Guido van Rossum

concurrently, especially things that had to do with IO and networking IO was especially a sort of a popular topic.

And in the Python standard library, we had a brief period where there was lots of development.

And I think it was late 90s, maybe early 2000s.

And like,

Two little modules were added that were the state of the art of doing asynchronous I.O.

or sort of non-blocking I.O., which means that you can keep multiple network connections open and sort of service them all in parallel like a typical web server does.

Also possible.

Yeah, like when you're writing a web server, when a request comes in, a user sort of needs to see a particular web page.

You have to find that page maybe in the database and format it properly and send it back to the client.

And there's a lot of waiting, waiting for the database, waiting for the network.

And so you can handle hundreds or thousands or millions of requests automatically.

concurrently on one machine.

Anyway, ways of doing that in Python were kind of stagnated and, uh, I forget, it might've been around 2012, 2014, uh,

when someone for the umpteenth time actually said, these async chat and async core modules that you have in a standard library are not quite enough to solve my particular problem.

Can we add one tiny little feature?

And everybody said, no, that stuff is not – you're not supposed to use that stuff.

Write your own using a third-party library.

And then everybody started a debate about what the right third-party library was.

And somehow I felt that there was actually a cue for, well, maybe we need a better state-of-the-art library

module in the standard library for multiplexing input-output from different sources.