Simple file-based non-volatile queue library for flutter.

Typical use-case scenario is for small to medium sized in-device buffers storing persistent yet temporary mission-critical data, until it can be efficiently and safely consumed / delivered permanently - such as for custom analytics and specialized logging applications.


Add dependency to pubspec.yaml:

  flutter_persistent_queue: ^3.0.0

Run in your terminal:

flutter packages get


import 'package:flutter_persistent_queue/flutter_persistent_queue.dart';

// instantiate a new queue `pq`, with auto-flush set at 10 items
final pq = PersistentQueue('pq', flushAt: 10, onFlush: (list) async {

// push 25 Map<<String, dynamic>> items (await is optional)
for (int i = 0; i < 25; ++i) {
  pq.push(<String, dynamic>{'i': i});

// trigger a final manual flush for the remaining items
pq.flush((list) async => print('manual-flush\n$list'));

// check if all items have been flushed
assert(await pq.futureLength == 0);

// deallocate resources, if needed
await pq.destroy();


For integration / performance tests:

git clone
cd ~/flutter_persistent_queue/test
flutter drive --target=test_driver/app.dart

Building documentation files

rm -rf doc
dartdoc --exclude 'dart:async,dart:collection,dart:convert,dart:core,dart:developer,dart:io,dart:isolate,dart:math,dart:typed_data,dart:ui'

How it works

Each JSON-encodable item when enqueued goes to its own indexed key in a single non-volatile file (one file per uniquely named queue) to an unspecified location on the permanent storage of a flutter-compatible device.

This particular design choice limits potential use cases requiring very long queues but otherwise provides good performance within a limited resource footprint, as it doesn't require serializing and deserializing contiguous or chunked dart:collections to the filesystem.

Every push() call triggers an asynchronous and isolated write to the permanent storage. If the caller awaits until its completion, it is notified via Completers and afterwards there's no possibility of data loss from the application perspective.

After user-defined flush() operations (either manual or auto-triggered) the queue is fully emptied if and only if the user-defined onFlush does not throw any errors.

All PersistentQueue methods are consumed sequentially (according to the strict chronological order of its calls) one at a time on a dedicated producer / consumer event loop.

Performance of persisted queues on standard simulated devices must be of at least 500 writes/second. This constraint is verified on integration tests.

All filesystem and JSON operations are abstracted by the minimalistic flutter_localstorage library.




A simple file-based non-volatile persistent queue library for flutter. [...]