[Python-Dev] PEP 492: async/await in Python; version 5 (original) (raw)

Guido van Rossum guido at python.org
Tue May 5 23:25:01 CEST 2015

On Tue, May 5, 2015 at 2:01 PM, Paul Moore <p.f.moore at gmail.com> wrote:

On 5 May 2015 at 21:00, Yury Selivanov <yselivanov.ml at gmail.com> wrote: > On 2015-05-05 3:40 PM, Jim J. Jewett wrote: >> >> On Tue May 5 18:29:44 CEST 2015, Yury Selivanov posted an updated PEP492. >> >> Where are the following over-simplifications wrong? >> >> (1) The PEP is intended for use (almost exclusively) with >> asychronous IO and a scheduler such as the asynchio event loop. > > Yes. You can also use it for UI loops. Basically, anything > that can call your code asynchronously.

Given that the stdlib doesn't provide an example of such a UI loop, what would a 3rd party module need to implement to provide such a thing? Can any of the non-IO related parts of asyncio be reused for the purpose, or must the 3rd party module implement everything from scratch? To me, this is an important question, as it cuts directly to the heart of the impression people have that coroutines and async are "only for asyncio". I'd be interested in writing, for instructional purposes, a toy but complete event loop. But I'm not really interested in trying to reverse engineer the required interface.

This is a great idea. What kind of application do you have in mind?

I think the main real-life use case for using coroutines with a UI event loop is newer Windows code. C# (and IIUC VB) has coroutines very much along the lines of PEP 492, and all code that does any kind of I/O (whether disk or network) must be written as a coroutine. This requirement is enforced by the C# compiler: the basic system calls for doing I/O are coroutines, and in order to get their result you must use an await expression, which in turn may only be used in a coroutine. Thus all code that may invoke an I/O call ends up being a coroutine. This is exactly the type of constraint we're trying to introduce into Python with PEP 492 (except of course we're not making all I/O primitives coroutines -- that would be madness, we're going with optional instead).

-- --Guido van Rossum (python.org/~guido) -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://mail.python.org/pipermail/python-dev/attachments/20150505/f3cb5600/attachment.html>

More information about the Python-Dev mailing list