[SciPy-user] New scipy release before 8/20?

David Cournapeau david@ar.media.kyoto-u.ac...
Wed Aug 8 01:30:09 CDT 2007


Matthieu Brucher wrote:
>
>     The friendly guys at Enthought go to the trouble already, so why not
>     let your students use
>
>     http://code.enthought.com/enstaller
>
>     Enthough currently allows you to mirror their eggs, even, so you can
>     download them on behalf of your students and create a local
>     repository.
>
>     See, for example, the instruction at the bottom of
>
>     http://dip.sun.ac.za/courses/ComputerVision/
>
>
> It seems that these eggs are too old for him and I cannot blame him 
> for this.
> I think there are some bugs in weave that must be fixed before a 
> release, but I really don't know anything about it to help solve them :(
>
IMHO, the most problematic thing is numpy. I think that a numpy release 
should be done as soon as possible, since the current one does not work 
with scipy. For example, if I take a look at the 1.0.4 milestone, I see 
no big bugs except a few 64 bits things which should not be too 
difficult to handle. Now that my midterm PhD evalution is done, I can 
spend a bit more time to fix those bugs :) I think I also found a way to 
compile an ATLAS without SSE on my machine, too, which means that 
releasing binaries for numpy should be doable soon.

For scipy,there are not many big bugs anymore I think, most serious 
problems (unbuildable on mac os X, memory corruption, bugs in basic 
functionalities) were handled. *I* think that the current trunk can be 
tagged as 0.5.3 as soon as numpy is released, basically, but maybe some 
other developers are working on other things which I didn't see on svn logs.

cheers,

David



More information about the SciPy-user mailing list