[SciPy-User] Contributing to SciPy was Re: Least-squares fittings with bounds: why is scipy not up to the task?

Skipper Seabold jsseabold@gmail....
Thu Mar 15 08:53:02 CDT 2012


On Wed, Mar 14, 2012 at 3:16 PM, Cameron Hayne
<cameron.hayne@dftmicrosystems.com> wrote:
>
> On 14-Mar-12, at 6:12 AM, Scott Sinclair wrote:
>
>> I still think that recipes which tell me how to do task X, with
>> package Y are better hosted in the documentation/online resources of
>> package Y.
>
>
> But that is only useful when you know that you need to use package Y.

I agree with this.

> The cookbook should answer the question "How do I do X ?" (without any
> reference to specific packages).

Here's my problem with this. What if the question is, the fairly
common, "how do I solve a non-linear programming problem?" The answer,
with numpy/scipy, is that really you don't (or you drop in the
entirety OpenOpt source code...). I don't think we should purge the
cookbook of examples like these. Don't get me wrong, I don't think we
should duplicate other package's documentation. But Dmitry
(presumably) took the time to write up these examples, and they can be
helpful in pointing someone towards topical software.

Comes down to SciPy vs. scipy I think. And maybe I'm saying I'd prefer
the former for the cookbook, and I think scikits, of which openopt was
formerly one, and closely related packages fall under SciPy.

This raises the question of what to do when the example becomes stale.
Well, for the future, maybe you could link your e-mail with the
posting of the original recipe and have a "Problem with this recipe"
button. I don't know.

That said, I'm also not doing any heavy lifting on this. Just my thoughts,

Skipper


More information about the SciPy-User mailing list