[cairo] Glitz surface?

[Clemens Eisserer]

> [*] The 4x improvement might be almost entirely from avoiding the
> overhead of sending trapezoids to the X server, and then incrementally
> rasterizing them. An interesting experiment would be to change
> cairo-xlib to rasterize client-side and push the mask to the server
> before compositing, (which would also allow for doing spans-based
> rasterization of polygons). The original idea behind tessellation and
> trapezoid-rasterization was to take advantage of hardware-based
> trapezoid-rasterization, but since that isn't is sight now, (even after
> all this time), we should probably cut out all the unneeded overhead.

Does this means  there's no shader based rasterization in sight, not
even for the intel driver?

Thanks, Clemens