[Matrix-SIG] Numerical python on SourceForge?
Travis Oliphant
Oliphant.Travis@mayo.edu
Sat, 8 Jan 2000 19:44:33 -0600 (CST)
>
> - What parts? Is it important to be compatible with jpython or viper?
> If nD-arrays go into the core, become they part of the language?
My vote is to eliminate Python Numeric Types and make them all rank-0
arrays. This would possibly involve changing rank-0 arrays in Numeric so
that don't have so much overhead. At the very least, arrays and
ufuncs should become standard types and umath and multiarray in the
standard library.
>
> - I think all the higher level stuff (fft, lapack) should remain an
> external module.
I agree.
>
> - Can these modules be separated in a similar form to other modules. I
> mean, is it achievable to integrate e.g. the cephes module at the
> same level as Numeric? Or deintegrate e.g. fft on the other hand?
I don't understand what you mean. Are you talking about distributing
cephes in a "standard" package or tying it more closely to Numeric. All
of this depends on what goes in to Python itself and what stays a separate
package.
>
> - Should there be provisions made for the inclusion of new packages
> into the source tree? This inflicts with the progress of the
> distutils package.
I think a separate project could handle "new packages" and take the Python
core and extend it with all kinds of Math goodies.
>
> - Should the documentation be made compatible to the normal python
> module documentation? I would volunteer to reformat the
> documentation.
Yes, definitely if it is going to be part of the language.
>
> - Are our needs fulfilled by sourceforge? I mean a first step would
> be a public CVS (anonymous checkout, restricted checkin)
I'm not sure. Lot's of people seem to be using sourceforge and it sure
looks nice. Certainly a CVS server would be a start with multiple people
allowed to check changes in.
>
> - Will gist be also part of this ``outsourcing''?
Gist is very nice but should definitely be an add-on,
>
> - Can some parts be restructured? I think about the array_map from
> Travis and some code from the arrayfns module.
What do you mean restructured? Do you mean that array_map and other
functions from arrayfns should go into the Python core along with the
basic modules? If so then I think that would be useful.
>
> - Is it possible to include also some more IO stuff at the ``basic''
> module level?
I agree. If the NumPy type is in the Python core then we need something
like numpyiomodule to get data into arrays quickly.
>
> Stop!
Please don't, let's keep going and get the ball rolling again.
>
> Just as a side note, our perl counterpart pdl is already at
> sourceforge :-)