Message-Id: <3.0.32.19971031131659.008a2350@pop.dial.pipex.com>
Date: Fri, 31 Oct 1997 15:58:06 +0000
To: Jacob Palme <jpalme@dsv.su.se>
From: Graham Klyne <GK@ACM.ORG>
Subject: Re: Can different messages have the same Message-ID
Cc: IETF working group on revision of mail standards <drums@cs.utk.edu>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"

At 20:57 30/10/97 +0100, Jacob Palme wrote:
>Graham Klynes idea, that if the identical content is sent with revised
>headers, then the new message should have a new Message-ID but the same
>Content-ID seems like a very interesting idea. However, does any existing
>mail system implement that principle? Most existing mailers probably keeps
>the same Message-ID when the same content is resent or forwarded. Some
>systems depend on the Message-ID. For example, Usenet News uses the
>Message-ID (not the Content-ID) for duplicate supression. And "In-Reply-To"
>and "References" refer to the Message-ID, not the  Content-ID. If Graham's
>idea is made a standard, then "In-Reply-To" and "References" should in most
>cases use the Content-ID, since the replying message is a reply to the
>content. (A recipient of a reply should be able to follow the thread even
>if he has a different version of the message than the person who wrote the
>reply.)  Usenet News ought to start using Content-ID instead of Message-ID
>for duplicate supression.
>
>My feeling is that this idea is nice, but perhaps too difficult to get
>accepted because it requires so many changes to existing software. It will
>never get accepted unless the drums document strongly says that this is the
>new standard.
>
>Question to drums: Do we really want to make this radical change to how
>messaging standards are presently implemented and used? We have to decide
>on this issue, before we can continue the discussion of what changes in a
>message should and should not require a new Message-ID.

I think a major obstacle to such change would to effectively mandate the
use of MIME with these protocols.

Without attempting to answer the question, a thought...  *IF* such a change
is to be considered, then maybe it should go further to permit the use of
URLs (e.g. 'cid:' and 'mid:' per <draft-ietf-mhtml-cid-v2-00.txt>).

As a transitional measure, maybe ( Message-ID | URL ) might be permitted in
fields which currently use just Message-ID?

GK.
---

------------
Graham Klyne