[AstroPy] ANN: numarray-1.0 released

Perry Greenfield perry at stsci.edu
Mon Jul 12 15:01:56 CDT 2004



> -----Original Message-----
> From: owner-astropy at stsci.edu [mailto:owner-astropy at stsci.edu]On Behalf
> Of Stephen Walton

> What are the chances of numarray-1.0 being integrated with SciPy,
> replacing numeric?  My quick read of the release notes indicates that
> the first two of the issues in
> http://www.scipy.org/documentation/numarraydiscuss.html have been
> addressed in the new version of numarray.
> 
> -- 
> Stephen Walton <stephen.walton at csun.edu>
> Dept. of Physics & Astronomy, CSU Northridge

There are really two different issues that are raised by your
question:

1) numarray being integrated with scipy
2) Numeric being replaced with numarray

Doing 2 would presumably handle 1, but 1 can be done without
doing 2.

The single biggest sticking point to 2 is the small array
performance issue (the three others referred to by the url
above are not serious obstacles and should be dealt with
before long). Removing the remaining difference in performance
for small arrays is likely to be a lot of work and requires
resources we (STScI) don't have to deal with an issue that
isn't (currently anyway) a problem for us.

To the extent that there is a large enough
community of users unwilling to suffer the slower performance
for small arrays (at least to the extent that they are
willing to continue maintaining Numeric) then both Numeric
and numarray may coexist for a while. The current move
seems to be to unify their behavior so that they can be 
used interchangeably. 

That being said, we still are very interested in making
scipy work with numarray as well as Numeric. We were hoping
to do that in the past year. Our efforts along those lines
have been sidetracked by work on contributing to a plotting
package that will work with numarray (that being matplotlib)
which we are still treating as a higher priority than 
scipy functionality. I'm hoping that in the next few months
that some of the effort being devoted to that and other 
projects can be redirected to making scipy work with numarray.
A small part of work along those lines has been done,
and the next week or two is being set aside to complete
work with f2py and numarray (f2py is used by a large part
of scipy).

I see the scipy effort as being done in phases. It could take a 
while to see all of it ported. It would be useful to get feedback
about what is most important to astronomers, so feel free to 
let us know.

Perry


_________________________________________________
AstroPy mailing list     -      astropy at stsci.edu
http://www.astro.washington.edu/owen/AstroPy.html


More information about the AstroPy mailing list