Errors-to: owner-drums@CS.UTK.EDU
Date: Thu, 17 Aug 1995 11:20:04 +0200 (MET DST)
From: Jacob Palme <jpalme@dsv.su.se>
To: Eric Thomas <ERIC@SEARN.SUNET.SE>
Cc: ietf-drums <drums@CS.UTK.EDU>
Subject: Re: "Reply-To"
In-Reply-To: <199508161703.TAA23586@mars.dsv.su.se>
Message-Id: <Pine.SUN.3.91.950817111400.27446H-100000@ester>
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII

On Wed, 16 Aug 1995, Eric Thomas wrote:

> I'm afraid  this is  not ok at  all. You can't  take something  away from
> millions  of  users  who  have  been  using  it  for  9  years  to  their
> satisfaction  without providing  a replacement.  If you  can't provide  a
> replacement, you can't  take away the solution people are  using now. End
> users don't  care about standards. If  they need feature X  and company Y
> doesn't provide it any longer because  the standards were just changed to
> say that there  is no way to  do it that doesn't violate  the specs, they
> will pay company  Z to do it. The  one thing they won't even  think of is
> stopping their use  of a feature they want just  because a committee said
> it isn't a good idea.
> 
> Personally, I think it's a lost  cause. A replacement is only useful (and
> the current usage can  only be deprecated) if it is  available to a large
> majority of  users. This is  going to take at  best 3-5 years.  RFC822 is
> already much more complex than it needs  to, and that's the reason why so
> many implementations are incorrect. If you make it even more complex, you
> run the  risk of  having implementations  that do  support the  new reply
> features, but incorrectly. And if that  is the case, people will continue
> to use "Reply-To:" because it works  all the time. Conversely, the reason
> it works all the  time is that it is very simple  to understand. If there
> is  a "Reply-To:"  field and  the user  presses the  normal/default reply
> button, you send the reply there.

Do I understand you to say that the practice of using "Reply-To" to
redirect discussion to the mailing list instead of the author is
(a) in common usage (b) works well?

My experience is quite the opposite, and that is the main cause for
taking up this for discussion in the ietf-drums group.

We implemented, a few years ago, a message system making this assumption.
I.e. we hade the system such that replies by default are sent to the
"Reply-To" address, unless the writer of the reply did something special
to send the reply somewhere else.

Our experience was that this implementation was a disaster, and that
we had to change it. The disaster occurred because a lot of replies
intended for the list were wrongly sent only to the author, since
so many existing messages, also messages to mailing lists, have only
some kind of address for the author in the "Reply-To" field.

Would you really like e-mail systems to be designed so that the
address in the "Reply-To" field is the default recipient set
by the mail client, both when the user gives the "personal reply"
or "reply all" command? Have you really experience with mail
systems designed in this way which users are satisfied with??

------------------------------------------------------------------------
Jacob Palme <jpalme@dsv.su.se> (Stockholm University and KTH)
for more info see URL: http://www.dsv.su.se/~jpalme