[SciPy-user] where is models.py?

Jarrod Millman millman@berkeley....
Tue Oct 21 02:28:55 CDT 2008


On Mon, Oct 20, 2008 at 11:19 PM, David Cournapeau
<david@ar.media.kyoto-u.ac.jp> wrote:
> IIRC, the main problem of the models package was its dependency on
> weave. I am not familiar with weave much, but I think the consensus was
> it was not adapted for scipy packages: it causes some build issues
> (since weave itself is in scipy, there is a bootstrap problem for once).
> So if you rewrote the part which depends on weave, it could be
> reincluded in scipy.

Chris Burns and Tom Waite rewrote the weave bit just before I decided
to pull it:
http://projects.scipy.org/scipy/scipy/changeset/4602
http://projects.scipy.org/scipy/scipy/changeset/4630
http://projects.scipy.org/scipy/scipy/changeset/4631
http://projects.scipy.org/scipy/scipy/changeset/4632

While they were removing the weave dependency, it became apparent that
the packages wasn't ready for release.

> But to be honest, I think we should not accept more code in scipy: scipy
> is already quite big, and there is not enough man power. More code means
> more maintenance problems. Would it be a big problem for you if the code
> was put in a scikit ?

I don't think anyone in my group is going to have time to make it into
a scikit.  Unless there is some serious interest and people willing to
work on it, I expect that we will work on the models code in the nipy
trunk going forward.

> In any case, I am strongly against any new code for scipy 0.7. We are
> already really late, there are tons of bugs, and not many people fixing
> them. Adding code just make it worse at this point.

+1.  Any new code or functionality should be worked on outside of
trunk at least until 0.7 is released.

-- 
Jarrod Millman
Computational Infrastructure for Research Labs
10 Giannini Hall, UC Berkeley
phone: 510.643.4014
http://cirl.berkeley.edu/


More information about the SciPy-user mailing list