Ferry Cache-Based Pagination

Cache-Based Pagination

Use Ferry’s cache to serve already-fetched entities instantly. This replaces the StreamGroup approach.

Ferry Cache-Based Pagination

Why Not StreamGroup?

The StreamGroup approach adds complexity: multiple streams, manual page tracking, merging logic, race conditions. Ferry’s cache-based approach is simpler.

Implementation

Notifier with Cache-Based Pagination

@riverpod
class AttendeeListNotifier extends _$AttendeeListNotifier {
  static const int _pageSize = 15;
  int _currentTake = _pageSize;

  @override
  Stream<List<AttendeeModel>> build() {
    _currentTake = _pageSize;
    final repository = ref.watch(attendeeRepositoryProvider);
    return repository.queryList(
      skip: 0,
      take: _currentTake,
    );
  }

  void loadMore() {
    _currentTake += _pageSize;
    ref.invalidateSelf();
  }

  void filter({String? fullName}) {
    _currentTake = _pageSize;
    ref.invalidateSelf();
  }
}

How Ferry Cache Makes This Work

1. First load: take: 15 fetches 15 items, caches them
2. Load more: take: 30 - Ferry serves first 15 from cache, fetches next 15 from network
3. Filter change: Fresh request with reset take

Each entity is cached by __typename:id, so Ferry reassembles lists efficiently.