Core concepts

This section is the mental model behind diwire. If you prefer learning by doing, start with the Tutorial (runnable examples) tutorial and use these pages as the “why it works” reference.

Recommended order

1. Container - auto-wiring, what gets resolved, and how keys work

2. Registration - explicit registration, decorators, and interfaces

3. Lifetimes - transient vs singleton vs scoped

4. Scopes & cleanup - request-like scopes and deterministic cleanup

5. Function injection - injecting into functions with Injected()

6. Named components - multiple implementations via Component("name")

7. Open generics - open generic registrations and type-safe resolution

8. Async - async factories, async cleanup, and aresolve()

9. container_context - global context container for framework integration

10. Compilation - precomputing the graph for speed

11. Errors & debugging - common error modes and how to debug them

12. Integrations - dataclasses/pydantic/attrs/msgspec notes

• Container
  • Auto-wiring (default)
  • Strict mode (no auto-registration)
  • Service keys
  • Next
• Registration
  • Class / factory / instance
  • Decorator registration
  • Interface/protocol binding
  • Re-registering (overrides)
  • Next
• Lifetimes
  • Transient vs singleton
  • Scoped
  • Auto-registration lifetime
• Scopes & cleanup
  • Creating a scope
  • Scoped lifetime
  • Generator factories (deterministic cleanup)
  • Nested scopes
  • Imperative close
• Function injection
  • Basic injection with Injected()
  • Per-call scopes for request handlers
  • Decorator style
• Named components
  • Example
  • Notes
• Open generics
  • Example
  • How it works
• Async
  • Async factories + aresolve()
  • Async cleanup with async generators
  • Parallel resolution
• container_context
  • The idea
  • Basic usage
  • Scoped usage
  • Framework integration
• Compilation
  • When to use it
  • Basic usage
  • Runnable example
  • Notes and limitations
• Errors & debugging
  • Missing dependencies
  • Circular dependencies
  • Scope mismatch
  • Async in sync context
• Integrations
  • Tested integrations
  • pydantic-settings (BaseSettings)
  • pytest