To: djb@koobera.math.uic.edu (D. J. Bernstein)
Cc: drums@cs.utk.edu
Subject: Re: timezones & Date 
In-Reply-To: Your message of "24 Mar 1996 08:57:22 GMT."
             <19960324085722.11600.qmail@koobera.math.uic.edu> 
Date: Sun, 24 Mar 1996 22:40:40 +1100
Message-Id: <5910.827667640@munnari.OZ.AU>
From: Robert Elz <kre@munnari.oz.au>

    Date:        24 Mar 1996 08:57:22 GMT
    From:        djb@koobera.math.uic.edu (D. J. Bernstein)
    Message-ID:  <19960324085722.11600.qmail@koobera.math.uic.edu>

    The difference is how you handle violations.

I'm not sure I believe that.

    I know this is heresy in the standards world, but publishing a standard doesn't create instant
    compliance, especially if you add stupid requirements to the standard.

That's fine.

    You have to acknowledge this and figure out what will happen if the
    requirements aren't universally followed.

Sure.
    
    Certainly (1) the sender must include Date in every message. But various
    people, considering the effects of violations, also want to require that
    (2) an MTA will pass along a message even if it doesn't have Date.

OK.

    If Date is in the syntax, (2) becomes a syntax error.

Well, perhaps, but if the text says "must include Date" then
the difference in error is pretty minor.

    Then you don't end up blaming the MTA for the MUA's failures.

Ideally MTAs (other than adding Received, and Return-Path) should
never go near any of the message content, shouldn't check it,
modify it, etc, at all.   Having the MTA check the message makes
about as much sense as having an anon FTP server syntax check C
programs before handing them out.

If anything is to object to the Date not being there, it would
be the receiving UA (or perhaps a gateway along the way, I
view gateways as an in/out UA pair tightly coupled).   But
whether that UA does this check or not, is pretty much orthagonal
to how the requirement is stated in the spec, it depends more
on the mood of the UA author and how pedantic they wanted to
be (or how much they really need the Date in order to function).

kre

ps: yes, I realise that some of the above is pretty strange
from someone who publically just admitted violating exactly the
principle espoused, but ocasionally pragmatics get in the way...