Message-Id: <199803250409.XAA15983@spot.cs.utk.edu>
From: Keith Moore <moore@cs.utk.edu>
To: Pete Resnick <presnick@qualcomm.com>
cc: Keith Moore <moore@cs.utk.edu>, Jacob Palme <jpalme@dsv.su.se>,
        IETF working group on revision of mail standards <drums@cs.utk.edu>,
        moore@cs.utk.edu
Subject: Re: Need for multiple reply destination sets? 
In-reply-to: Your message of "Tue, 24 Mar 1998 21:38:43 CST."
             <v04010f04b13e258e21ba@resnick2.qualcomm.com> 
Date: Tue, 24 Mar 1998 23:09:26 -0500
Sender: moore@cs.utk.edu

> >(If you just want Mail-Followup-To to specify where followups go,
> >the author's user agent shouldn't be automatically generating it.
> 
> That doesn't make sense. Of course, I should never do anything in my MUA
> which hamstrings the user into doing something my way or no way at all. But
> certainly if I keep a list of the mailing lists to which the person
> subscribes, it is a good convenience feature to set the Mail-Followup-To
> for the user to some default that works most of the time and leave it to
> the user to change it if they desire, based on content.

If done on a per-list basis, it might work okay.

> >Yes, but that 99% of the time doesn't include the cases where the
> >author has specified an alternate reply address.
> 
> I don't understand. Are you saying that Mail-Followup-To doesn't provide
> semantics for an alternate reply address for the author? Of course it
> doesn't. That's got to be the job of some other field.

No.  I'm saying that the 99% case is the one for which there is
no Reply-To or Followup-To or Respond-To or any preference at all
set by the author of the subject message.  For that 99% of the time,
the majority of the responder's reply wishes can be lumped
into "reply to author" or "reply to everybody".

But when the author wishes to specify a reply preference, it's not
nearly so often that the author's reply preference can be lumped
into "author" or "everybody" categories.

> >I agree that 99% of the time, (in the absence of an explicitly
> >specified reply address) a responder wants to reply either to
> >just the author(s) or to all recipients of the subject message.
> >We don't need additional reply/respond headers to facilitate that.
> 
> Obviously mailing lists and some users think we do given how they use
> Reply-To. Giving them a mechanism to make that semantics explicit (which
> Reply-To clearly does not) can only help.

Users may think that they need the ability to specify such preferences,
but in reality they don't need this ability unless their user agent
is so broken that it won't let them specify From.  Since that user
agent has to be fixed to incorporate any changes to reply behavior
anyway, it might as well be fixed to let the user specify From at the
same time.

Lists may think that they need the ability to specify where replies
go, but such a reply is neither "reply to everybody" nor "reply to 
author" nor even "reply to group" (since there can be multiple
groups and there's no way for the MUA to know which group is being
referred to)   The responder shouldn't be misled into thinking that 
he is doing a "reply to X" when the reply doesn't go to X. 

> >But for the small fraction of the time where the author  of the
> >subject message has recommended an alternate reply address, a
> >significant portion of those recommendations are not properly
> >labeled either "author/sender" nor "all/everyone/group".
> 
> I don't see that at all. Any time I can think of when I wanted to specify a
> different address than From or From/To/Cc, it's either (a) a place where I
> wanted group discussion to take place, but not redirect mail intended to be
> privately to me; (b) an address I will receive mail at temporarily that's
> not mine; or (c) a person I want to handle all further correspondence about
> this particular topic. Mail-Followup-To handles (a). Certainly solutions
> for (b) and (c) are not, but Reply-To or other fields might serve for them.

Reply-To won't serve for (c) because it is too widely interpreted 
as alternate From address.  Mail-Followup-To or Respond-To or 
whatever will serve equally well for (c) as for (a) if we get 
rid of the notion that a "followup" is inherently to "everyone" 
or to "the group".


> >If you want to define some abstract "Respond" operation, which is
> >explicitly NOT deemed to be synonymous with "reply to all" or
> >"reply to everyone" or "reply to the group" but which is instead:
> >
> >IF the responder specifies a set of recipient addresses, use those,
> >ELSE IF the author of the subject message specified a set of
> >	Respond-To recipient addresses, use those,
> >ELSE use the From/To/Cc fields of the subject message.
> >
> >I think that would be fine.
> 
> Actually, in our discussions about Eudora internally, we're thinking about
> defining three replies in the UI:
> 
> 1. Reply to Author - replies to From. There will be an option in the
> configuration for whether another field (likely Reply-To for the time being
> for most users) will be used instead of From.
> 
> 2. Reply as followup - replies to the Mail-Followup-To field if present. If
> it isn't present, replies to From/To/Cc. Again, there will be an option for
> whether something like Reply-To will be used instead of From, and perhaps
> an option to use Reply-To if Mail-Followup-To is not present. We'll give
> the user as much freedom as possible.
> 
> 3. Reply specially - brings up a dialog with all of the addresses we can
> find with a little check box next to each. Depending on some user settings,
> certain addresses will be checked by default.
> 
> Again, notice that 1 and 2 are the most common operations that most users
> perform and they will be happy to have them automated in this way without
> having to resort to 3.

This sounds very promising.  I think you could probably even combine
2 and 3 as long as you had a checkbox that allowed the responder
to ignore mail-followup-to.  (you do have to decide whether toggling
that checkbox overrides or ignores all of the per-address toggles 
that the user has already done, but either one seems better than
forcing the user to commit to either "followup" and "special" before
he's seen for himself what "followup" will do.)


> >This, I think, is the real sticking point.  MUA authors want their
> >software to be able to make the decision of where to send replies
> >on behalf of the responder.
> 
> No. You keep getting hung up on this, but it is emphatically false. MUA
> authors want their software to be able to automatically do some things that
> the user instructs them to do, and we want those instructions to be quick
> and easy to access for the user. I want the user to be able to easily
> access a command which follows the rule: "Send this message to the
> discussion group that is having this discussion. If the sender has made
> mention of where a good place for that is, use it. If not, send to everyone
> in the From, To, and Cc fields." 

As long as the user has the flexibility to reply to whomever he wants,
(including the ability to ignore the author's wishes)
and he isn't surprised by the behavior of his UA when he hits the
"reply to FOO" button, I have no problem with it.  

Keith