Structure de dossiers axée sur les fonctionnalités et monodépôts Melos
Organisez les fonctionnalités en paquets indépendants et gérez-les avec un monodépôt Melos.
Structure de dossiers axée sur les fonctionnalités et monodépôts Melos est une leçon Flutter Mobile Development gratuite sur CoddyKit. Ceci est la leçon 3 sur 4. Tu peux lire la leçon complète ci-dessous gratuitement — puis la pratiquer en direct dans le navigateur avec un éditeur de code intégré et un tuteur IA 24/7. Elle fait partie du parcours d'apprentissage Flutter Mobile Development, et ta progression se synchronise sur le web et l'application CoddyKit. Le cours Flutter Mobile Development comprend 4 leçons au total.
Certaines parties de cette leçon n'ont pas encore été traduites et s'affichent en anglais.
Why Feature-First?
As a Flutter app grows, the classic layer-first structure (top-level screens/, widgets/, models/, services/) starts to hurt. To touch one feature you jump across five folders, and unrelated teams collide in the same files.
Feature-first flips this: the top level is organized by business capability (auth, cart, profile), and each feature owns its own layers internally.
- High cohesion — everything for a feature lives together.
- Low coupling — features depend on contracts, not each other's internals.
- Scalability — a feature can later be extracted into its own package with almost no churn.
Anatomy of a Feature Folder
Inside lib/features/<feature>/ we mirror Clean Architecture's three layers. The domain layer is pure Dart with no Flutter or plugin imports, which is exactly what makes a feature easy to extract later.
A typical auth feature:
data/— DTOs, data sources, repository implementations.domain/— entities, repository interfaces, use cases.presentation/— widgets, pages, state (BLoC/Riverpod).
lib/
├── core/ // shared cross-cutting code
│ ├── error/
│ └── network/
└── features/
└── auth/
├── data/
│ ├── models/
│ ├── datasources/
│ └── repositories/
├── domain/
│ ├── entities/
│ ├── repositories/
│ └── usecases/
└── presentation/
├── bloc/
├── pages/
└── widgets/The Domain Layer Is Pure Dart
The key discipline that makes a feature portable: the domain layer imports nothing from Flutter or plugins. Entities are plain immutable Dart classes, and repository contracts are abstract interfaces.
Because this code has no framework dependencies, it is trivially unit-testable and can run on any Dart VM — including an online judge.
// domain/entities/user.dart
class User {
final String id;
final String email;
const User({required this.id, required this.email});
}
// domain/repositories/auth_repository.dart
abstract class AuthRepository {
Future<User> signIn(String email, String password);
}
void main() {
const user = User(id: '1', email: 'ada@example.com');
print('Loaded ${user.email}');
}Use Cases Encode Business Rules
A use case is a single-responsibility object that orchestrates one application action. It depends only on the repository interface, never on its implementation. This keeps the presentation layer dumb and the rules testable.
The common Flutter convention is to give each use case a call method so it can be invoked like a function.
// domain/usecases/sign_in.dart
class User {
final String email;
const User(this.email);
}
abstract class AuthRepository {
Future<User> signIn(String email, String password);
}
class SignIn {
final AuthRepository repository;
const SignIn(this.repository);
Future<User> call(String email, String password) {
if (!email.contains('@')) {
throw ArgumentError('Invalid email');
}
return repository.signIn(email, password);
}
}
class FakeRepo implements AuthRepository {
@override
Future<User> signIn(String email, String password) async => User(email);
}
void main() async {
final signIn = SignIn(FakeRepo());
final user = await signIn('ada@example.com', 'pw');
print('Signed in: ${user.email}');
}From Folders to Packages
Feature-first folders are a great first step, but folders don't enforce boundaries — any file can still import any other. The next level is to make each feature a real Dart package with its own pubspec.yaml.
Now the compiler enforces isolation: a package can only use what it explicitly declares as a dependency. This is where a monorepo comes in — many packages, one repository.
packages/feature_auth/packages/feature_cart/packages/core/apps/mobile/(the runnable Flutter app that wires features together)
What Is Melos?
Melos is a CLI tool for managing Dart/Flutter monorepos with multiple packages. It solves the pain of running pub get, tests, and version bumps across dozens of packages.
- Bootstrap — links local packages together and resolves all dependencies in one command.
- Scripting — define named scripts that run across every package.
- Versioning — conventional-commit based version bumps and changelogs.
You install it once globally and configure it per repository with a melos.yaml.
dart pub global activate melos
# from the monorepo root
melos bootstrap # or: melos bsConfiguring melos.yaml
At the repo root, melos.yaml declares the workspace name and where packages live (glob patterns). Modern Melos (3+) also supports declaring the workspace inside the root pubspec.yaml, but a dedicated file remains common.
The packages globs tell Melos which folders are managed members of the monorepo.
# melos.yaml
name: shop_workspace
packages:
- apps/**
- packages/**
command:
bootstrap:
usePubspecOverrides: trueWiring Local Path Dependencies
Inside the app, you depend on local feature packages by name. With usePubspecOverrides: true, Melos injects the local path: overrides automatically during bootstrap, so your committed pubspec.yaml can reference versions while local development still uses source.
The result: editing a file in feature_auth is instantly visible to the app — no publishing, no copy-paste.
# apps/mobile/pubspec.yaml
name: mobile
dependencies:
flutter:
sdk: flutter
feature_auth: ^1.0.0
feature_cart: ^1.0.0
core: ^1.0.0
# Melos generates pubspec_overrides.yaml on bootstrap:
# dependency_overrides:
# feature_auth:
# path: ../../packages/feature_authDefining Workspace Scripts
The biggest day-to-day win is scripts. You define a command once and run it across every package that matches a filter. This replaces brittle shell loops.
Common filters include --dir-exists=test (only packages that have tests) and --depends-on=core (only packages depending on a given package).
# melos.yaml
scripts:
analyze:
run: melos exec -- dart analyze .
description: Analyze every package.
test:
run: melos exec --dir-exists=test -- flutter test
description: Run tests only where a test/ dir exists.
# usage:
# melos run analyze
# melos run testKeeping Features Decoupled
Packages enforce boundaries, but you still must design the contracts. The rule of thumb:
- Features never import each other directly.
- Shared contracts and primitives live in a low-level
corepackage that everything may depend on. - Cross-feature navigation goes through an abstraction (a router or an event bus) injected at the
applevel.
This dependency direction — features → core, app → features — keeps the graph acyclic. A cycle between two feature packages will fail to resolve, which is a useful early warning.
// packages/core/lib/src/app_route.dart (shared contract)
abstract class AppRouter {
void goToCart();
void goToProfile();
}
// feature_auth depends on `core`, never on feature_cart.
class LoginController {
final AppRouter router;
const LoginController(this.router);
void onLoginSuccess() => router.goToCart();
}
void main() => print('Auth depends on core, not on cart.');A Typical Developer Workflow
Putting it together, the daily loop in a Melos monorepo is short and uniform regardless of how many packages you have:
melos bootstrapafter pulling new changes to relink packages.melos run analyzeandmelos run testin CI and locally.melos versionto bump changed packages from conventional commits.
Because the app is the only Flutter-runnable target and features are plain packages, your CI can test domain logic on the pure Dart VM (fast) and only spin up Flutter for presentation tests.
# clone + first run
melos bootstrap
melos run analyze
melos run test
# ship a release
melos version # bumps + changelogs from commits
cd apps/mobile && flutter build apkQuick Check
You have a Melos monorepo with feature_auth and feature_cart packages. After login, auth needs to navigate the user to the cart screen. What is the cleanest way to keep the dependency graph acyclic?
Recap
You learned how to scale a Flutter codebase from folders to packages:
- Feature-first structure groups code by business capability, each feature mirroring data/domain/presentation layers.
- The domain layer stays pure Dart, making it portable and easy to extract.
- Promoting features to Dart packages lets the compiler enforce boundaries that folders cannot.
- Melos manages the resulting monorepo:
bootstraplinks packages,scriptsrun commands across all of them, andversionautomates releases. - Keep the graph acyclic: features depend on
core, the app depends on features, and cross-feature concerns go through shared contracts.
Questions Fréquemment Posées
La leçon « Structure de dossiers axée sur les fonctionnalités et monodépôts Melos » est-elle gratuite ?
Oui — le texte complet de « Structure de dossiers axée sur les fonctionnalités et monodépôts Melos » est gratuit à lire ici sur le web. Pour la pratiquer de manière interactive (un éditeur de code intégré et un tuteur IA 24/7) et déverrouiller le reste du cours Flutter Mobile Development, passe à CoddyKit PRO. Le cours Flutter Mobile Development comprend 4 leçons au total.
Qu'est-ce que j'apprendrai dans « Structure de dossiers axée sur les fonctionnalités et monodépôts Melos » ?
Organisez les fonctionnalités en paquets indépendants et gérez-les avec un monodépôt Melos. Tu pratiques Flutter Mobile Development avec du code pratique que tu exécutes directement dans le navigateur, et un tuteur IA 24/7 répond à tes questions au fur et à mesure que tu avances dans la leçon.
Dois-je avoir de l'expérience pour commencer Flutter Mobile Development ?
Aucune expérience préalable n'est requise. Flutter Mobile Development sur CoddyKit est structuré pour les débutants jusqu'aux apprenants avancés, donc tu peux commencer ici ou depuis le début et avancer à ton rythme. Ceci est la leçon 3 sur 4.
Combien de temps prend la leçon « Structure de dossiers axée sur les fonctionnalités et monodépôts Melos » ?
La plupart des leçons CoddyKit prennent environ 5–10 minutes. Chacune est courte et interactive, tu progresses régulièrement et tu repiques exactement où tu t'es arrêté sur le web et l'app.
Peux-tu écrire et exécuter du code dans cette leçon Flutter Mobile Development ?
Oui. Chaque leçon Flutter Mobile Development inclut un éditeur de code intégré, tu écris et exécutes du vrai code directement dans ton navigateur et tu reçois des retours IA instantanés — aucune configuration locale requise.
Toutes les leçons de ce cours
- Limites entre les couches domaine, données et présentation
- Injection de dépendances avec get_it et injectable
- Structure de dossiers axée sur les fonctionnalités et monodépôts Melos
- Either, types d’échec et gestion fonctionnelle des erreurs