Deployment¶

Call setup() once per process after all provider modules have been imported. In multi-worker deployments, each worker process must wire its own registry.

Plain scripts and workers¶

import asyncio

from wepositive_di import setup


async def main() -> None:
    ...


if __name__ == "__main__":
    setup()
    asyncio.run(main())

ASGI applications¶

Use your framework lifespan hook.

from collections.abc import AsyncIterator
from contextlib import asynccontextmanager

from fastapi import FastAPI

from wepositive_di import registry, setup


@asynccontextmanager
async def lifespan(app: FastAPI) -> AsyncIterator[None]:
    setup()
    yield
    await registry.shutdown_resources()


app = FastAPI(lifespan=lifespan)

Each Uvicorn or Gunicorn worker process runs the lifespan independently.

WSGI applications¶

For Flask or other sync frameworks, call setup() at module import time or from a worker startup hook.

from flask import Flask

from wepositive_di import setup


setup()
app = Flask(__name__)

If sync request handlers depend on async providers, they can be resolved only when no event loop is already running in that thread.

Shutdown¶

Plain singleton providers do not need DI shutdown. Singleton context manager providers do.

from wepositive_di import registry


registry.shutdown_resources()        # sync singleton resources
await registry.shutdown_resources()  # async singleton resources

Match the shutdown call to the kind of resources you registered.

Context storage in workers¶

The default InMemoryContextStorage is per process. In a multi-process deployment, workers do not share in-memory context. Override context_storage_singleton with a distributed storage implementation when contexts must be shared across workers.