[SciPy-User] Central File Exchange for SciPy

John Hunter jdh2358@gmail....
Mon Nov 1 19:34:56 CDT 2010


On Mon, Nov 1, 2010 at 7:29 PM, william ratcliff
<william.ratcliff@gmail.com> wrote:
> I think for now, let's try going with BSD or CC0 and allowing people to link
> to other code if they so desire (but not put it on the site).   Another
> question:
> For usability, I really like how the stack overflow answers with the most
> votes appear at the top of the page.  However, over time, the previous top
> answer may become less relevant.  Should we have "aging" for scores?   Also,
> a friend suggested requiring doctests for the code (with the eventual long
> term goal of being able to run the doctests on a vm somewhere like EC2, but
> that would be a much longer term goal with the latest versions of scipy,
> numpy, and matplotlib).    Does it sound reasonable to require doctests or
> is that too high of a burden?

Way too high.  I may have some useful code laying around I would
upload if it were easy.  I will certainly not do it if I have to
retroactively go add tests.  I think we need the minimum barrier to
entry.  Let the description, reviews, rankings and the end user decide
if the code is suitable for a purpose.   With something like this the
key to use will be to get people to actually upload something.  Once
you have critical mass, and if you have a problem with too many low
quality submissions, consider tightening the standards then.

JDH


More information about the SciPy-User mailing list