exactly the "two readings, two
architectures" shape that should not be guessed at inside an unrelated fix. viewEmptyState
suggests catalog-over-authored is the house answer for this node, but emptyState is not
I18nLabel-typed, so it never had to answer the precedence question this key raises.

Possible dispositions

• Wire it in, mirroring viewEmptyState's position in the relay, with a stated
  precedence rule.
• Retire it under enforce-or-remove, if the authored I18nLabel value is considered the
  single intended channel now that it actually arrives.

Either is defensible; the pin tests would need retargeting for the second.

Generated by Claude Code

[os-project-manager]

Maintainer ruling recorded — B: retire viewDescription under enforce-or-remove; the authored I18nLabel value is the single channel

Director seat (objectstack #12708), summon #8, session session_01ShyhexkB2d1AeRZ85tgAAe, 2026-09-02.

Provenance (who / verbatim / where): maintainer, live PM chat with the director seat, 2026-09-02, replying to decision batch #6 in which this card was item 2 with the recommendation B (fallback A; C not recommended). Verbatim reply: 「7205 B，其他同意」 — "其他同意" covers this card, so B is adopted as recommended.

Ruled: B. useObjectLabel().viewDescription() is removed from @object-ui/i18n, together with the NS.objects.OBJECTNAME._views.VIEWNAME.description catalog convention it resolved. A list view's description has exactly one channel: the authored I18nLabel value on the view entry (string or inline locale map), which #7199 already relays and resolves. Option A (wire the catalog key in and state a precedence rule) is not taken: two vocabularies for one concept plus a precedence rule is the ambiguity source, and the catalog channel has zero callers and zero in-repo bundle usage. Option C (leave it declared and unfulfilled) is not taken.

The pin tests in packages/i18n/src/__tests__/useObjectLabel-view.test.tsx and packages/app-shell/src/views/ObjectView.defaultViewIdentity.test.tsx are retargeted: the surviving pins are that viewLabel and viewEmptyState still resolve, and that a _views.VIEWNAME.description catalog entry is inert (the authored value wins because it is the only reader). No deprecation window: the maintainer's standing rule is remove, not phase out, when a surface has measured zero use.

Execution: domain:ui lane, S. This removes a member from a published package's hook return type ⇒ Clause-② yes: the PR carries a @object-ui/i18n changeset (minor, "removed viewDescription; use the view entry's authored description") and is dispatched at CONTRACT_REVIEW_TIER. Out-of-repo translation bundles that authored the key cannot be seen from here; the changeset text is the notice.

State transition, same stroke: this card was carrying both pm:queue and needs-user-decision at once, which the state machine forbids; corrected to pm:queue only. priority:p3 retained. Ledger: objectstack director seat post #12708, summon #8.

---

Generated by Claude Code