Re: Tesselating complex figures with Render

Around 12 o'clock on May 11, Carl Worth wrote:

> From an interface point of view, I think we definitely want the
> pre-adding as is done in the case of the text primitives.

It's optional in the case of text primitives, I'll make it optional in 
this case as well.

> So, if you can implement it efficiently, I don't see any compelling
> reason not to take this approach.

I think I'll do this by adding an optional field to the picture structure
inside the server which marks the non-zero extents per scanline within the
alpha channel. That will make it work easily for overlapping traps as well
as abutting ones.

The polygon code would then allocate a picture with this field and keep it 
updated as trapezoids are rendered.  The underlying compositing code could 
then avoid transferring the known-zero values.

Keith Packard        XFree86 Core Team        HP Cambridge Research Lab