status safari merge

[David Hyatt]

On Oct 19, 2003, at 1:44 PM, Dirk Mueller wrote:

> Note I'm not at all objecting to getting rid of the collapsing of the
> individual dirtyrects into a big boundary dirty rect. This is hitting 
> us bad
> as well, and a solution is needed. However, when there are complex 
> regions
> (regions are assembled as unions of rects in X11), stuff slows down
> drastically, as it all boils down to array operations (elements being 
> the
> individual rects of the region). the boundary rectangle implementation 
> is not
> something we put big design thoughts in. Its a hack, and it served the
> purpose so far.
>
> You were saying that Panther does this simplification of complex dirty 
> regions
> already for you? Or did I misunderstand something?
>

I am not sure what Panther does with the individual rects, i.e., 
whether it does any smart coalescing.  I'll need to delve into the 
source to find out.  Panther basically gives you a choice when drawing 
of using the bounding box that encloses all the individual rects or of 
walking the individual rects as a list.  I am unsure whether it does 
any optimizing/coalescing on that list.  Safari's drawing code is not 
using the individual rects yet.