Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!ucsd!ucsdhub!esosun!seismo!uunet!auspex!guy
From: guy@auspex.UUCP (Guy Harris)
Newsgroups: comp.sys.next
Subject: Re: How to set the time zone on a NeXT (under 0.8 at least)
Message-ID: <990@auspex.UUCP>
Date: 10 Feb 89 09:49:03 GMT
References: <10176@ut-emx.UUCP> <1676@mace.cc.purdue.edu>
Reply-To: guy@auspex.UUCP (Guy Harris)
Organization: Auspex Systems, Santa Clara
Lines: 25

>You must also teach the kernel your time zone.  Otherwise it will require
>a PST bias when you enter the time with date(1) and sendmail will continue
>to stamp your letters with PST.

That sounds like a botch.  If a machine is running the Arthur Olson timezone
code, as NeXT appears to be doing, programs using the standard
"localtime", "ctime", etc. routines should be getting the time zone
information from there, not from the kernel.  Only binaries built with a
library that doesn't use the Arthur Olson code should be fetching it
from the kernel.

Perhaps NeXT switched to the Olson code recently, and either 1) haven't
updated their C library yet or 2) if the programs in question don't use
a shared C library, haven't relinked them against the new library yet. 

Given that it's not FCS software (right?), this may be the case (the
lack of "zic" or documentation for same, as indicated by the original
poster, indicates that this might be the case).  I would presume they'll
fix it for FCS (and will provide the source to the time zone data bases,
not just the compiled files, so that when your legislature changes the
rules on you you're not stuck with the old rules - admittedly, since the
files are supposed to be in a machine-independent format, you could, if
necessary, update the databases on a Sun running 4.0 or later or a Mac
II running A/UX - assuming Apple provides the source to the databases; I
know Sun does - and zip the data files over to the NeXT).