Using Asyncio in Python

Chapter 2. The Truth About Threads

Let’s be frank for a moment—you really don’t want to use Curio. All things equal, you should probably be programming with threads. Yes, threads. THOSE threads. Seriously. I’m not kidding.

Dave Beazley, “Developing with Curio”

If you’ve never heard of threads before, here’s a basic description: threads are a feature provided by an operating system (OS), made available to software developers so that they may indicate to the OS which parts of their program may be run in parallel. The OS decides how to share CPU resources with each of the parts, much as the OS decides to share CPU resources with all the other different programs (processes) running at the same time.

Since you’re reading an Asyncio book, this must be the part where I tell you, “Threads are terrible, and you should never use them,” right? Unfortunately, the situation is not so simple. We need to weigh the benefits and risks of using threads, just like with any technology choice.

This book is not supposed to be about threads at all. But there are two problems here: Asyncio is offered as an alternative to threading, so it’s hard to understand the value proposition without some comparison; and even when using Asyncio, you will still likely have to deal with threads and processes, so you need to know something about threading.

Warning

The context of this discussion is exclusively concurrency in network programming applications. Preemptive multithreading is also used in other domains, ...