[Cairo] Rectangle-based clipping for surfaces

[Keith Packard]

Around 21 o'clock on Oct 28, Carl Worth wrote:

>	pixman_region_t *
>	cairo_current_path_region (cairo_t *cr);

I'm curious how "tight" this region is guaranteed to fit around the actual 
path; if we're really going to compute all of the pixels which are 
affected by the operation, the region will be very complex.  Is this 
really a useful operation?  Or do we just want to stroke the path to an 
alpha-only destination and look at the resulting alpha values?

> I don't like the name cairo_stroke_path, so I have already proposed:

Let's leave the PS name in place; it's not horrible, and 
cairo_outline_path isn't enough of an improvement to justify being 
different.

-keith