Message-Id: <199704190138.VAA19307@ig.cs.utk.edu>
From: Keith Moore <moore@cs.utk.edu>
To: schaefer@nbn.com
cc: drums@cs.utk.edu, moore@cs.utk.edu
Subject: Re: How Reply SHOULD Work 
In-reply-to: Your message of "Fri, 18 Apr 1997 17:24:30 PDT."
             <970418172430.ZM28595@candle.brasslantern.com> 
Date: Fri, 18 Apr 1997 21:38:48 -0400
Sender: moore@cs.utk.edu

> > Subject: How Reply SHOULD Work
> > 
> > 1. For a large class of messages -- those with a single From address,
> >    no Reply-To field, and only one recipient -- everybody's notion of
> >    "reply" does the Right Thing.
> > 
> > 2. For EVERY OTHER MESSAGE not in the above class, I want to
> >    consciously choose the set of recipients of any reply that I send.
> 
> That's fine, but what does that imply for 822bis?  That mention of the
> relationship between From: and Reply-To: should be omitted?

I'm still thinking about how to put it in a protocol specification.
For now, I'm interested in people's reaction to the basic idea.

One possibility is that the 822bis simply describes the meanings
of From and Reply-To, along with some advice for appropriate use
by the sender of a message, leaving the user interface issues to
a separate Informational or BCP document.

> In my experience, most users of mail UAs do want the software to make
> an informed choice for them, and want to be able to direct the software
> to make the same choice every time so they can stop thinking about it.

My current UA lets me edit the message headers.  This lets me make
deliberate choices, but editing the headers is a pain.  It would 
be really nice if my UA made default choices for me, but let me
change them easily.  For instance, if my idealized UA were replying 
to your message, it could display something like

From: Keith Moore <moore@cs.utk.edu>
  To: [x] schaefer@nbn.com (reply-to address)
      [ ] "Bart Schaefer" <schaefer@candle.brasslantern.com> (author address)
  Cc: [ ] drums@cs.utk.edu, [ ] moore@cs.utk.edu
Subj: [Re: How Reply SHOULD Work]

...where the [x] or [ ] boxes would be toggle buttons, allowing me 
to turn off any reply address at will.   But the toggles would
be pre-set (as shown) according to the instructions in the Reply-To
header and my personal preferences.

Of course, I'd also like to be able to add other addresses not 
present in the original message header.


> Telling them that they ought to be thinking about it doesn't go over
> well; it's like telling them they ought to obey the speed limit and not
> accelerate on yellow.

Yep.  But these are still deliberate acts; people's cars don't 
speed or run yellow lights by themselves.

Keith