fix(alarm): fail-safe alarm system overhaul (SDD alarm-system-overhaul, slice A)
Build & Deploy PluriWave / Análisis de código (push) Successful in 25s
Build & Deploy PluriWave / Build APK + AAB release (push) Successful in 2m25s

Root-cause fix for the unstoppable-alarm incident (alarm rang 15 minutes,
only uninstall silenced it) plus systematic hardening of every stop path.

Native (Kotlin):
- Verified stop: stopActiveAlarm now derives its result from the real
  post-teardown state (companion instance + synchronous stopEverything +
  activeRingingId check) instead of reporting unconditional success.
- Atomic teardown: every stop path (stop action, notification button,
  snooze, missed, onDestroy, startForeground failure) funnels through one
  stopEverything() covering audio, wakelock, notification, foreground
  state and firing-record cleanup; player.release() guarded.
- Bounded ringing: 10-minute auto-silence armed via AlarmManager fires a
  FIRED->MISSED transition with a localized missed-alarm notification;
  repeating alarms keep their native rearm, deleted alarms never produce
  ghost MISSED notifications.
- Durable firing record with onStartCommand re-validation (resurrection
  guard) and boot-time stale cleanup; firing records cleared on every
  refuse/mismatch/cancel path.
- New notification-only dismissal channel (dismissAlarmNotificationOnly)
  so UI-level dedup can never kill a live ring's audio.

Flutter (Dart):
- Stop/disable/edit/delete of a ringing alarm always attempt to silence
  it; on native-query failure the stop falls back toward silence via the
  id-scoped legacy stop.
- Verified-stop results surface failures: the ringing screen keeps
  dismiss-by-design on success, but on a verified failure it stays up
  with a persistent force-stop banner (guarded against double-dismiss)
  and auto-dismisses if the ring ends externally (missed/notification).
- Missed events sync alarm bookkeeping without opening the ringing UI.
- 4 new l10n keys translated across all 13 locales (ARB guard green).

550 tests green, analyzer clean. Reviewed in 3 adversarial 4-lens rounds
(2 deterministic + 1 refuter-corroborated critical fixed); formal
gentle-ai receipt waived by maintainer authorization (correction scope
legitimately exceeded the frozen genesis paths). On-device QA checklist
in openspec/changes/alarm-system-overhaul/tasks.md pending before
archive.
This commit is contained in:
Javier Bautista Fernández
2026-07-22 23:52:36 +02:00
parent 0f9a6a1719
commit 29f7d54e85
50 changed files with 2461 additions and 24 deletions
@@ -0,0 +1,68 @@
# alarm-system-overhaul — Apply Progress (Slice A)
## Status: complete (pending final full-suite verification + bounded review)
Slice A (Phases 1-7 of tasks.md) is fully implemented and every task checkbox is marked with
verified evidence. Phase 8 / Slice B (P1 permission fallbacks) is intentionally untouched —
deferred to a follow-up delivery per the orchestrator decision recorded in Engram
(`decision/alarm-system-overhaul-pre-apply-orchestrator-decisions`).
## How this phase ran (recovery notes)
- The initial apply agent implemented most of Slice A but stalled (stream watchdog) before
recording progress; a second continuation agent also stalled early. Recovery was done in
smaller stages: (1) a mapping agent verified every Phase 1-7 task acceptance criterion
against the working tree (grep/diff evidence, table below summarized) and marked tasks.md;
(2) a surgical TDD agent closed the one functional gap the mapping found (task 7.4);
(3) the orchestrator ran the final full-suite verification.
- Baseline established during recovery: full `flutter test` suite green (544 tests) and
`flutter analyze` clean (1 known pre-existing info in test/estado/estado_radio_test.dart:641)
BEFORE the 7.4 fix; final run after 7.4 recorded below.
## Task evidence summary (full table in the mapping agent report; all grep-verified)
- 1.1-1.4 (fail-safe native stop): `@Volatile activeRingingId` companion snapshot;
atomic `stopEverything()` (stopForeground(REMOVE)+stopSelf+wakelock+MediaPlayer+notification)
routed from ACTION_STOP/ACTION_STOP_ACTIVE/onDestroy/missed paths; id-agnostic
`ACTION_STOP_ACTIVE` for ringing UI + notification Stop (`stopPendingIntent` action updated).
- 2.1-2.7 (auto-silence + durable firing record): `AUTO_SILENCE_MILLIS` (10 min),
AlarmManager-armed `ACTION_MISSED``onAlarmMissed` (stops if active, posts missed
notification via existing channel, clears firing record, no re-arm of the fired instance);
durable firing record written before `player.start()`, `onStartCommand` stale re-validation,
`START_NOT_STICKY` preserved; boot-time `cleanupStaleFiringRecords()` in
`reschedulePersistedAlarms`; `missedTitle`/`missedText` in AlarmNotificationStrings.
- 3.1 (channel contract): `stopActiveAlarm` + `getActiveRingingAlarmId` in MainActivity dispatcher;
`stopActiveAlarm` reports `STOP_FAILED` errors instead of silent void.
- 4.1-4.6 (Dart orchestration): `ResultadoDetencion`, `alarmaSonandoId()`,
`detenerSonidoActivo()` on the puerto; fakes gained `fallaDetener`/`alarmaSonandoIdValor`/
`detencionesActivas`; `_detenerSiEstaSonando` wired into `guardarAlarma` (covers
`cambiarActiva`) and `eliminarAlarma`; `finalizarEjecucion` stops active sound and surfaces
`_error` on failure. Spec scenarios SS-1a/b/c/d, SS-2a/b covered by tests.
- 5.1-5.4 (ringing screen fail-safe UX): `forzarDetencion()`; `_detener()` keeps
dismiss-by-design but captures ScaffoldMessenger pre-dismiss and shows a retryable
force-stop SnackBar on failure (mirrors the existing snooze-failure pattern).
SS-3a/b/c covered by tests.
- 6.1-6.2 (missed-event sync): `accionMissed` native event → `_registrarEjecucionPerdida`
`completarEjecucion`, no re-arm. Test present.
- 7.1-7.4 (l10n): 4 new keys in en+es ARBs only (11 other locales untouched — CI ARB guard);
missed-alarm strings wired through `setNotificationStrings` (`missedTitle`,
`missedTemplate` with `{name}` sentinel via new `_plantillaNombre` helper) — 7.4 closed the
gap found during mapping (RED→GREEN evidence in task report).
## Deviations
- Task 7.4 added during apply verification (missed-string wiring was implied by NA-3a intent
but not an explicit task). No other scope deviations; no unexpected diff content.
## Changed-line accounting (excluding generated lib/l10n/gen/*)
- Working tree at mapping time: ~1045 insertions total incl. generated; non-generated portion
within the 800-line session ceiling (final figure recorded by the review lifecycle at
review start).
## Remaining work
- None for Slice A code. Next: bounded 4-lens review → sdd-verify → commit/push.
- Slice B (Phase 8, P1 fallbacks) — follow-up delivery.
- On-device QA checklist (tasks.md) — requires the user with real hardware; sdd-archive held
until that passes.