Sender: Chris Newman <chrisn+@CMU.EDU>
Date: Tue, 26 Mar 1996 15:24:17 -0500 (EST)
From: Chris Newman <chrisn+@CMU.EDU>
Subject: Re: timezones & Date
To: drums@cs.utk.edu
In-Reply-To: <199603261929.OAA17321@CS.UTK.EDU>
References: <199603261929.OAA17321@CS.UTK.EDU>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Message-ID: <827871857.1525.0@nifty.andrew.cmu.edu>

The problem situations we have identified with times are:

1) GMT time is known, but the timezone is unknown.

2) local time is known, but the timezone is unknown.

3) GMT time known, timezone may or may not be correct.

4) local time known, timezone may or may not be correct.

In cases 1 & 3, it is probably best to generate the Date header in GMT
time and forget about communicating the sender's timezone.  In case 4,
assuming the timezone is correct is least likely to cause problems.
However, case 2 should have some indication that the timezone is
unknown.

Is case 2 a serious problem with RFC 822 dates?  What systems does
this happen on?

I've heard three suggestions for indicating case #2 -- LDL, +0001, and
-0000.  Were we to decide that #2 was a serious real-world problem I'd
vote for "-0000" on the grounds that I think it is least likely to
break anything.

First datapoint: the Macintosh, prior to System 6.0.4 (approximately),
was in category #2, but all modern Macintosh systems fall in category
#4.  I would consider #2 largely irrelevant for the Macintosh.
-----
Chris Newman <chrisn+@cmu.edu>, http://www.contrib.andrew.cmu.edu/~nifty/
CMU & Compu$erve: Your freedom of expression ends at their newsfeeds.