[SciPy-dev] Merged numarray branch

Todd Miller jmiller at stsci.edu
Mon Jan 17 07:14:57 CST 2005


On Sun, 2005-01-16 at 16:49 -0800, Stephen Walton wrote:
> Todd Miller wrote:
> 
> >As part of an effort to make scipy work with numarray,  I merged the
> >numarray branch of CVS onto the main trunk about an hour ago. 
> >
> I just downloaded a clean copy of the current CVS for SciPy, built and 
> installed it.  scipy.test(level=1) is generating a lot of failure 
> messages, mainly related to a missing scipy.linalg.eig module.  If I set 
> the environment variable NUMERIX to numarray, I can't import scipy at 
> all:  the traceback complains that _na_imports.py cannot import name isinf.

Quoting me:

There are still errors in the self-tests using numarray so this is
mostly a curiosity announcement and head's up to be on the lookout for
other problems.  If you're curious, get numarray from CVS on source
forge first.

So:

I didn't mean to imply that the numarray port of SciPy is finished,
merely that the port of scipy_base to numarray has been moved to the
head of CVS;  scipy_base.test() does AFIK complete w/o errors.
scipy.test() had ~70 errors the last time I looked and it will likely
take at least a few weeks to straighten them out.  Beyond that, major
scipy facilities remain to be ported such as weave and scipy_distutils.
Beyond that,  some of the scipy functions need to be "re-wrapped" using
the numarray UFunc API.

Regards,
Todd




More information about the Scipy-dev mailing list