> The problem is the complexity of writing a replacement and all the > compatibility layers needed to make it usable in the productive world.
The "window to a framebuffer" thing appeals to me because it would be simple, performant, easy to implement, easy to code for, easy to port libs and apps to use it. I don't need partly obscured windows or enforced clipping. Also don't need every bogus app under the sun to work on it right away.
