tobold.org

correct • elegant • free

△ comp.mail.* △

◅ Sender: vs. Resent-From: for bounced messages

rfc822 structured headers ▻

Sender: vs. Resent-From: for bounced messages

In article <D7s7M9.Hz2@unicom.com>, Chip Rosenthal <chip@unicom.com> wrote:
>In article <3nqtnh$oq0@pipe.pipex.net>, Tim Goodwin <tim@pipex.net> wrote:
>>However, in my opinion it is better to avoid the Resent-* headers
>>altogether---there are several problems with them.  It is much more
>>satisfactory to use MIME's message/rfc822 content type to resend a
>>message.
>
>Can you elaborate?  What problems?

STD 11 (RFC 822) does not define what constitutes "resending" a message,
and the Resent-* headers have been used for a variety of quite distinct
actions.  I know of one UA that uses Resent-* headers and actually
allows you to edit the message you are resending!  I can (and have)
argued that this is thoroughly misleading, but there's nothing in the
standard to say it's illegal.

There are some STD 11 headers that contain trace information: `Received:'
and `Return-Path:'.  When a message is resent, should these be stripped
(throwing away information), left alone (potentially confusing), or
munged (e.g. changed to something non standard such as `Old-Received:')?

If you reply to a resent message, should you use the addresses in
the `Resent-Reply-To:' / `Resent-From:' fields?  Or `Reply-To:' /
`From:'?  Do you use the `Message-ID:' or the `Resent-Message-ID:' in
the `In-Reply-To:' field?

Given all this, it's clear that there can be no way to recover the
original message (before it was resent), so there's no sanctioned way to
resend a message twice.  STD 11 doesn't define Resent-Resent-* headers,
and it makes explicitly undefined the interpretation of multiple
Resent-Foo fields of the same type.

All of this can be avoided by encapsulating the message to be resent
using MIME's message/rfc822 content type.

>                                    I don't think MIME encapsulation
>is such a hot way of handling mailing list redistribution.  The Resent-*
>headers seem just dandy for that.

Erp.  I see mailing list expansion as an MTA function.  The STD 11
headers should not be altered at all.

Tim.
--
Tim Goodwin        | "BIND does not like VM." -- Paul Vixie
Public IP Exchange |

Original headers:

From: tim@pipex.net (Tim Goodwin)
Newsgroups: comp.mail.headers
Subject: Re: Sender: vs. Resent-From: for bounced messages
Date: 2 May 1995 13:51:19 +0100
Organization: PIPEX, 216 Science Park, Cambridge, England
Message-ID: <3o59s7$8q2@pipe.pipex.net>
References: <3np8a1$79h@news.aero.org> <3nqtnh$oq0@pipe.pipex.net>
  <D7s7M9.Hz2@unicom.com>

△ comp.mail.* △

◅ Sender: vs. Resent-From: for bounced messages

rfc822 structured headers ▻