[cairo] bumping pixmap dependency

[Chris Wilson]

On Fri, 2009-06-05 at 12:06 -0400, Thomas Jaeger wrote:
> I don't think there's really a reliable way to tell whether the drivers
> are fixed, so in my current version of the patch (attached)

Nice work, pushed.

> As for the gradient patches, I'm not so sure anymore that David
> Reveman's approach was the right one:  Wouldn't it make more sense to
> handle gradients like solid fills and add a new function to
> cairo_surface_backend_t for backends that want to accelerate gradient fills.

Yes and no. The current system of fallbacks does feel quite limited and
complicated to extend. There is a growing desire to replace the internal
rendering pipeline with something more explicit and with greater control
for the backends, but nothing concrete has been proposed yet. (On the
other hand, the current system works and is easy to use from the
backend's perspective!)

I find that the function that I miss the most (and therefore end up
writing paint,fill,mask stubs) is a generic composite_region() that may
handle any of the source or mask types. One of the interesting things
about writing a new GL (or drm) backend is asking exactly these
questions: just how should Cairo be structured to achieve full h/w
acceleration and provide s/w fallbacks whenever necessary?
-ickle