Re: comment-start redefined in cc-langs.el

[Alan Mackenzie]

Hi, Martin!

On Tue, 2 May 2006, martin rudalics wrote:

>Hi, Alan!

> > I'm not sure why it reports cc-engine.el, ...

>... because symbol-file finds cc-engine.elc first in load-history, I
>presume.

I'm sure now!  Near the start of cc-engine.el, there's a macro
c-declare-lang-variables, which generates a defvar for each symbol stored
inside c-lang-variable-inits.  3 of these symbols are comment-start and
friends.

> > A quick grep for "(c-lang-defvar " reveals that these are
> > the only non-CC Mode variables redefined in this way.

>While you're here: What about C-h v for non-non-CC Mode variables
>defined this way?

Some of them (more precisely, those intended to be available to derived
modes) have doc-strings.  Others are specifically marked as _not_ having
doc strings.  This has been a little contentious on emacs-devel recently,
but if it is to change, it should be done over the whole code atomically.

>.. And maybe

>(put 'c-lang-defconst 'doc-string-elt 2)

What's doc-string-elt?  (OK, I can guess, but I can't find it in the
Elisp manual).

>and eventually

>(put 'c-lang-defvar 'doc-string-elt 3)

>for ease of reading cc-langs.el?

I'm in the middle of documenting the entire language variable mechanism.
This thing is a tremendous achievement and saves hours of work, but it is
somewhat sparsely commented, and hence difficult to debug.  Give me a few
days!

>Sincerely, martin.

-- 
Alan (Munich, Germany).