[cairo] glitz repo access and fixes

[Chris Wilson]

On Mon, 14 Jun 2010 03:39:30 -0700 (PDT), karx11erx <karx11erx at googlemail.com> wrote:
> Q: glitz doesn't seem to be maintained well. I have read about problems with
> floating and fixed point formats and gradient handling as well as
> interfacing cairo among others. Would someone knowledgeable inform me about
> current problems with glitz, particularly when using it with the latest
> cairo version?

As far as cairo is concerned cairo-gl has superseded cairo-glitz [and the
glitz backend removed]. The fundamental issue was that glitz never lived
up to its potential performance and even the initial implementation of the
gl backend plugging into the high level api [rather than implementing
RENDER on top of GL as glitz tried] was an order of magnitude faster.
Since then, both the implementation within cairo has been improved and so
has mesa, as cairo-gl has proven to be an exceptional stress test. Given
the artificial limitation on glitz has been removed by reimplementing a gl
backend within cairo, that also opens up the possibility of handling paths
more intelligently and for requirements of direct rendering to drive the
design of Cairo much harder.
 
> Background is that I am currently evaluating mapnik + cairo + glitz on
> Windows 7 with the target of making it fully work on that platform.

Hmm, my experience with mapnik is that it is used to generate very large
images, on the order of 12k x 12k (and larger). In order to use the GPU
for such images, one needs to tile and blit the collage of subimages back
into the whole. Map rendering is a subject close to many hearts on the
Cairo list, if you find an area we can improve, or just have some pretty
pictures!, please do let us know.

I hope you have fun using Cairo!
-ickle

-- 
Chris Wilson, Intel Open Source Technology Centre