[cairo] new pixman questions

[David Hill]

hey - a couple of questions related to the integration of the new pixman 
code. These came from my first, very hack and slash to get it to compile 
pass. I am sure I will have more questions as I get back to all the code 
I commented out.

1) in the new pixman repo, the source is in pixman/pixman, and in the 
cairo it is pixman/src. Anyone mind if I leave it pixman/src ?

2) why pixman_region_num_rects() -> pixman_region_n_rects() ? Is there a 
semantic difference I need to be worried about ?

3) in the new code, pixman_region_rects() exists but is not in pixman.h. 
Is this an oversight or on purpose ?

4) there does not seem to be a pixman_image_destroy() and I would think 
there needs to be one to free image internal data. Oversight ?

5) pixman_image_set_transform() does not return an error condition when 
it detects out of memory. I am guessing we should return a bool_t at a 
minimum, and possibly an error code ? After all the recent error 
handling additions, perhaps a few error codes and an int return make sense.

6) pixman_bits_t is gone. It was a typedef for uint32_t, which is used 
in a few places in the code. This is going to force a cast when dealing 
with RGB565 and other short data. Do we want to leave it as uint32_t or 
perhaps use a typedef to hide things, or perhaps even make it a void* so 
that no cast is needed on the public interface ?

thanks,
Dave (happily hip deep in code).