problem with tzcode96l

Paul Eggert eggert at twinsun.com
Tue Oct 29 19:21:17 UTC 1996


   Date: Tue, 29 Oct 1996 11:35:14 -0700
   From: "Todd C. Miller" <Todd.Miller at cs.colorado.edu>

   we're trying to get OpenBSD to pass the NIST POSIX test suite and
   it tries to do a mktime() with TZ=dst0std,J057,J059 with a time of
   Tue Feb 26 01:59:59 1991 (which should return 667533600).

This makes it sound like the NIST Posix test suite is incorrect.
With those arguments, mktime should return 667533599 if tm_isdst is
negative or 0, and an implementation-defined value (typically either
667533599 or 667529999) if tm_isdst is positive.

   mktime() from tzcode96l (and earlier versions) returns -1 with that
   TZ setting.

If mktime returns -1 then that is a bug, but I can't reproduce the bug
when building tzcode96l on Solaris 2.5.1.  I used

	make cc=gcc CFLAGS='-DHAVE_GETTEXT=1 -DHAVE_LONG_DOUBLE=1 -DHAVE_SETTIMEOFDAY=3 -DHAVE_STRERROR=1 -DHAVE_UTMPX_H=1'

to build tzcode96l.  mktime had the behavior described above
(it returned 667529999 for the positive tm_isdst case).



More information about the tz mailing list