Sender: jwz@netscape.com (Jamie Zawinski)
Message-ID: <342D6351.F5F95191@netscape.com>
Date: Sat, 27 Sep 1997 12:49:37 -0700
From: Jamie Zawinski <jwz@netscape.com>
Organization: Netscape Communications Corporation, Mozilla Division
MIME-Version: 1.0
To: "Eric S. Raymond" <esr@snark.thyrsus.com>
CC: Dave Crocker <dcrocker@brandenburg.com>, schaefer@brasslantern.com,
        drums@cs.utk.edu
Subject: Re: Reply-To (was Re: Issues from the past few weeks)
References: <199709271906.PAA00160@snark.thyrsus.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Eric S. Raymond wrote:
> 
> The present language of 4.4.4 reads:
> 
>             o   If the "Reply-To" field exists, then the reply  should
>                 go to the addresses indicated in that field and not to
>                 the address(es) indicated in the "From" field.
> 
> I propose that the editor enroll the following correction without further
> debate:
> 
>             o   If the "Reply-To" field exists, then it overrides the "From"
>                 field and MUST be used instead of the "From" addresses for the
>                 purpose of assembling reply address lists.

I agree that this seems a sensible clarification of what Dave says 
was the original intent.  

It tells me clearly that one thing I should *not* do is just ignore 
the Reply-To header, which is something I was suspecting I was going 
to end up having to do.

It also tells me all I need to know about how I should implement 
my "reply to author" header, which is: direct replies to
(Reply-To OR From OR Sender).

I think that it doesn't tell me clearly enough what to do about 
"reply to all".  I think that, with that language, it's still open 
to debate  whether my "reply to all" command should be implemented 
as ((Reply-To OR From OR Sender) AND To AND Cc) or as
(Reply-To OR ((From OR Sender) AND To AND Cc)).  Because, while
that language talks about the From field, it doesn't talk about 
To or Cc.  

Perhaps it's appropriate that 822bis not talk about how Reply-To 
relates to the To and Cc fields, but I don't believe that.

In the interest of removing ambiguity, I would suggest that one 
of the following three pieces of text be appended to the preceeding
paragraph.

   o   The "Reply-To" field overrides only the "From" field. 
       In the case where it is desired to include "To" and/or "Cc"
       addresses in a reply, and the "Reply-To" field exists, then
       the "Reply-To" field does not override the "To" and "Cc" fields.

   o   In the case where it is desired to include "To" and/or "Cc"
       addresses in a reply, and the "Reply-To" field exists, then
       the "Reply-To" field overrides the "To" and "Cc" fields as
       well.

   o   The behavior of the "Reply-To" header with respect to any other
       address header, such as "To" or "Cc", is undefined.  In the case
       where it is desired to include "To" and/or "Cc" addresses in a
       reply, and the "Reply-To" field exists, an implementation may
       choose to direct replies to only the "Reply-To" field, or it may
       choose to direct replies to the union of the "Reply-To" field
       and the other desired address header(s).

Based on Dave's latest comment:

Dave Crocker wrote:
> 
>> I think this most closely matches case (1b).  That is, 'Reply-to'
>> overrides 'From', but doesn't affect any other possible recipients.
> 
>         This is exactly what Reply-to was intended to do.

I think we should go with the first.

-- 
Jamie Zawinski         http://people.netscape.com/jwz/      about:jwz