Checked in the initial OpenGL 3.2 support, haven't done *any* checking on it yet.  So buyer beware on bzr trunk :) .  Really do need to get the pyglet-generator branch working some day.

Oh, also ran 2to3 across the package.  It finished (used to blow up on the larger files), so it may be possible some day to run on Python 3.x, though honestly my interest in the test rapidly approached 0 as I finished the run and considered the idea of integrating the changes.  Can I really afford to spread my already rather thin resources over supporting a platform that almost no-one is using yet (almost doubling the release workload)?

Seems like one of those "after we have a perfectly functional, perfectly *automated* test suite" things.