Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10 5/3/83; site asgb.UUCP
Path: utzoo!watmath!clyde!bonnie!akgua!sdcsvax!bmcg!asgb!nkn
From: nkn@asgb.UUCP
Newsgroups: net.unix-wizards
Subject: TZ in Other Countries: Summary
Message-ID: <587@asgb.UUCP>
Date: Mon, 26-Nov-84 11:38:29 EST
Article-I.D.: asgb.587
Posted: Mon Nov 26 11:38:29 1984
Date-Received: Wed, 28-Nov-84 02:39:24 EST
Reply-To: nkn@asgb.UUCP (Neil Nelson)
Organization: Burroughs Corporation, Boulder Colo.
Lines: 28

<Is there really a line-eater bug?>

A month ago I posted a query about how TZ is handled in countries outside
of the U.S.A.  This is a summary of the responses.

First the problems: (1) Some countries have time zones which are not an
integral number of hours past GMT.  AT&T's UNIX(tm) cannot handle a zone
which is, say, 3 1/2 hours past GMT.  (2) Different countries begin/end
daylight savings time on different days of the year.  AT&T's UNIX assumes
the begin/end dates are the same everywhere.

The solution: (1) I wasn't too worried about this one, since there are
many possible fixes.  Most people suggested something along the lines
of "MST7:30MDT" to indicate that MST is 7 1/2 hours past GMT.  This
requires a simple fix to tzset() in ctime.c.
(2) This one is a little tougher.  Berkeley has a good solution here.
Under AT&T's UNIX, "daylight" is an on/off flag which indicates whether
or not daylight savings time is in effect.  Berkeley's "tz_dsttime"
not only gives an on/off indication, but also, if on, which table to
look up the begin/end dates in.  I guess if someone wanted to do this
with AT&T's UNIX they would have to add these tables and alter the
meaning of "daylight" accordingly.

My thanks to all of you who responded.

Neil Nelson
Burroughs Advanced Systems Group
. . .!sdcsvax!bmcg!asgb!nkn