[Date Prev][Date Next][Thread Prev][][Date Index][Thread Index]

Re: bogus change(s) in cl-macs.el



Katsumi Yamaoka wrote:

   First of all, I protest strongly that such a big change has not
   been recorded on the ChangeLog file.  Second, the change at least
   to the `labels' macro breaks emacs-w3m.  Emacs-w3m uses some cl
   macros including `labels' in order to work together with the
   external w3m command asynchronously.  The patch below should be
   applied.  Third, I'm not sure whether a problem is limited only
   to it.

It was not me who installed that change to cl-macs, but I believe that
it was made in response to problems with compiler warnings I reported.
I myself suggested the following alternate type solution, which I
believe would not have broken anything:

!       (let ((var (with-no-warnings (gensym))))

but after I saw Stefan's patch, I did not pursue (and did not double
check) this.  I could still do it given that Stefan's solution seems
to give problems.  It would eliminate the compiler warnings with no
side effects.  I guess that this should be done everywhere since if
the change breaks things in one place, it probably does in other
places too.  I could post a patch for review, if desired.

I do not immediately understand why Stefan's patch breaks things.  It
fails to adjoin a unique identifying number, like gensym does, but I
do not understand why that is necessary.  Distinct uninterned symbols
with the same name are still distinct.  I guess that if I knew the cl
package better, I would understand.

Sincerely,

Luc.