Errors-to: owner-drums@CS.UTK.EDU
Message-Id: <199508252237.SAA13527@wilma.cs.utk.edu>
From: Keith Moore <moore@CS.UTK.EDU>
To: Pete Resnick <presnick@qualcomm.com>
cc: Mark Crispin <MRC@CAC.Washington.EDU>, Keith Moore <moore@CS.UTK.EDU>,
        drums@CS.UTK.EDU
Subject: Re: From the Chair: the Reply-To issue 
In-reply-to: Your message of "Fri, 25 Aug 1995 17:02:05 CDT."
             <v03002c05ac63f221af91@resnick1.isdn.uiuc.edu> 
Date: Fri, 25 Aug 1995 18:36:54 -0400
Sender: moore@CS.UTK.EDU

> Also, I think we
> should be a little leary of defining what a UA default behavior is like in
> the first place, let alone using the SHOULD NOT language. I think this is a
> better replacement:
> 
>         A Reply-To: header indicates the sender's preference of an
>         address to be used for replies in place of the address in
>         the From: header. When a Reply-To: header is present, a UA
>         SHOULD use the Reply-To: address when generating replies and
>         SHOULD NOT use the From: address. Many UA's also provide an
>         additional reply function where all recipients (including
>         those in From:, To:, and Cc: headers) may be addressed in
>         the reply. When such a function is used, if a Reply-To: header
>         is present, the Reply-To: address should be used in place of
>         the From: address.


I agree that it is dangerous to specify user agent behavior.
But the above statement doesn't get us out of that trap.

My view is that the To, Cc, From, and Reply-To headers are intended for 
communication between sender and recipient -- they're supposed to tell 
the *recipient* who the message is from, who it's being sent to, and to 
where replies should be addressed.  

If this is the case, we're not defining a "protocol" with these headers 
so much as a "language" -- a standardized way for the sender to communicate 
such things to the recipient.  So we need to define these headers according 
to what they *mean* as opposed to how user agents act upon them.

--

So on this particular issue, I want people to very clearly state
whether Reply-to expresses the author's preference about (pick one):

[ ] where replies to the message should be sent

    (thus for the recipient to honor the author's wishes, the reply
    the addresses in the reply-to header rather than to any of the 
    other combinations)

[ ] where replies to the *author* should be sent

    (thus a reply might reasonably be sent to the
    reply-to address + addresses from the to and cc headers,
    without violating the author's intent)

I understand Pete to be in the latter camp.  But I want to keep this
discussion focused on this crucial choice rather than getting bogged 
down in what happens when a particular *kind* of reply is sent. 

Keith