On 10.5.2013, at 08:39, Achim Gratz <strom...@nexgo.de> wrote: > Carsten Dominik writes: >> by decoration you mean font-lock support? > > Yes, but I also don't think these should ever become indented in the > first place. That's debatable of course, the syntax as defined by > org-element does not require this IIRC.
Well, which are the ones you think should never become indented? OPTIONS, TITLE, of maybe you mean the whole suite of keywords? I sometimes put the setup below a major headline "* setup" or so, to make them hide away and give the buffer a clean look. Depending on indentation setting it then does make some sense to allow indentation. So I think the font-lock fix is more important than the indentation one. Please read on below. > >> Maybe this would be a better fix: >> >> Modified lisp/org.el >> diff --git a/lisp/org.el b/lisp/org.el >> index 745fb82..43df094 100644 >> --- a/lisp/org.el >> +++ b/lisp/org.el >> @@ -5867,7 +5867,8 @@ by a #." >> ((or (equal dc1 "+results") >> (member dc1 '("+begin:" "+end:" "+caption:" "+label:" >> "+orgtbl:" "+tblfm:" "+tblname:" "+results:" >> - "+call:" "+header:" "+headers:" "+name:")) >> + "+call:" "+header:" "+headers:" "+name:" >> + "+options:")) >> (and (match-end 4) (equal dc3 "+attr"))) >> (add-text-properties >> beg (match-end 0) > > Sure, that helps too and has certainly less potential for controversy. > > I think we should make an effort to shift most if not all the regex > stuff in org.el into org-element. There's far too much duplication with > subtle differences sprinkled all over the place to get match data to > work with and it's almost hopeless to try and find all such uses for a > single element. What do you mean? Do you meant to use the org-elemnt parser and base also font-lock on it? Or do you mean all the definitions of regexp constants. This sounds desirable - but it also sounds like an extremely daunting task with possibilities for problems in side effects of regexp matching that will be difficult to find and might only show after a long time. I guess we could start such a process one regexp at a time. - Carsten