[IPython-dev] Feature and scope creep in IPython

[Thomas Kluyver]

On 15 January 2013 21:48, Nicholas Bollweg <nick.bollweg at gmail.com> wrote:

> This is a very good point, but since i'm on a kick: what is the means for
> determining quality? For the kernel? For the notebook? For the various
> frontends? Does every committer weigh in on every decision? What is the
> role of non-committers? "Plebiscite" style decision making would probably
> be crippling. If a project can start determining what these heuristics are,
> maybe the data from its various feeds can help.
>

At least one core developer has to agree with it, to do the merge.
Generally, the larger the change, the more people we'll want to confirm it
- but nothing requires the input of the whole core team. We also have a
rough guideline that if core contributors disagree about something, we're
probably missing something important, and we should spend more time
thinking about it.

We do, of course, listen to non-committers, but the core developers
sometimes decide that a particular feature would be best developed outside
of IPython. There are various technical mechanisms to allow separate pieces
of code to work with IPython. I think this is a key part of what Brian was
talking about - we can't and shouldn't maintain code for everything anyone
wants to do in IPython.


> I liked some of the discussion on your blog post (I think) by the SymPy
> guys: their users "only use 20%" of SymPy... but every user has a different
> 20%! What is the broader community's usage of IPython like? Has this been
> captured before?
>

Yes, I have some concerns around this as well - I suspect that many of our
features are useful to a large minority of our users, and I don't think
that's necessarily a problem. Where the user-threshold should be, I'm not
sure.

We did do a simple survey in 2011
(http://ipython.org/usersurvey2011.html). We hope to repeat it this
year with a bit more detail.

Best wishes,
Thomas
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/ipython-dev/attachments/20130116/53120d93/attachment.html>