error("No match %d in highlight %S" 2 (2 'org-headline-done prepend))

[yantar92]

I doubt that font-lock is the root to the problem here. Kind of hoping to be proven wrong though!

I did a benchmark from your branch @yantar92. No performance improvement. Slower actually. Output of the tests:

With org-mode master:
Elapsed time: 1.016928s (0.102247s in 3 GCs)
Elapsed time: 115.648919s (3.886891s in 85 GCs)
Elapsed time: 0.726889s (0.054811s in 1 GCs)

With yantar92/org-fold-universal-core:
Elapsed time: 1.184400s (0.213159s in 7 GCs)
Elapsed time: 116.078592s (4.321491s in 70 GCs)
Elapsed time: 0.770061s (0.094068s in 1 GCs)

First row is with new customization on. Second with it off (showing current performance) and third is with all files already loaded into buffers.

I can note that benchmarking org-roam--list-all-files shows that's actually a significant piece of the time spent already there:

(benchmark 1 '(org-roam--list-all-files))

results:

Elapsed time: 0.390924s

These numbers are with my personal setup though, Emacs 28.0.50 on Windows 10 with native compilation on. Surely they can differ in another persons config but I suspect the relative performance difference is around the same. (i.e. around 100x faster with this customization)

Sidenote: Got an error with the org-fold-universal-core branch. error("No match %d in highlight %S" 2 (2 'org-headline-done prepend)) Can discuss that offline if needed, to not take over this pull-request.

Originally posted by @Whil- in alphapapa/org-ql#200 (comment)

[yantar92]

It looks like match-data is changed somewhere where it is not supposed to.

[yantar92]

@Whil- By any chance, do you know what kind of org file triggers the error?

[Whil-]

FIle sent offline

[yantar92]

FIle sent offline

Thanks! Though I cannot reproduce neither with my config nor with clean emacs. I guess, there is some strange interaction between this branch and org-roam.

[Whil-]

Hmm, yeah. Sucks to debug other ppls's problems... :( Sorry for not being more specific here.

If it helps it could have to do with org-startup-folded being set to show2levels. Also, there were a couple of compile warnings related to org-fold (I believe was the name). Don't have the record of exactly what now though.

[yantar92]

If it helps it could have to do with org-startup-folded being set to show2levels. Also, there were a couple of compile warnings related to org-fold (I believe was the name). Don't have the record of exactly what now though.

Thanks! I tried show2levels, but still cannot reproduce. Anyway, I pushed a change preventing recursive calls to font-lock, which might help here.

[Whil-]

Ok, sorry for not being of more help here. Let's assume it's a problem with my setup for now then since it's difficult to reproduce.