Errors-to: owner-drums@CS.UTK.EDU
To: Chris Newman <chrisn+@CMU.EDU>
Cc: ietf-drums <drums@CS.UTK.EDU>
Reply-To: ietf-drums <drums@CS.UTK.EDU>
Subject: Re: 
In-Reply-To: Your message of "Mon, 18 Sep 1995 10:46:54 -0400."
             <811435614.3668.0@nifty.andrew.cmu.edu> 
Date: Tue, 19 Sep 1995 05:57:55 +1000
Message-Id: <18170.811454275@munnari.OZ.AU>
From: Robert Elz <kre@munnari.oz.au>

    Date:        Mon, 18 Sep 1995 10:46:54 -0400 (EDT)
    From:        Chris Newman <chrisn+@CMU.EDU>
    Message-ID:  <811435614.3668.0@nifty.andrew.cmu.edu>

    I am opposed to loosening the requirement that "Reply-To" always
    overrides "From" because: 1) This will break some user's expectation
    that when "Reply-To" is present, the "From" address will never be
    used.

Any user under that impression should have it corrected
immediately.   "Never" is way too strong.   Consider that
I have received a message, its from Joe, a friend, I have
nothing whatever to say about the content of the message, so
I am not going to reply, but I do want to know how his last
vacation was, as I'm considering going to the same place.
The message I received reminds me I wanted to ask.  In that
event I am likely to compose a new message, but cut & paste
Joe's address - and I'm going to do that using the From: line,
I will take no notice whatever of a Reply-To (this message is
not a reply).

I doubt anyone would dispute that as being a legitimate thing
to do (From: lines are required to have valid addresses in
them remember, even when there is a Reply-To).

But once we admit that From: can sometimes be used to generate an
address for messages, then it starts becoming pretty meaningless
to draw a line and say "ok for that message, but not this other
one".

This is partly why I believe that Reply-To should always (and
only) indicate the original sender's (author's, whatever)
view on the address to which replies to the message should be
sent.   That is, if you are really replying to the message, and
not just sending a new message, on either a related, or
different, subject, that is the address (or addresses) you
are encouraged to use.

    2) This will encourage even greater use of the wide-reply use
    of the "Reply-To" header, leaving all correctly implemented RFC822
    clients unable to do personal replies even more often.

I think I understand what you mean there, but I sympathise not
at all.   If you want to Reply to this message, you should
follow the Reply-To header I have inserted, and send the
message only there.   If you want to send me, and only me,
a message on this (or some other) topic, you should use the
From: line, but that message would not be regarded (by me anyway)
as a reply to this message.

    Currently, "Wide-Reply-To" is one of the permitted meanings of
    "Reply-To".

Yes.

    However, we already have concensus that banning the other
    meanings of "Reply-To" is not possible, so this would leave the
    status-quo ambiguity intact.

No we don't.   If we have consensus on anything on this, it
is certainly not that.

    I am opposed to any change in the meaning of "Reply-To" by this
    working group.  I believe the meaning cannot be changed to correct the
    current ambiguity without creating new headers.

This is another eay of saying that drums should do nothing,
and may as well go away.  New headers are outside the charter,
they're clearly new functionality.  What we're supposed to be
doing is removing the ambiguities.   Yes, really doing that, not
just abandoning them as too hard to bother with.   This will
sometimes mean hard choices.

Personally, I'm willing to listen to arguments that Reply-To
should be defined as "override From", or that it should be
defined as "replies are intended for these addresses and no
others", and perhaps others like that if any.   I don't believe
it is open to us to simply say "too hard, forget it".

kre