Re: [Emacs-diffs] /srv/bzr/emacs/trunk r109512: Inline functions to examine and change buffer overlays.

[Stefan Monnier]

> I think that it's reasonable to have just one chain of overlays per
> buffer, much like the markers and intervals chains per buffer text.

Why?
The point of having 2 is that they're sorted in opposite order, so that
finding overlays close to the division point is faster than O(n).

I think the only meaningful improvement for overlays would be to move
them to the interval tree, thus replacing the O(n) worst case by an
O(log n) worst case for most operations.


        Stefan