[cairo] GSoC: Scan converting rasteriser update

[Carl Worth]

On Thu, 2008-10-02 at 12:32 +0300, M Joonas Pihlaja wrote:
> The existing rendering pipeline goes via the fallback surface
> where there's a lot of code that implements drawing differently
> based on the nature of the path, the compositing operation, and
> the clip state, so plugging a spans rendering method alongside
> the trapezoid rendering in cairo-surface-fallback.c seemed like
> the right choice to me.

Sure. We can't blame you for following what's there.

> Reorganising the existing rendering framework was out of scope
> for the project, which is what changes such as the proposed
> image_surface_fill() sounds like:

And it would seem almost unfair to say "Your new code looks great, but
you have to clean up the existing code before you can add it".
However...

> Having said that, I'm not particularly happy about the state of
> cairo-surface-fallback.c, especially now that there are at least
> three different ways of rendering paths (regions, traps, and
> spans.)

This is the tricky part.

We haven't done that cleanup of the existing code, and now that we've
got some attention on the code, it would be an ideal time to do it. If
we miss this chance then the code gets even more complex and will be
even hard to fix later.

I guess I just don't want to miss this window of opportunity.

> Thanks for the review.  I'll get on the easy cleanups and dead
> code elimination and make noise when they're pushed.

Great. I'll look forward to that.

And like I mentioned earlier, please feel free to also send the whole
patch series here to the list if you'd like.

-Carl