🚀 go-pugleaf

Kevin> I've been looking into adding font-lock-keywords for other modes.
Kevin> It's a real pain to have to add something like:- [real pain code]

OK, I addressed the problem in font-lock.el by adding the following new
variable:

(defvar font-lock-defaults-alist
  '(...)
  "*Alist of major mode and Font Lock defaults.
Each item should be a list of the form:
 (MAJOR-MODE . (FONT-LOCK-KEYWORDS KEYWORDS-ONLY CASE-FOLD FONT-LOCK-SYNTAX))
where both MAJOR-MODE and FONT-LOCK-KEYWORDS are symbols.  If KEYWORDS-ONLY is
non-nil, syntactic fontification (strings and comments) is not performed.
If CASE-FOLD is non-nil, the case of the keywords is ignored when fontifying.
FONT-LOCK-SYNTAX should be a list of cons pairs of the form (CHAR . STRING), it
is used to set the local Font Lock syntax table for keyword fontification.")

one such item is:

    (c-mode . (c-font-lock-keywords nil nil ((?\_ . "w"))))

Kevin> Another pain is having to set font-lock-no-comments if the mode
Kevin> doesn't have a comment syntax defined.

See above.

The above solution does mean that, if the new font-lock-defaults-alist is
not autoloaded, you have to do things like:

 (eval-after-load "font-lock"
  '(or (assq 'foo-mode font-lock-defaults-alist)
       (setq font-lock-defaults-alist
             (cons '(foo-mode . (foo-font-lock-keywords t nil t ((?\_ . "w"))))
                   font-lock-defaults-alist))))

to add foo-mode to the list.  The difficulty with font-lock.el is that a
loading a package (to satisfy, say, an autoload for foo-mode) may or may
not need font-lock.el (foo-mode-hook has or has not 'turn-on-font-lock),
and font-lock.el may not be around.

I thought about building the variable name to use for font-lock-keywords
by (intern (concat (symbol-name major-mode) "-font-lock-keywords")), but
it (a) didn't seem reliable enough, and (b) didn't address the problem of
setting font-lock-no-comments, font-lock-keywords-case-fold-search, and
font-lock-syntax-table.

Setting font-lock-no-comments on the basis of the mode's comment-start (or
better comment-start-skip, at least font-lock-fontify-region uses that)
does of course assume that you never want to fontify strings if you don't
set comment-start[-skip].  I'm not sure about that, though I haven't an
example to show it;-)  It doesn't address the other two variables anyhow.

My solution is still verbose, but I couldn't think of a way to deal with
all the 4 font-lock.el variables that need setting.

What do you think?

Si.