Re: timezone variable not set correctly, bug blocking LSB Conformance

[Martin Zobel-Helas <zobel@ftbfs.de>]

Hi Aurelien,

On Mon, Jul 03, 2006 at 06:44:16AM +0200, Aurelien Jarno <aurelien@aurel32.net> wrote:

> Hi SRM team!
> 
> Would such a fix to the glibc be accepted for a stable release?

i would be first interested, which LSB versions are effected by this. I
don't see any need for fixing newer LSB issues in Sarge, as Sarge was
targeted LSB 2.0 compatibility.

Greetings
Martin

> Bye,
> Aurelien
> 
> 
> Martin Dittmar wrote:
> >Package: libc6
> >Version: 2.3.2.ds1-22sarge3
> >
> >Problem: after calling the C method "ctime" the global variable 
> >"timezone" is not set correctly with certain TZ environment variables.
> >With TZ=JKL3:10PNM4:40 set, the value of "timezone" is expected to be 
> >11400, but has a nonsense value of 18000.
> >
> >For a demonstration please see the attched file test.c
> >
> >This is a libc bug (already in libc Bugzilla (see 
> >http://sourceware.org/bugzilla/show_bug.cgi?id=2865 )), but can be fixed 
> >in Debian.
> >
> >This would be important for LSB compliance (LSB Runtime tests LSB 
> >runtime tests T.ctime_X 1, T.localtim_X 1, T.mktime_X 1).
> >LSB test failure message:
> >       with TZ=JKL3:10PNM4:40 ctime() did not set timezone correctly
> >       value of timezone was 18000, expected 11400
> >
> >A patch for sarge is attached.
> >
> >Kernel on tested system: 2.6.14-2-686-smp
> >libc version: 2.3.2.ds1-22sarge3