ietf-822
[Top] [All Lists]

Re: the most obvious failure in To-NoReply

2004-08-30 11:34:23

Simon Josefsson writes:
At this point, perhaps it is useful to enumerate requirements on the
solutions.

You mentioned two requirements:

   (1) letting the sender exclude the reply address (normally From or
       Reply-To) from the followup list;
   (2) letting the sender include a public recipient while excluding
       that recipient from the followup list.

Here are two more requirements:

   (3) letting the sender _include_ the reply address in the followup
       list---consider non-subscriber messages to mailing lists;
   (4) letting the sender specify more than one address for followups---
       consider messages crossposted to two mailing lists.

To-NoReply flunks #1. The notion of having a ``post to list,'' separate
from replies and followups, flunks both #3 and #4.

Mail-Copies-To:nobody flunks #2. As for #1, Mail-Copies-To:nobody might
seem at first glance to be simpler than Mail-Followup-To, but in fact
the code to do

   Mail-Followup-To || (To + Cc + (Mail-Reply-To || Reply-To || From))
   Mail-Reply-To || Reply-To || From

tends to be a trivial adaptation of the code already written to do

   To + Cc + (Reply-To || From)
   Reply-To || From

whereas Mail-Copies-To:nobody breaks the address-list encapsulation and
requires some new low-level code.

Mail-Followup-To supports #1, #2, #3, and #4, and it does so in a way
that's reasonably easy for both ends to automate. It already has heavy
use in the real world. Nobody has suggested a simpler way to meet the
requirements.

---D. J. Bernstein, Associate Professor, Department of Mathematics,
Statistics, and Computer Science, University of Illinois at Chicago