RR bounce test

Jeroen C. van Gelderen jeroen@vangelderen.org
Wed, 12 Apr 2000 12:26:45 -0400


"Pascal J. Bourguignon" wrote:
[...]
> Actually, about every user agent I know of has defined its own receipt
> request header. Probably, it would be best to configure your mail list
> software to  filter out (silently)  all headers but  some specifically
> selected one. Perhaps there is some RFC defining standard ones.

That would *really* be a bad idea. Headers have a purpose (ever tried to
trace a mail loop or spam or a character set problem without headers?)
and the fact that some mailers/users are so stupid as to return receipts
to the wrong address is IMHO not a good reason to mutilate message
headers. 

Another IMO very good reason for simply bouncing offending messages is
user education: silently stripping notification request headers will not
do anything to discourage people from sending messages with a request
for receipt.

Fortunately there is a finite number of mailers (which can interoperate
to some extent) out there so it should be easy enough to compile a list
of the relevant headers. For Netscape 4.x it is:
  Disposition-Notification-To
as I just tested.

I personally think it is up to the user to handle receipt requests
sensibly but at the same time I can see where Jonathan is coming from.
Mutilating message headers however is not a good idea and I'd encourage
to just continue bouncing offending messages.

Cheers,
Jeroen