Message-Id: <199712300433.XAA20520@spot.cs.utk.edu>
From: Keith Moore <moore@cs.utk.edu>
To: Jacob Palme <jpalme@dsv.su.se>
cc: Keith Moore <moore@cs.utk.edu>,
        IETF working group on revision of mail standards <drums@cs.utk.edu>
Subject: Re: Table of all the issues and all the major proposed solutions 
In-reply-to: Your message of "Sun, 28 Dec 1997 03:36:45 +0100."
             <v03110702b0cb66a5c722@[130.237.150.138]> 
Date: Mon, 29 Dec 1997 23:33:09 -0500
Sender: moore@cs.utk.edu

> > The biggest issue I have with this table is that you've biased against
> > the "fix the definition of Reply-To" solution by saying that
> > implementors won't accept it.  Well, no solution can solve a problem
> > if implementors don't accept it, but the entire purpose of the
> > exercise is to *persuade* implementors (and in general, the entire WG)
> > of the superiority of a particular solution.  Your table pre-judges
> > such a conclusion, assuming that WG members would reject a particular
> > solution out-of-hand even if it became obvious that it were the best
> > solution.  Frankly, I think this is insulting to the WG.
> 
> We have tried to discourage mailing lists owners from munging Reply-To
> for many years, without success. This shows that a standard, to succeed,
> must give what implementors need, otherwise they will not follow the
> standard.

We agree on this much.

> Any attempt to fix the existing "Reply-To" has the disadvantage that
> we will, for many, many years into the future, not be able to know
> whether a Reply-To header is according to old practice or the new
> standard. 

Right.  But if the new practice is an obvious improvement from the old,
it might not matter too much.   Right now we have the situation that
almost all MUAs handle reply-to poorly when given a "reply all" command -- 
by using Reply-To+To+CC, they behave in a way that isn't terribly useful.  
If this were changed to using only Reply-To, they immediately become
more useful without creating a significant amount of pain.

The reason this works is that Reply-To is used a lot more often
to mean "where I want replies to go" (by both authors and lists)
than to mean "please use this address instead of my From address".
And most MUAs these days allow authors to set their From address.


Keith