Handling error recovery in dependent providers when the consumer only knows the parent provider

[saibotma]

When a parent provider (AProvider) depends on a child provider (BProvider), any error in BProvider propagates to AProvider. However, consumers of AProvider have no direct way to refresh/invalidate BProvider since they only interact with AProvider.

For example:

• BProvider fails due to a network error.
• AProvider (which depends on BProvider) also enters an error state.
• The UI (consumer of AProvider) shows an error and a "Retry" button that triggers ref.refresh(aProvider).

Current Behavior:
Refreshing AProvider re-executes its logic but reuses the existing (failed) state of BProvider, causing AProvider to fail again immediately.

Example code

import 'package:flutter/material.dart';
import 'package:flutter_riverpod/flutter_riverpod.dart';
import 'package:riverpod_annotation/riverpod_annotation.dart';

part 'main.g.dart';

@riverpod
Future<String> b(BRef ref) async {
  print("refreshing b");
  throw Exception("Failed because of reason");
}

@riverpod
Future<String> a(ARef ref) async {
  print("refreshing a");
  final bValue = await ref.watch(bProvider.future);
  return "a + $bValue";
}

void main() {
  runApp(
    ProviderScope(
      child: MyApp(),
    ),
  );
}

class MyApp extends ConsumerWidget {
  @override
  Widget build(BuildContext context, WidgetRef ref) {
    final aValue = ref.watch(aProvider);

    return MaterialApp(
      home: Scaffold(
        body: Center(
          child: Text(
            "${aValue.valueOrNull} ${aValue.hasError}",
          ),
        ),
        floatingActionButton: FloatingActionButton(
          onPressed: () => ref.refresh(aProvider),
          child: const Icon(Icons.refresh),
        ),
      ),
    );
  }
}

Consumers of a parent provider should be able to recover from errors in child providers without explicit knowledge of the dependency chain. A simple ref.refresh(parentProvider) should propagate retries to all relevant dependencies.

One solution i came up is making AProvider a notifier with a member function retry which invalidates BProvider.

Example code

import 'package:flutter/material.dart';
import 'package:flutter_riverpod/flutter_riverpod.dart';
import 'package:riverpod_annotation/riverpod_annotation.dart';

part 'main.g.dart';

@riverpod
Future<String> b(BRef ref) async {
  print("refreshing b");
  throw Exception("Failed because of reason");
}

@riverpod
class A extends _$A {
  @override
  Future<String> build() async {
    print("refreshing a");
    final bValue = await ref.watch(bProvider.future);
    return "a + $bValue";
  }

  void retry() {
    ref.invalidate(bProvider);
  }
}

void main() {
  runApp(
    ProviderScope(
      child: MyApp(),
    ),
  );
}

class MyApp extends ConsumerWidget {
  @override
  Widget build(BuildContext context, WidgetRef ref) {
    final aValue = ref.watch(aProvider);

    return MaterialApp(
      home: Scaffold(
        body: Center(
          child: Text(
            "${aValue.valueOrNull} ${aValue.hasError}",
          ),
        ),
        floatingActionButton: FloatingActionButton(
          onPressed: () => ref.read(aProvider.notifier).retry(),
          child: const Icon(Icons.refresh),
        ),
      ),
    );
  }
}

Is there a recommended pattern or planned feature in Riverpod to handle such scenarios more elegantly? A built-in way to cascade retries/invalidations would simplify error recovery in dependent providers significantly.

[gdurandrexel]

I think ref.invalidate(providerA) can do the trick if providerB is AutoDispose and is watched only by providerA (it will be destroyed because no more watched, and then rebuilt)

[saibotma]

This is not the case, unfortunately. See example in question.

[rrousselGit]

One solution i came up is making AProvider a notifier with a member function retry which invalidates BProvider.

That's fine