Re: Overlays as an AA-tree

[Andreas Politz]

Stefan Monnier <address@hidden> writes:

> What's missing for it to be "good enough for master" (which AFAIK means
> mostly that the code is clean, there are no known bugs, and the
> performance is varies between "no worse" and "much better" than what we
> currently have)?

Don't know about "clean".

I found the bug I was mentioning earlier, but it literally took hours.
I needed to construct a tree with 7 nodes while performing a series of
operations in a very specific manner.  It is really difficult to
anticipate these kinds of bugs within a test-suite.

Benchmarks I've posted in the past:

https://lists.gnu.org/archive/html/emacs-devel/2017-02/msg00488.html
https://lists.gnu.org/archive/html/emacs-devel/2017-02/msg00598.html
https://lists.gnu.org/archive/html/emacs-devel/2017-02/msg00736.html

> I don't think we should care too much about those degenerate cases.

Yes, we could wait and see if these cases (mostly many overlapping
overlays) make a difference in real applications.

-ap