Errors-to: owner-drums@CS.UTK.EDU
Date: Thu, 14 Sep 1995 17:28:50 -0700 (PDT)
From: Mark Crispin <MRC@Panda.COM>
Sender: Mark Crispin <mrc@Ikkoku-Kan.Panda.COM>
Subject: re: summing up: the meaning of the reply-to header
To: Keith Moore <moore@CS.UTK.EDU>
Cc: drums@CS.UTK.EDU, moore@CS.UTK.EDU
In-Reply-To: <199509150001.UAA21752@wilma.cs.utk.edu>
Message-Id: <MailManager.811124930.2030.mrc@Ikkoku-Kan.Panda.COM>
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; CHARSET=US-ASCII

On Thu, 14 Sep 1995 20:01:39 -0400, Keith Moore wrote:
> 1. There is rough concensus that the Reply-To header specifies
> where replies to the *message* should go, as opposed to replies
> to just the author.
>
> In particular, I can find only limited support for the idea
> that Reply-To overrides only the address(es) in the From header.
> So there's a lack of concensus on this sub-issue.

I'm not sure that any of these necessarily conflict in the real world.
Consider this model, which is actually what many MUAs follow:
	The author of the message is its Sender.
	The address that receives replies is the Reply-To.
	The Sender and Reply-To, if not specified, default to the From.
This is silent on the matter of whether or not replies should go to the To/cc
addresses.  In most MUAs, this is left up to user choice, as stated here:

> 2. There is rough concensus that the recipient can, in principle,
> reply to anyone she wants to, and that user agents can provide
> functions to make it easy for her do so.

One way of stating the argument is that some people evidentally think that the
Reply-To header should prevent a reply, even under user choice, from going to
the To/cc addresses as well.  If you accept the principle stated in (2), then
you must reject this argument, and the issues in (1) become moot.

Put another way; it does not matter whether you say "Reply-To specifies where
replies to the message should go, as opposed to replies to just the author",
since the dichotomy is meaningless.  Say instead, "replies go to the reply
address, which *defaults* to the author" and "the decision of whether or not
replies go to the To/cc addresses is up to the user."

> (However, the Chair is willing to consider proposals to deprecate
> Resent-* and replace it with a similar but better defined mechanism
> using new field names.  If these new fields were useful for mailing
> lists, the Chair would still consider them within the scope of this
> group.  Otherwise, interested parties are invited to develop proposals
> for new headers for mailing lists outside of the DRUMS group.)

I think that ReSent is too entrenched.  I wonder, though, if it would be
possible to introduce a requirement that all ReSent-* headers be written in a
block starting with ReSent-Date, and that it is:
	1) mandatory that new ReSent headers be at the end of a header
	2) forbidden to relocate or reorder ReSent headers