[PATCH weston 2/7] pixman-renderer: Move shadow buffer into renderer

[Bill Spitzak]

Ander Conselvan de Oliveira wrote:

> If I understand 
> correctly, pixman doesn't consider the transform of the destination 
> image when compositing, so we also need a shadow buffer to support 
> transformed outputs.

I don't understand this. The caller of pixman can "consider the 
transform of the destination image when compositing", there is no need 
for pixman to do this (all pixman could do is divide two transform 
matrices and produce the same result as the caller doing it).

I would guess the real problem is that telling pixman to compose with no 
transform followed by a hand-written copy of the shadow buffer to the 
output that applies a 90 degree rotation is faster than telling pixman 
to apply that rotation while composing. But there is no reason for 
pixman to "consider the transform of the destination" because the caller 
can do so.