(The thread has blossomed. I think I'm all caught up on reading.)

The Pygame on ctypes project has gotten me more excited about Pygame
than anything. Alex has done an amazing job not only at getting this put
together, but exploring various avenues of development that did and
didn't work out.

The crown jewel in the ctypes pygame code is the sdl layer. I'd like to
get this released soon, without the pygame stuff. It sounds like this
would be best handled as a completely separate project.

Coding for straight SDL in Python is still an ugly prospect, but this is
the level where I'd like to do some experimenting and development.

The "pygame compatibility layer" would be something separate from these
as well. Probably distributed as an egg in some form that won't
automatically get confused with "Pygame proper".

I think the results of ctypes-pygame will be a more agile and
lightweight library. It will easily mix with a variety of other
libraries for things like rendering and transforms. It will be easier
and faster to develop with, and more effective at trying new things.

There are a few hurdles left for ctypes. Primarily platform stuff like
ARM and Win64.