T10/T11: location seam and recording pipeline; fix a native crash-recovery bug

T10 -- LocationSource abstraction plus FakeLocationSource, so the riskiest code
in the app is testable with no device. Also found that geolocator ships its own
Android foreground service with enableWakeLock, which could drop
flutter_foreground_task and both its deprecation paths; deferred to T12 since it
costs notification actions.

T11 -- the pipeline. Kotlin's unbounded Channel plus blocking-receive writer
becomes a List buffer plus a periodic Timer; Dart's event loop makes a blocking
receive unnecessary and the guarantee is unchanged, since the fix callback only
appends and returns. 24 tests.

Found a real bug in the native app while porting. restoreAfterProcessDeath says
it resumes into a new segment because the dead time is a real gap, but it calls
resumeTrip, which adopts the segment a crash left open. After a pause that is
right; after a crash nothing closed it. Points either side of the dead time then
share a segment, and computeSummary -- the authoritative pass that overwrites the
live estimate on completion -- measures straight through the gap. Reverting the
fix and running the guard shows 111,217 m of phantom distance.

Fixed via TripRepository.resumeIntoNewSegment, which closes the stale segment at
its last recorded point rather than at now, so the dead time is not billed as
ride time either. A deliberate, documented departure from parity: the code
contradicted its own comment and produced silently wrong data.

The first version of that guard passed with the bug still present -- the fixture
used pause(), which closes the segment and stops reproducing the crash. Caught
only by reverting the fix and checking the test failed.

145 tests passing, analyze clean.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
This commit is contained in:
2026-08-15 11:19:21 -05:00
parent 3fd93cf8a6
commit 5a7c598610
5 changed files with 1087 additions and 0 deletions

View File

@@ -0,0 +1,440 @@
import 'package:drift/drift.dart' show driftRuntimeOptions;
import 'package:drift/native.dart';
import 'package:flutter_test/flutter_test.dart';
import 'package:rippr/src/data/database.dart';
import 'package:rippr/src/data/trip_repository.dart';
import 'package:rippr/src/domain/models.dart';
import 'package:rippr/src/recording/location_source.dart';
import 'package:rippr/src/recording/recording_engine.dart';
/// Covers `TrackingServiceLifecycleTest`'s ground, but on the Dart VM.
///
/// The Kotlin equivalent drove a real Android `Service` through all five actions on a
/// device, polled the database with a 25 s timeout, and was flaky enough that the timeout
/// was once raised to hide what turned out to be a real `stopSelf()` race. None of that
/// applies here: the pipeline is plain Dart behind a fake location source, so these run
/// deterministically in milliseconds.
void main() {
late AppDatabase db;
late TripRepository repo;
late FakeLocationSource source;
late RecordingEngine engine;
late int fakeNow;
setUp(() {
driftRuntimeOptions.dontWarnAboutMultipleDatabases = true;
db = AppDatabase(NativeDatabase.memory());
repo = TripRepository(db);
source = FakeLocationSource();
fakeNow = 1000;
engine = RecordingEngine(
repository: repo,
locationSource: source,
clock: () => fakeNow,
);
});
tearDown(() async {
await engine.dispose();
await source.dispose();
await db.close();
});
/// Lets the broadcast stream deliver. Fixes reach the engine on a microtask, not
/// synchronously, so every emission must be followed by this before asserting.
Future<void> settle() => Future<void>.delayed(Duration.zero);
/// Emits [n] fixes a second apart, moving ~11 m north each time.
Future<void> ride(int n,
{int startTs = 1000, double startLat = 51.0, double speedMps = 11.11}) async {
for (var i = 0; i < n; i++) {
source.emitAt(
timestamp: startTs + i * 1000,
latitude: startLat + i * 0.0001,
speedMps: speedMps,
);
}
await settle();
}
group('the fix path', () {
test('a started ride buffers fixes without writing immediately', () async {
await engine.start();
await ride(5);
expect(engine.pendingCount, 5,
reason: 'the fix path must not touch the database');
expect(await db.countPointsForTrip(engine.currentTripId), 0);
});
test('fixes are ignored before start and after stop', () async {
await ride(3);
expect(engine.pendingCount, 0, reason: 'source is not running yet');
await engine.start();
await engine.stop();
await ride(3);
expect(engine.pendingCount, 0);
});
test('fixes worse than the accuracy budget are dropped', () async {
await engine.start();
source.emitAt(timestamp: 1000, accuracyM: 500);
await settle();
expect(engine.pendingCount, 0);
source.emitAt(timestamp: 2000, accuracyM: 5);
await settle();
expect(engine.pendingCount, 1);
});
test('speed is converted and the noise floor applied', () async {
await engine.start();
source.emitAt(timestamp: 1000, speedMps: 10.0); // 36 km/h
source.emitAt(timestamp: 2000, speedMps: 0.1); // 0.36 km/h, under the floor
await settle();
await engine.stop();
final points = await db.allPoints();
expect(points[0].speedKmh, closeTo(36.0, 1e-9));
expect(points[1].speedKmh, 0.0,
reason: 'jitter under the noise floor must read as zero');
});
test('live telemetry updates at fix rate', () async {
await engine.start();
source.emitAt(timestamp: 1000, speedMps: 20.0);
await settle();
expect(engine.pendingCount, 1);
// 20 m/s is 72 km/h.
// LiveTelemetry is a singleton; the value is what the record screen reads.
await engine.stop();
});
});
group('persistence', () {
test('stop drains everything buffered', () async {
await engine.start();
final tripId = engine.currentTripId;
await ride(7);
await engine.stop();
expect(await db.countPointsForTrip(tripId), 7);
expect(engine.pendingCount, 0);
});
test('aggregates are persisted and then reconciled on stop', () async {
await engine.start();
final tripId = engine.currentTripId;
await ride(11); // ten hops of ~11.12 m
await engine.stop();
final trip = (await repo.tripById(tripId))!;
expect(trip.pointCount, 11);
expect(trip.distanceM, closeTo(111.2, 3.0));
expect(trip.state, TripState.completed);
expect(trip.endedAt, isNotNull);
});
test('a ride that captured nothing is discarded, not saved', () async {
await engine.start();
final tripId = engine.currentTripId;
final result = await engine.stop();
expect(result, isNull);
expect(await repo.tripById(tripId), isNull,
reason: 'an empty trip is noise in the history list');
expect(await db.countTrips(), 0);
});
});
group('pause and segments', () {
test('the full lifecycle produces one trip and two segments', () async {
await engine.start();
final tripId = engine.currentTripId;
await ride(3, startTs: 1000);
await engine.pause();
await ride(3, startTs: 5000); // ignored: the source is stopped
fakeNow = 6000;
await engine.start(); // resume
await ride(3, startTs: 6000, startLat: 51.001);
await engine.stop();
expect(await db.countTrips(), 1);
final segments = await repo.segmentsForTrip(tripId);
expect(segments.length, 2);
expect(segments.every((s) => !s.isOpen), isTrue);
expect(await db.countPointsForTrip(tripId), 6,
reason: 'fixes emitted while paused must not be recorded');
});
test('a fix in flight at pause lands in the segment it belongs to', () async {
// This is the guarantee that stamping ids at creation exists for. The fix is
// buffered before the pause and written after it; it must carry the OLD segment id.
await engine.start();
final tripId = engine.currentTripId;
final firstSegment =
(await repo.segmentsForTrip(tripId)).single.id;
await ride(2); // buffered, not yet written
expect(engine.pendingCount, 2);
await engine.pause();
final points = await db.pointsForTrip(tripId);
expect(points.length, 2);
expect(points.every((p) => p.segmentId == firstSegment), isTrue,
reason: 'a queued fix must not be re-homed into the next segment');
});
test('distance does not span the pause', () async {
await engine.start();
final tripId = engine.currentTripId;
await ride(3, startLat: 51.0);
await engine.pause();
fakeNow = 6000;
await engine.start();
// Resumed a full degree of latitude away — a trailered gap of ~111 km.
await ride(3, startTs: 6000, startLat: 52.0);
await engine.stop();
final trip = (await repo.tripById(tripId))!;
expect(trip.distanceM, lessThan(200.0),
reason: 'the pause gap leaked into distance: ${trip.distanceM} m');
});
test('pause then stop completes the trip normally', () async {
await engine.start();
final tripId = engine.currentTripId;
await ride(3);
await engine.pause();
final result = await engine.stop();
expect(result, tripId);
expect((await repo.tripById(tripId))!.state, TripState.completed);
});
});
group('discard', () {
test('discard deletes the trip and everything buffered', () async {
await engine.start();
final tripId = engine.currentTripId;
await ride(5);
await engine.discard();
expect(await repo.tripById(tripId), isNull);
expect(await db.countTrips(), 0);
expect(engine.pendingCount, 0);
expect(engine.state, RecorderState.idle);
});
test('discard leaves earlier completed rides alone', () async {
await engine.start();
final keep = engine.currentTripId;
await ride(3);
await engine.stop();
fakeNow = 10000;
await engine.start();
await ride(3, startTs: 10000);
await engine.discard();
expect(await repo.tripById(keep), isNotNull);
expect(await db.countTrips(), 1);
});
});
group('idempotency and state', () {
test('start twice adopts rather than duplicating', () async {
await engine.start();
final tripId = engine.currentTripId;
await engine.start();
expect(engine.currentTripId, tripId);
expect(await db.countTrips(), 1);
expect(await db.countSegmentsForTrip(tripId), 1);
});
test('pause twice is safe', () async {
await engine.start();
await engine.pause();
await engine.pause();
expect(engine.state, RecorderState.paused);
});
test('stop with nothing running is a safe no-op', () async {
expect(await engine.stop(), isNull);
expect(engine.state, RecorderState.idle);
});
test('the source is started and stopped in step with the lifecycle', () async {
await engine.start();
expect(source.isRunning, isTrue);
await engine.pause();
expect(source.isRunning, isFalse);
await engine.start();
expect(source.isRunning, isTrue);
await engine.stop();
expect(source.isRunning, isFalse);
});
test('state transitions are observable', () async {
final seen = <RecorderState>[];
final sub = engine.stateStream.listen(seen.add);
await engine.start();
await engine.pause();
await engine.start();
await engine.stop();
await Future<void>.delayed(Duration.zero);
await sub.cancel();
expect(seen, [
RecorderState.recording,
RecorderState.paused,
RecorderState.recording,
RecorderState.idle,
]);
});
});
group('process death', () {
test('a recording trip is resumed into a new segment', () async {
await engine.start();
final tripId = engine.currentTripId;
await ride(5);
// Deliberately no stop(): dispose without completing simulates a process kill.
await engine.dispose();
// A brand new engine over the same database, as after a process restart.
final revived = RecordingEngine(
repository: repo,
locationSource: source,
clock: () => fakeNow,
);
addTearDown(revived.dispose);
fakeNow = 20000;
final state = await revived.restoreAfterProcessDeath();
expect(state, RecorderState.recording);
expect(revived.currentTripId, tripId);
final segments = await repo.segmentsForTrip(tripId);
expect(segments.length, 2,
reason: 'the dead time is a real gap and must render as one');
});
test('the dead time is never measured as distance', () async {
// The bug this guards is in the native app: after a crash the open segment is
// adopted rather than closed, so computeSummary -- the authoritative pass at trip
// completion -- measures straight across the gap. A rider who crashes in town and
// relaunches downtown would have those kilometres added to their ride.
await engine.start();
final tripId = engine.currentTripId;
final crashedSegment = (await repo.segmentsForTrip(tripId)).single.id;
// Write points the way the flush loop would have, *without* going through pause or
// stop — so the segment is left OPEN, which is exactly what a crash leaves behind.
// Using pause() here would close the segment and quietly stop reproducing the bug.
await repo.appendPoints([
for (var i = 0; i < 3; i++)
TrackPoint(
tripId: tripId,
segmentId: crashedSegment,
timestamp: 1000 + i * 1000,
latitude: 51.0 + i * 0.0001,
longitude: -114.0,
speedKmh: 40,
altitudeM: 1000,
),
]);
await engine.dispose();
final revived = RecordingEngine(
repository: repo,
locationSource: source,
clock: () => fakeNow,
);
addTearDown(revived.dispose);
fakeNow = 900000;
await revived.restoreAfterProcessDeath();
// Relaunched a full degree of latitude away — ~111 km from where it died.
await ride(3, startTs: 900000, startLat: 52.0);
await revived.stop();
final trip = (await repo.tripById(tripId))!;
expect(trip.distanceM, lessThan(200.0),
reason:
'the dead time leaked into distance: ${trip.distanceM} m — the crash gap '
'was measured as if it had been ridden');
});
test('restored totals continue rather than restarting from zero', () async {
await engine.start();
final tripId = engine.currentTripId;
await ride(11);
// Force the buffered points and their aggregates to disk without completing.
await engine.pause();
final before = (await repo.tripById(tripId))!;
expect(before.distanceM, greaterThan(100.0));
await engine.dispose();
final revived = RecordingEngine(
repository: repo,
locationSource: source,
clock: () => fakeNow,
);
addTearDown(revived.dispose);
fakeNow = 30000;
await revived.start(); // resumes the paused trip
await ride(3, startTs: 30000, startLat: 53.0);
await revived.stop();
final after = (await repo.tripById(tripId))!;
expect(after.pointCount, 14, reason: 'earlier points must be kept');
expect(after.distanceM, greaterThanOrEqualTo(before.distanceM),
reason: 'distance must not restart from zero');
});
test('a paused trip is restored as paused, not resumed', () async {
await engine.start();
await ride(3);
await engine.pause();
await engine.dispose();
final revived = RecordingEngine(
repository: repo,
locationSource: source,
clock: () => fakeNow,
);
addTearDown(revived.dispose);
final state = await revived.restoreAfterProcessDeath();
expect(state, RecorderState.paused);
expect(source.isRunning, isFalse,
reason: 'a paused ride must not silently start consuming GPS');
});
test('no active trip restores to idle', () async {
final state = await engine.restoreAfterProcessDeath();
expect(state, RecorderState.idle);
});
});
}