Bastien <b...@gnu.org> writes: > Hi Aaron and Tod, > > Aaron Ecay <aarone...@gmail.com> writes: > >> This seems like an excellent use case for the parser: basically a bunch >> of uses of org-*-regexp and org-re-property need to be augmented with >> a check like: >> (not (memq (org-element-type (org-element-at-point)) '(src-block >> example-block ...))) >> >> A better alternative might be to use the parser to find the property >> drawer in the first place (instead of a regex). Either way, it seems >> like the best strategy might be to fix all uses of these problem >> variables at once, which is a big undertaking. > > Also don't forget the cost in terms of speed. It's fine to fix the > behavior of Org for such cases, but those cases are rare, and could > be explicitely prevented. If the general fix does not slow down the > parsing too much, then I'm all for it. Nicolas might have better > insight here than me.
With newly introduced parser cache, the cost is amortized when using `org-element-at-point' and `org-element-context'. With a little care, the overhead is negligible. Therefore, these functions should be used whenever possible. I'm slowly working on introducing these functions in basic parts of Org (e.g. fontification, indentation). Then I will target other functions not using them yet. There's a lot to do. Help is welcome. IMO, at the bare minimum, new or rewritten functions should use them. For example, a way to correctly find the property drawer associated to the current section is to use something like the following: (save-excursion (outline-previous-heading) (unless (bobp) (let ((limit (save-excursion (outline-next-heading) (point)))) (catch 'exit (while (re-search-forward org-property-start-re limit t) (let ((drawer (org-element-at-point))) (when (eq (org-element-type drawer) 'property-drawer) ... (throw 'exit ...)))))))) Regards, -- Nicolas Goaziou