Date: Wed, 25 Mar 1998 10:41:08 -0800 (PST)
From: Chris Newman <Chris.Newman@innosoft.com>
Subject: Re: Reply-To Harsh realities
In-reply-to: <v04003a04b13ec6747280@[130.237.150.138]>
To: Jacob Palme <jpalme@dsv.su.se>
Cc: Detailed Revision/Update of Message Standards <drums@cs.utk.edu>
Message-id: <Pine.SOL.3.95.980325090543.3495A-100000@elwood.innosoft.com>
MIME-version: 1.0
Content-type: TEXT/PLAIN; charset=US-ASCII
Originator-Info: login-id=chris; server=THOR.INNOSOFT.COM

On Wed, 25 Mar 1998, Jacob Palme wrote:
> At 11.43 +0100 98-03-25, Chris Newman wrote:
> > (10) I am not comfortable with the current Mail-Followup-To proposals.  I
> > agree with Keith that it doesn't seem to match what _senders_ wish to
> > express.
> 
> Explain in more detail what is wrong with it, and I will try to improve it.

Every time I manually set a reply-to header, the semantics I always want
are "this is the complete set of addresses were I want *all* replies
sent."  I don't do this very often, so I don't care if the procedure is
made more complex, but I think this is the singular common case for manual
reply-to.  I'm not sure how to express this with Mail-Followup-To as it
forces a dicotomy which may be natural for the recipient, but is quite
unnatural for the sender.  Examples:

* I send to multiple lists and want responses to go to only one of the
  lists and not to me.
* I send to one list and don't want responses to go to me
* I send to a list and want to collect all responses personally and
  summarize, so I don't want responses to go to the list
* I reply to "service" email as myself (I'm the author -- not Innosoft
  service), but want to make sure "service" is the primary recipient of
  all replies so they don't get black-holed if I go on vacation.

How would I express these with any of the "Mail-Followup-To" proposals?

In addition, Keith's complaint about overloading Mail-Followup-To both as
a potentially automated way to notify the recipient not to send duplicates
to list mail as well as a manual way to notify where replies are sent
makes me nervous.  It's not a show-stopper in my book, but it doesn't feel
safe.  I also think allowing mailing lists to set Mail-Followup-To is
quite troublesome.  We've had a lot of experience that overloading
Reply-To causes lots of problems.  I don't want to see the introduction of
a new header until I'm sure it is specified well enough and correctly
enough that we won't repeat the Reply-To fiasco to a greater or lesser
degree.

I'd like to see simple proposal overviews which address the following points:

* How the common-case personal/group (or public/private) 
  auto-reply-address-generation model interacts with the proposal
* How it expresses "send all replies to this complete set of addresses"
* How it can assist prevention of duplicates
* How it can co-exist with the installed base of clients, servers, 
  gateways and munging-listservers
* Why new list munging will/won't happen and why that's OK
* Why the new header(s) will not have any of the problems we've seen with 
  Reply-To
* Why the new header(s) will not be abused or implemented incorrectly.

I note there is a cognative dissonance between the first two points and I
suspect this is the rat in the rathole.

		- Chris