At 1:09 PM -0700 5/13/97, John W. Noerenberg wrote:
Tony, if you have some ideas that can address Valdis' concerns, I'd be very
interested in hearing more. I have to be honest and say that I think it is
highly unlikely you can guarantee the cancel in the distrbuted environment
of the Internet. And that is what will be required.
I even think you would be hard pressed to prove that the nntp CANCEL is
100% effective. By that I mean that no one at any site had read a message
that was also cancelled.
My understanding of NNTP CANCEL is that it is far from guaranteed; but then
that also goes for NNTP in general. CANCEL messages get lost, messages get
cancelled after they are read, so what? Also, NNTP specifically says that
sites should not do much authentication on the CANCEL message; certainly
forging one is quite easy.
If there were to be an SMTP CANCEL (maybe RESCIND?) I think it would have
to be modelled on receipts -- strictly advisory, no guarantees to the
sender that anything will happen, up to the receiver how to deal with one.