[cairo] Solid pattern caching revisited

Chris Wilson chris at chris-wilson.co.uk
Thu Apr 19 06:26:53 PDT 2007

The issue that I've just been bitten by is that attributes are
stored directly on the surface or implicitly in the graphics state. The
notable one, as it is the first failure in cairo-test, is the clip mask,
but all of cairo_surface_attributes_t are at issue, as well as any
backend specific properties.

It seems that the best approach would be to use a COW surface.
Implementing such a beast would be quite invasive, so I'm open to
alternatives or further use cases that would benefit from COW surfaces.
Chris Wilson

More information about the cairo mailing list