bug#64177: outline-minor-mode-highlight fails to activate for custom headings

[uzibalqa]

------- Original Message -------
On Wednesday, June 21st, 2023 at 6:13 PM, Juri Linkov <juri@linkov.net> wrote:


> > > > (defvar elisp-hglevels
> > > > '( (";;; H1" . 1) (";;; H2" . 2) (";;; H3" . 3) (";;; H4" . 4)
> > > > (";;; H5" . 5) (";;; H6" . 6) (";;; H7" . 7) (";;; H8" . 8) ))
> > > > ...
> > > > ((eq major-mode 'emacs-lisp-mode)
> > > > ...
> > > > (setq outline-regexp
> > > > (concat (regexp-opt (mapcar 'car hrklevels)) "\\>"))
> > > > (setq outline-heading-alist hrklevels))) ))
> > > > Please also try:
> > > 
> > > (setq-local outline-level 'outline-level)
> > > 
> > > because 'emacs-lisp-mode' overrides the default value
> > > with own 'lisp-outline-level'.
> > 
> > Now that I have tried your suggestion, I can the the requested highlighting.
> > 
> > I see that the value of outline-level is lisp-outline-level. Whereas with 
> > your change
> > it is outline-level.
> > 
> > In principle, shouldn't outline-level be directly determined by the level 
> > specified
> > in outline-heading-alist ? If emacs-lisp-mode sets outline-heading-alist 
> > itself
> > properly then any level information can be extracted from there. Then the 
> > setup would
> > run more smoothly because it would be consistent with the way one would 
> > actually
> > customise it.
> 
> 
> This makes sense. OTOH, if we will change the logic, it might break existing 
> code?
> 
> Current situation:
> 
> (defun outline-level ()
> (or (cdr (assoc (match-string 0) outline-heading-alist))
> (- (match-end 0) (match-beginning 0))))
> 
> but many modes override this default with own function that ignores
> outline-heading-alist:
> 
> lisp-outline-level, c-outline-level, cperl-outline-level, perl-outline-level,
> tcl-outline-level, cfengine2-outline-level, conf-outline-level,
> nroff-outline-level, latex-outline-level
> 
> - all these ignore outline-heading-alist.
> 
> To support outline-heading-alist unconditionally the current calls
> 
> (funcall outline-level)
> 
> in many places should be replaced with
> 
> (or (cdr (assoc (match-string 0) outline-heading-alist))
> (funcall outline-level))
> 
> I don't think this is feasible, so probably we need only
> to add a warning to the docstring of outline-heading-alist.

In the modes that override the default, outline-level could be
set up to use outline-heading-alist, avoiding the conditional.