PyOpenGL 3.x has been in development for something on the order of 3.25 years, from the first tentative experiments in making a ctypes wrapper to something that's almost a drop-in replacement for PyOpenGL 2.x.  In that time I've gone through a significant number of code generator approaches and have altered the method I use for wrapping code manually a number of times.

Which is to say, I should probably go through the code and make sure we're using the same wrapping style throughout.  It might even be worth it to regenerate the core "raw" API, as it turns out the docstrings with just the arguments/return values aren't all that useful (thus just waste space).

If I'm going to do that, I probably should use the pyglet code generator (I'm already using it for the wgl, xgl and agl libraries) so that at the very least the tool-chain is the same, even if we're not sharing the same core library instance.