Re: [Brett Presnell] Re: Fedora Core 3: File mode specification error

Ralf Angeli <[email protected]> writes:

> * David Kastrup (2005-01-20) writes:
>
>> So the only question that remains is how we are going to deal with
>> this ourselves.  It would appear that Fedora Core 3 has shipped with
>> an Emacspeak causing the problem.  We had several reports of this by
>> now if I remember correctly.
>
> Two.
>
>> Probably there are not many calls of regexp-opt that can be passed
>> "nil".  A grep turns up quite a few calls, however.  As far as I can
>> see we have stuff in font-latex.el, tex-font.el (what is that?  The
>> header says "keep this in synch with Emacs-21.1 which we haven't, and
>> the changelog has no entry except this having been checked into
>> AUCTeX), tex.el and tex-buf.el.
>
> tex-font.el is a rip-out of fontification code from tex-mode.el.  You
> can activate it by customizing `TeX-install-font-lock'.  If there
> still is demand for that we could update it when Emacs 21.4 will be
> out.  But I doubt many people are still using it.
>
> Regarding `regexp-opt': The only place where it certainly will get
> passed nil is `font-latex-set-syntactic-keywords' because the
> variables `font-latex-verbatim-macros' and
> `font-latex-verbatim-macros-local' are nil per default.  So I think it
> would be enough to check only there, that it is only called with a
> non-nil value.  I don't see a need for replacing every call to
> `regexp-opt' with a to-be-implemented `TeX-regexp-opt' which could
> handle the nil case.  (There are likely other ways to do this.)  And
> the problem will go away in the not so far future anyway.
>
> A proposed patch is attached.  Yes, it looks ugly, but still better
> than let-binding another set of variables.

How about (let ((this (append whatever bla))
                (that (append blubb blah)))
             (setq this (and this (regexp-opt this))
                   that (and that (regexp-opt that)))

Of course, this depends on nil being a valid value for this and that.

If not, we need

(setq this (if this (regexp-opt this) "")
      that (if that (regexp-opt that) ""))

That needs no additional let-binding.  On the other hand, it also does
not appear much of an improvement.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum