bug#20179: 24.4; Crash when S-TAB'ing on Org entry

[Sebastien Vauban <sva-news-D0wtAvR13HarG/iDocfnWg@public.gmane.org>]

Eli Zaretskii wrote:
>> From: Sebastien Vauban <sva-news-D0wtAvR13HarG/iDocfnWg@public.gmane.org>
>> Date: Tue, 24 Mar 2015 12:17:24 +0100
>> 
>>> Please attach GDB to Emacs right after you start a new session, type
>>> "continue", and let Emacs run as usual.
>> 
>> Just to be sure, nothing else to type: no "source .gdbinit" or
>> "xbacktrace" or something along those lines?
>
> It doesn't really matter.  If needed, you can always do that when the
> problem occurs and GDB regains control.
>
>> And, is it OK to try to reproduce those problems with the above
>> (release) version?
>
> It is better to try with the pretest of 24.5, if you can.  But if not,
> 24.4 is also OK.

I'm testing GNU Emacs 24.4.91.2 (x86_64-pc-mingw32) of 2015-03-13 on
LEG570 for a couple of days, directly from GDB, without crash.  I have
the _impression_ that my problem is not present (not at all, or not as
often) in the pretest of 24.5 as it was in 24.4 -- good news if this is
confirmed over time.

Though, this morning, in the above configuration (Cygwin GDB + pretest
of Emacs 24.5), I had a problem with the Org timestamps: they were one
hour late, like if the computer clock hadn't been updated with the
"summer" times here in Europe (since this Saturday).

I finally found that the problem does not come from Emacs itself, but
from the fact it's launched from Cygwin GDB!

When launched directly, the times (result of `format-time-string') are
correct...

See http://screencast.com/t/lFMrkqBgDR for a comparison.

Any idea how this problem can be circumvented?

Best regards,
  Seb