Re: noverlay branch

[Eli Zaretskii]

> From: Ihor Radchenko <yantar92@gmail.com>
> Cc: monnier@iro.umontreal.ca,  emacs-devel@gnu.org
> Date: Sat, 08 Oct 2022 10:36:28 +0800
> 
> There are a couple of glitches though:
> 1. hl-line-mode overlay sometimes disappear
> 2. notmuch search buffer hides text in wrong places (see the attached video)
> 
> Should I report these things as proper bugs or is it OK to discuss them
> within this thread?

Separate bug reports, please.  And please mention there that the bugs
pertain to the noverlay branch.  Thanks.

> Also, I did a small test comparing Org mode folding performance (using
> overlays) on master vs feature/noverlay branch using the methodology
> described in https://blog.tecosaur.com/tmio/2022-05-31-folding.html:
> 
> On large 20Mb Org file using built-in Org version:
> 
> master: Elapsed time: 142.940028s (1.428150s in 55 GCs)
> feature/noverlay: Elapsed time: 4.369854s (1.451378s in 74 GCs)
> 
> for comparison, the Org version from main yields:
> 
> Emacs master:
>  - main Org branch using text properties for folding: Elapsed time: 2.185786s 
> (0.639190s in 4 GCs)
>  - main Org branch using overlays for folding: Elapsed time: 27.244284s 
> (0.731581s in 5 GCs)
> 
> Emacs feature/noverlay:
>  - main Org branch using text properties for folding: Elapsed time: 1.586936s 
> (0.476606s in 3 GCs)
>  - main Org branch using overlays for folding: Elapsed time: 2.039803s 
> (0.724576s in 5 GCs)
> 
> The improvement is very significant.

Thanks, looks very encouraging.

> We also have tests for marker scaling. If the overlay tree
> implementation can be reused for markers, it would also be great.

Not in Emacs 29, I'm afraid.