Ihor Radchenko <yanta...@gmail.com> writes: > Kévin Le Gouguec <kevin.legoug...@gmail.com> writes: > >> Since the changes in Org 9.4 aimed at improving performance, is there a >> test case somewhere in the "Mitigating the poor Emacs performance on >> huge org files" thread that could help ensure that a tentative fix will >> not degrade performance? > > The first message in the thread ;) I believe it was also used to > benchmark the change in 9.4.
Thanks for the pointer! I've looked at your branch for inspiration, and my takeaway is that the isearch-open-invisible-temporary route might be too involved for a bugfix, especially if it's going to be reverted wholesale when your branch gets merged. Then again, maybe I'm not smart enough to devise a solution. I wonder if the path of least resistance couldn't be found in org-cycle-hide-drawers: right now this function just skips over drawers which are covered with an invisible overlay, but maybe it should not skip a drawer if the overlay starts before it (i.e. the overlay is not specific to this drawer but covers a whole containing section).
