feat(web,mobile): Today scheduling clarity, race-safe completion, URL day-nav

[thomasluizon]

Three correlated issues from one PRD (today-scheduling-concurrency-daynav), bundled into one frontend PR. Paired backend PR: thomasluizon/orbit-api#196.

#149 — Scheduling status on the Today list

• Overdue marker now also shows on child rows (dropped the !isChild gate).
• New future hint: relative in 6d (≤7 days), absolute scheduled 13 Jun beyond.
• The check-circle is disabled on recurring/flexible rows that can't be logged for the date, via a shared canLogHabitOnDate predicate mirroring the backend ValidateTargetDate (one-time tasks stay tappable). No API change.

#150 — Race-safe completion (client half)

• Replaced the decaying recentlyCompletedPromptIdsRef + 1400 ms timer with a deterministic, authoritative-cache-derived parent-prompt check + a promptedParentIdsRef once-guard (re-arms on done<total). The "complete the parent?" prompt fires exactly once per burst. Backend idempotency is in the paired PR.

#151 — URL-driven day navigation

• The selected day is now the URL (?date=YYYY-MM-DD); bare route = today; back steps days. selectedDate/followToday removed from persisted store state; midnight/foreground roll-to-today preserved via a today-tick.

Parity & validation

• web ↔ mobile mirrored (parity-checker ✅); shared helpers + i18n (en+pt-BR) in packages/shared (i18n-syncer ✅); canLogHabitOnDate ↔ backend ValidateTargetDate (contract-aligner, 0 drift).
• type-check ✅ · lint ✅ (0/0) · Vitest web 1465 / shared 850 / mobile 411 ✅.
• Playwright day-nav spec authored (apps/web/e2e/tests/22-day-nav.spec.ts); runs in CI.

Closes #149
Closes #150
Closes #151

🤖 Generated with Claude Code

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Actions	Updated (UTC)
orbit-ui-mobile-web	Ignored		Jun 8, 2026 6:28pm

[sonarqubecloud]

Quality Gate passed

Issues
1 New issue
0 Accepted issues

Measures
0 Security Hotspots
92.9% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[claude]

Three well-scoped features land together coherently and the implementation is tight throughout.

#149 – Scheduling clarity. Removing the !isChild gate on the overdue token is the right root-cause fix; the old gate was an arbitrary restriction. canLogHabitOnDate correctly mirrors LogHabitCommand.ValidateTargetDate — one-time tasks (no frequencyUnit) are still tappable for future dates, recurring/flexible are blocked; the disabled dot (opacity 0.4, non-interactive) is a clean UX signal. computeHabitFutureHint is pure, well-tested, and the 7-day threshold for relative-vs-absolute is sensible.

#150 – Race-safe completion. The replacement of the decaying recentlyCompletedPromptIdsRef + 1400 ms timer with promptedParentIdsRef as a once-guard is a genuine root-cause fix: the old approach raced against cache invalidation timing; the new one reads authoritative cache state and fires the prompt exactly once per burst, re-arming only when progress drops below total. The burst test (3 siblings → 1 dialog) covers the key invariant.

#151 – URL day navigation. Removing selectedDate/followToday from persisted store state is the right call — ephemeral navigation state has no business in localStorage/SecureStore. The today-tick refactor (useState(getTodayDate) + handleTodayRollover) handles midnight rollover and app-foreground correctly on both platforms without the store coupling. Store version bump to 2 and migration that strips the legacy keys is clean.

Parity, i18n, tests. Web ↔ mobile mirrored throughout; both locales updated together; shared helpers canLogHabitOnDate and computeHabitFutureHint live in packages/shared with their own unit tests; the E2E Playwright spec covers the golden path for URL-driven day nav; Vitest counts (web 1465 / shared 850 / mobile 411) confirm no regressions.

One minor cleanup (non-blocking): apps/web/e2e/tests/22-day-nav.spec.ts line 13 has an empty test.afterAll(async () => {}) stub — no-op, but per CLAUDE.md rule 2 stub functions should be deleted. Worth a follow-up squash or a quick cleanup commit before merge.