On Sun, Jun 1, 2008 at 8:27 PM, Brian Jorgensen <[EMAIL PROTECTED]> wrote: > Hi, > > It's been a while, but I'm the one who committed the original sin of > including pyglet. In general, I agree with you. The intent back then was to > allow users to plot out-of-the-box, without any external dependencies. I've > since come to believe that our users are generally python programmers who > can handle the installation of dependencies. > > On the other hand, I want to avoid the situation that used to happen with > PyOpenGL on windows: you had to install specific, non-standard versions of > numeric, PIL, etc (see > http://www.visionegg.org/install-windows-details.html). Things to think > about: > > * How would this work for people using the windows installer? Can it handle > dependencies somehow, or should pyglet be a separate download? > > * Plotting hasn't been updated to use the latest versions of pyglet (we > depend on < 1.0, I think). If someone uses a more recent pyglet, how do we > know it's going to work? We might end up getting a lot of bug reports > relating to version mismatch. > > Regrettably, I haven't had enough time to contribute regularly, but I've > been toying here and there with different ways of doing plotting. I've been > thinking it might be a good idea to make a separate package sympy-plotting, > sympy-extras, or similiar. We could offer two windows installers, one with > plotting and one without, though it would create quite a bit of extra work > for each release. >
Thanks for starting this discussion and sharing the ideas. My own thoughts: * As to mpmath, it contains essential things that should imho be part of sympy, also it's just 8 files, no nested directories, so we just plain copy them to sympy and that's it. That's imho the best solution here and this is how it is done now. So I think mpmath is a non-issue. * pyglet: I think we could distribute sympy-pure.tar.gz (without pyglet) and sympy.tar.gz (with pyglet). Unfortunately this means more work for the release manager, as now he needs to test two tarballs, but if it's worthy, let's do it. The only real problem here is the sys.path hack, but that is imho only needed in python2.4, so this should soon become non-issue anyway, as I think in python2.5 it coud be fixed by relative imports and we don't need any sys.path things anymore. An argument could be the size of pyglet --- is there a problem with it? For me the size is ok. > So, is sympy an everything-but-the-kitchen-sink, batteries-included library, > or a svelte library with lots of optional add-ons? Both. It should be easy enough to get the job done. Imho it needs to be judged on case by case basis, so for mpmath and pyglet, see my thoughts above. Ondrej --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "sympy" group. To post to this group, send email to sympy@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sympy?hl=en -~----------~----~----~----~------~----~------~--~---
