[IPython-dev] IPython inputhook, higher rate?

Thomas Kluyver takowl at gmail.com
Wed Oct 1 12:44:47 EDT 2014

Hi Martin,

On 1 October 2014 00:39, Martin Teichmann <lkb.teichmann at gmail.com> wrote:

> To me it's a bit weird to say that we shouldn't push that straight into
> core
> Python - IPython is working around this broken functionality since more
> than a decade, so how long do we have to wait until something is not
> considered "straight" anymore? Apparently IPython is not the only one
> having this problem, as you mentioned others have also already written
> a library to circumvent this problem in the standard library.

InputHook has been around for a long time, but only recently has there been
a standardised event loop interface in the stdlib that it could use
instead. And integrating input() with that is presumably not trivial, so it
makes sense to work out the kinks in a separate package, not in a Python
builtin function.

> As for compatibility: I actually think that the way IPython is handling
> the problem - mostly hiding all the ugly details behind a nice API -
> would make it possible to have the new and old version nicely in
> parallel, especially as the new version has already mostly been written,
> for the case if IPython is not running in the terminal.

It probably could, but we'll probably still have to maintain the old API
for quite some time; this is true whether it's a separate package depending
on asyncio or a new Python feature. The real value for us would be being
able to drop the inputhook stuff altogether, which seems unlikely for the
foreseeable future.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/ipython-dev/attachments/20141001/1c62b8f9/attachment.html>

More information about the IPython-dev mailing list