[cairo] performance of cairo in a larger application

Carl Worth cworth at east.isi.edu
Wed Jul 28 07:09:57 PDT 2004


On Wed, 28 Jul 2004 08:04:33 +0200, Mathieu Lacage wrote:
> On Tue, 2004-07-27 at 16:15, Carl Worth wrote:
> > 1) Keith and I are involved in a rewrite of the tessellation and
> >    rasterization stages right now. We're doing this to make it correct
> >    and more robust, but it looks like the new version will also be a
> >    fair amount faster.
> 
> Where does this new code live ? Although I cannot promiss to contribute
> code, I would be definitely be interested in reading it and maybe
> commenting on it.

For tessellation, the algorithm is straight from a paper by John Hobby:

	http://cm.bell-labs.com/cm/cs/doc/93/2-27.ps.gz

        [Hobby93c] John D. Hobby, Practical Segment Intersection with
        Finite Precision Output, Computation Geometry Theory and
        Applications, 13(4), 1999.

The prototype we have so far is written in nickle, and is in the
src/tess module within the CVS tree at keithp.com. As we finish it and
port it to C it will appear in cairo of course.

Keith has a prototype of the new trapezoid rasterization as well. That
will go into fb/fbtrap.c in the X server eventually, (and a similar spot
in libpixman). I don't know if what he has so far is in CVS yet.

-Carl




More information about the cairo mailing list