Re: Re: Sub-pixel area computations

Around 17 o'clock on May 5, Allen Akin wrote:

> Sounds good.

One other thing -- I want to make the precise specification fast enough 
that no other software implementation is needed for performance reasons.

Because the current design is moving towards a prescriptive spec, it would
be nice to hear of alternative pixelization algorithms which might be
faster than what we've got so far -- there's every reason to look at
alternatives now before the spec is nailed down, alternatively, an
algorithm which is similar in performance but easier to understand would
also be welcome.  There will be at least two implementations of any such
spec, the one in the server and also one on the client side in an emulation
library; it would be nice to keep them both simple.

Keith Packard        XFree86 Core Team        Compaq Cambridge Research Lab