How the heck does async/await work in Python 3.5?
Being a core developer of Python has made me want to understand how the language generally works. I realize there will always be obscure corners where I don't know every intricate detail, but to be able to help with issues and the general design of Python I feel like I should try and understand its core semantics and how things work under the hood.
But until recently I didn't understand how async/await worked in Python 3.5. I knew that yield from in Python 3.3 combined with asyncio in Python 3.4 had led to this new syntax. But having not done a lot of networking stuff -- which asyncio is not limited to but does focus on -- had led to me not really paying much attention to all of this async/await stuff. I mean I knew that:
And I knew that asyncio was an event loop framework which allowed for asynchronous programming, and I knew what those words (basically) meant on their own. But having never dived into the async/await syntax to understand how all of this came together, I felt I didn't understand asynchronous programming in Python which bothered me. So I decided to take the time and try and figure out how the heck all of it worked. And since I have heard from various people that they too didn't understand how this new world of asynchronous programming worked, I decided to write this essay (yes, this post has taken so long in time and is so long in words that my wife has labeled it an essay).
Now because I wanted a properly understanding of how the syntax worked, this essay has some low-level technical detail about how CPython does things. It's totally okay if it's more detail than you want or that you don't fully understand it as I don't explain every nuance of CPython internals in order to keep this from turning into a book (e.g., if you don't know that code objects have flags, let alone what a code object is, it's okay and you don't need to care to get something from this essay). I have tried to provide a more accessible summary at the end of every section so that you can skim the details if they turn out to be more than you want to deal with.
