Using Sheppy without asyncio¶

SyncQueue wraps Queue for codebases that don't use asyncio. It mirrors the queue API with plain blocking calls.

from sheppy import task
from sheppy.experimental import SyncQueue


@task
def add(x: int, y: int) -> int:
    return x + y


def main() -> None:
    q = SyncQueue("memory://")  # use "redis://localhost:6379" in production

    t = add(2, 3)
    q.add(t)

    done = q.wait_for(t)
    print(done.result)

    q.close()


if __name__ == "__main__":
    main()

$ python sync_example.py
5

How it works¶

SyncQueue runs one asyncio event loop in a background daemon thread and submits every call to it, blocking the calling thread until the result is ready. One instance can be shared between threads. Call close() when done to stop the loop.

Two differences from Queue:

• The backend can only be given as a URL string (or the SHEPPY_BACKEND_URL environment variable), not as a Backend instance
• Long waits (like wait_for) block the calling thread, not the event loop

Workers are unchanged. sheppy work from the command line works the same regardless.