MirrorSyncClient promised that reads consult both backends "so the cutover order is irrelevant", but every read called self.primary unguarded. A Firebase outage therefore produced no data at all and the GitHub mirror was never consulted -- the union read silently degraded to nothing, precisely when the fallback existed to help. This is not hypothetical: screen-locker's workout sync hit it on 2026-08-10, where a DNS failure reading the primary took down the whole locker rather than falling back. Reads (list_directory, get_file_text, get_string_map) now degrade to the mirror when the primary raises, and only fail when NEITHER backend can answer -- returning an empty result would be indistinguishable from "no data", which callers act on. Writes keep the old asymmetry on purpose: a write the primary did not accept has not happened, so it must still fail the tick. A mirror that cannot bulk-read (GitHub) contributes nothing but is not a failure, so it cannot by itself make a revision-map read raise. Both language ports agree on that. Same fix in the Dart port, which had all three read paths wrong too and runs on the phone. Two older tests asserted the buggy read-fail-closed behaviour and are renamed to describe what they actually exercise: the both-sides-failed case. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01JS5V44jkHZiH1wC7BH5h5n |
||
|---|---|---|
| .. | ||
| lib | ||
| test | ||
| tool | ||
| .gitignore | ||
| analysis_options.yaml | ||
| LICENSE | ||
| pubspec.yaml | ||
| README.md | ||
crdt_sync
Shared CRDT merge scheme + GitHub-Contents-API sync transport, extracted for
reuse across four personal apps that each need cross-device sync: todo,
diet_guard's Flutter app, wake_alarm's phone_app, and screen-locker's
workout_app. Pure Dart -- no Flutter SDK dependency (package:http only)
-- so it works in both Flutter apps and plain Dart tooling. Mirrors the
Python crdt-sync package
module-for-module and test-for-test, so the merge algorithm canonically
agrees on both sides of any sync between a Dart client and a Python one.
The merge scheme
See the Python package's README for the full rationale; in short, a generic LWW-map-with-sticky-remove:
Hlc-- a Hybrid Logical Clock(wallTimeMs, counter, nodeId), totally ordered, monotonic even across clock skew.Record-- an id, a per-field last-writer-wins map (fieldName -> (value, Hlc)), and a stickydeletedflag (deleted = a.deleted || b.deleted, never the other way -- a delete can never be silently undone).Log--Map<String, Record>, merged by taking the union of ids and merging shared ones.
All three merges are commutative and idempotent -- see test/log_test.dart
for the concrete convergence-property tests.
Install
Consume via a git dependency on the utils monorepo:
dependencies:
crdt_sync:
git:
url: https://github.com/kuhyx/utils
ref: crdt_sync_dart-v0.2.0
path: crdt_sync_dart
Usage
import 'package:crdt_sync/crdt_sync.dart';
final nodeId = 'phone';
final clock = Hlc.newTick(nodeId);
final record = Record(
id: 'abc123',
fields: {'text': ('buy milk', clock)},
);
final client = GitHubClient(owner: 'kuhyx', repo: 'my-app-sync', token: token);
final merged = await syncLog(
client: client,
deviceId: nodeId,
pathPrefix: 'devices',
localLog: {record.id: record},
encode: (log) => jsonEncode(log.map((k, r) => MapEntry(k, r.toJson()))),
decode: (text) => (jsonDecode(text) as Map<String, dynamic>).map(
(k, v) => MapEntry(k, Record.fromJson(v as Map<String, dynamic>)),
),
);
syncLog pulls every other device's last-pushed log from
<pathPrefix>/<other-device-id>/..., merges each into the local log with
mergeLogs, then pushes this device's own merged result back up.
Local persistence (LogStore)
The merge scheme above is in-memory only. For an app that needs to keep its
Log on disk and drive a live UI off it, this package also ships a
domain-agnostic LogStore:
import 'package:crdt_sync/crdt_sync.dart';
import 'package:crdt_sync/crdt_sync_io.dart'; // FileLogPersistence (dart:io)
final store = LogStore(
persistence: FileLogPersistence(File('$docsDir/notes.json')),
nodeId: 'phone',
);
await store.load();
// Reactive read: re-derive a domain view whenever anything changes. Filtering,
// sorting and search live here, in the app -- the store is field-agnostic.
final Stream<List<Record>> visible = store.changes.map(
(_) => store.values.where((r) => !r.deleted).toList(),
);
await store.upsert(Record(id: 'n1', fields: {'text': ('hi', store.nextHlc())}));
await store.delete('n1'); // sticky tombstone, not a hard remove
// After a sync tick, swap in the merged result:
await store.replaceAll(await syncLog(/* ... */, localLog: store.snapshot()));
LogStore is pure Dart (import it from the main barrel); the filesystem-backed
FileLogPersistence lives behind the separate crdt_sync_io.dart entrypoint so
the core stays web-safe. Querying is deliberately not in the library — a
Record's fields are an opaque map, so the app filters store.values itself.
The Python side mirrors the on-disk format via crdt_sync.dump_log/read_log
(load/save only; no reactive stream — that's a Dart/Flutter convenience).
Development
dart pub get
dart analyze
dart test