Python: async programming

Last time, we saw you can leverage ThreadPool for multiprocessing:

Python: meet ThreadPool

jmau111⚡⚡⚡ ・ Jun 9

#python #programming #performance #beginners

You may also appreciate a convenient async/await syntax to execute tasks asynchronously, and that's exactly what asyncio does.

Async IO?

Asyncio allows writing concurrent code.

You will often see it in professional usages.

However, you might not know that Asynchronous IO uses a single thread in a single process.

It's a bit counterintuitive, and you may find many usages for "concurrent code," but tasks are not inherently concurrent.

It's another approach that is different from threading or multiprocessing, by design.

Async programming is not easy

The bad practice would be to use async/await for anything and everything.

To me, you should focus more on the asynchronous vs. synchronous code than "concurrency."

With async/async with, you will manipulate coroutines and use the await to suspend their execution until you get something you are waiting for.

That's pretty much what "coroutine" means: a function that can pause its execution for other operations.

To define a coroutine, just use async:

import asyncio

async def test():
    print("test")

if __name__ == '__main__':
    asyncio.run(test())

To pause the execution, use await:

import asyncio

async def test():
    await asyncio.sleep(7)
    return "test"

async def main():
    t = await test()
    print(t)

if __name__ == '__main__':
    asyncio.run(main())

You can see await as break points.

Pros and cons

In my experience, the benefits of Async IO can be multiple:

• non-blocking calls, especially with HTTP requests
• optimized CPU usage
• awaitable objects
• ability to chain coroutines

However, there are some cons:

• can be harder to debug
• it's easy to misuse it (e.g., using async along with blocking calls)