async-python-patterns

Use asyncio.gather() to run multiple coroutines concurrently by passing them as arguments and awaiting the combined result. Alternatively, use asyncio.create_task() to schedule tasks on the event loop and await them individually when needed.

Use asyncio.Semaphore to limit concurrency by wrapping operations in an async with semaphore block. Set the semaphore to your desired maximum, such as Semaphore(5), to allow only that many simultaneous operations like API calls.

Use asyncio for I/O-bound work like network requests and database queries where tasks wait on external resources. Use multiprocessing for CPU-bound computation, since asyncio runs single-threaded and cannot parallelize heavy calculations.

Blocking calls like time.sleep() or requests.get() freeze the entire event loop and stall all concurrent tasks. Wrap them with asyncio.to_thread() in Python 3.9+ or loop.run_in_executor() to run them in a thread pool instead.

This happens when you forget the await keyword, so the coroutine is created but never executed. Always call async functions with await inside another async function, or use asyncio.run() from synchronous entry points.

Use the pytest-asyncio plugin and mark test functions with @pytest.mark.asyncio so they run on an event loop. You can then await coroutines directly in tests and assert on results, including timeout behavior with asyncio.wait_for().