To: drums@cs.utk.edu
Subject: Re: replies, replies... 
In-Reply-To: Your message of "Fri, 07 Nov 1997 16:31:42 BST."
             <199711071531.QAA23213@beatles.cselt.it> 
Date: Sat, 08 Nov 1997 12:18:11 +1100
Message-Id: <7488.878951891@munnari.OZ.AU>
From: Robert Elz <kre@munnari.oz.au>

    Date:        Fri, 7 Nov 1997 16:31:42 +0100 (MET)
    From:        Maurizio Codogno <mau@beatles.cselt.it>
    Message-ID:  <199711071531.QAA23213@beatles.cselt.it>

  | I object to your statement "Reply-To has never been defined nearly well
  | enough for people to really understand how it should be used".

Well, it would help us all, considerably, if you would point us to the
definition which is good enough for us to understand how it is to be used.
Then perhaps we could end this discussion.   The rest of us, while we may
differ on how we think it ought be defined, I think all agree that there
is currently no rational definition.

  | We can argue whether the model which says
  |     The  "Reply-To"  field  is added by the message
  |     originator and is intended to direct replies.
  | is correct (we have already done it, therefore I won't add anything)
  | but this is a definition.

It's a definition, and it is useless.   It isn't a question of being
correct or not, it simply doesn't say enough to be useful.

  | Even if RFC822, section A.2.4, explicitly says
  |      A.2.4.  Committee activity, with one author

Note that title - that's an example of a single author (in the From header)
who wants replies to go to a bunch of people (who are possibly not
getting copies of the message, it doesn't matter).   Nothing strange
there.    If the committee were to be the author, the committee would
be in the From header.

  | You are quite correct, but this shows only that Reply-To has a definite
  | meaning different from that of the originator of the message.

I agree there.

  | Not at all. I am suggesting that Reply-To: should be an authenticated
  | address exactly like From:, that it indicates where the originator
  | wants any answer to the message to be sent,

That's what I am suggesting (except that "authenticated" is a loaded word,
probably impossible to enforce, and doesn't belong).   I am attempting to be
a little more explicit, and make it quiet clear that the content of the
Reply-To is the entire list of addresses where the originator wants any
answer to the message to be sent.

  | and that people answering to the
  | message should be aware that the originator added a Reply-To:.

That's a quality of UA issue, and I don't think belongs.

  | By the way, I personally read the sentence
	[...]
  | as a way to have more reply buttons,

Yes, I agree.

  | I still contend that the *default* button (what I
  | call Simple-Reply) should use it.

And if there is no such button?   I prefer not to design the UI here,
just define the headers and what they mean (I keep saying this...).   If
we do a good enough job of that, the UA people won't have many difficulties
designing reasonable UIs.

kre