[Mac-emacs-devel] "format" bug in Mac Emacs 21

[Dan S. <seg...@ma...>]

There is a very strange bug lurking somewhere in the Mac Emacs 21
code.  I exchanged some private email with Andrew Choi (thanks
again, Andrew!), and at least we're both getting the same results.

At 19:29 -0600 2001-10-30, Dan Sommers wrote:
> Copied and pasted from my *scratch* buffer:
> 
>     (format "%f" 1.0)  ; "%f" loses
>     "0.000000"
>     (format "%s" 1.0)  ; "%s" wins
>     "1.0"
> 
> I traced this one all the way down to the call to sprintf on
> line 3416 of editfns.c, and everything looks good except
> sprintf's output.  it happens with or without my .emacs file,
> and with various compiler optimization and PPC processor
> settings.  I'm using CodeWarrior 5.3 and [the] pre-packaged
> project, and Mac OS 9.0.4, if that makes any difference.  I'll
> give it a shot later with MPW, as soon as I finish downloading
> it (at 56K...).
> 
> I think this bug also existed in Mac Emacs 20.6.  I tracked it
> down because of strange output from ps-print (*everything* was
> black, including the background), and I had that same symptom
> in 20.6.

[ Aside:  ps-print loses because it depends on (format "%f ...)
to format RGB values, and if (format "%f" ...) keeps returning
"0.000000," then all those colors are *black*.

As a temporary work-around for this particular problem, I added
this to my .emacs file:

;; Mac Emacs (format "%f" 1.0) loses; these kluges prevent
;; ps-print from losing, too....
(setq ps-color-format " %s %s %s "
      ps-float-format " %s ")

End of aside. ]

At 13:58 +0800 2001-11-02, Andrew Choi wrote:
> Thank you for the bug report.  I've spent some time on it and I
> must say I'm baffled.  I added a simple block
> 
>   {
>     char s[20];
> 
>     sprintf(s, "%f", 1.0);
>   }
> 
> at the beginning of main() and that works correctly.  I also
> added this block right after the sprintf line in questions (in
> Fformat() in editfns.c) and it didn't work.  I also tried
> `(format "%f" 1.0)' in the MPW version and that gives a
> "0.000000" also.  Perhaps I can post this to mac-emacs-devel and
> see if other people have some ideas.

I had tried these things in the mean time, too, and got the same
results, and that's where we are.

Any ideas?

Thanks,
Dan

-- 
Dan Sommers, Esq., Ltd., The Third
mailto:seg...@ma...