Re: "Warning: `save-excursion' defeated by `set-buffer'" madness

[Barry Fishman <barry_fishman@acm.org>]

Stefan Monnier <monnier@iro.umontreal.ca> writes:
>>> > Is there any way I can get rid of these warnings globally?  It's
>>> > really annoying to have *Compile-Log* pop up every now and then.
>>> I put the following in my .emacs file:
>>> ;; Shut off compiler error pop-up warning about
>>> ;;    save-excursion defeated by set-buffer
>>> (setq byte-compile-warnings '(not suspicious))
>> Oh, great, thanks!
>> I knew there had to be a way...
>
> I really hope you did first report the warning to get it fixed.  I
> currently only know of one external package (BBDB) where this warnings
> pops up like this, so if you're using BBDB, try to install a newer
> version.

You think a new version would help?  The CVS head of BBDB still looks
like it would generate the compiler warning.  Not to mention w3m, w3,
and possibly other packages used with Gnus, that I don't feel like
scanning through at the moment.  They are easy enough to find just using
grep.

I do report problems that I find to the appropriate developers.  In this
case the issue is not necessarily a true bug but more a statement about
how the Emacs code should be written.  It is not the kind of thing I
feel I should be enforcing.  If you can't get these things fixed
upstream, why do you think I can?  I did fix all of my own code, and the
packages that I use that no longer had upstream support.

I send an off-list message to Štěpán because he did not seem to be
getting a reply from anyone on the list, and I thought it seemed cruel
to have him suffer because the Emacs developers want to send an annoying
message to the external package developers.  There already had been
discussion on the list about the issue, and I saw no value in going over
it again.  I just showed him what he wanted, which is how to shut off
the message, which nobody else seemed to have posted to the list.

I run repo versions of Emacs and the external Emacs packages that I use,
when they are avaliable.  In part it is in self defence.  The
development versions tend to be reliable, and when interfaces are
changed it usually happens one at time, rather that all at once.  This
gives me an opportunity to fix them in my code.  I do report bugs that I
find (and patches if I can) to the upstream developers.

If I just used releases, I would see multiple problems all at once, and
have to go back through the development mailing list to figure out what
to do, long after any discussion was over.  I have too much of my own
code for that approach to work.

-- 
Barry Fishman