> > > > The alternative is to configure systems that reside on the same link > > to have the same prefixes. > > there are lots of other ways to solve that problem. none of them can be > adequately described in one sentence - there are too many cases to consider.
Point taken. There are other ways to solve the problem that the on-link assumption was originally trying to solve. Do you have suggestions on how the draft can better address other possible solutions (if at all)? > > There's no expectation that apps > > communicate using link-local addresses. > > so the scenario described above was one in which hosts communicate > using link-local addresses for some other purpose than to support apps? No, the scenario described in the draft doesn't necessitate using link-local addresses for any purpose. Even if the on-link assumption were taken out of the Neighbor Discovery sending algorithm, there would be no need for apps to communicate using link-local addresses given adequately configured hosts. -Seb -------------------------------------------------------------------- IETF IPv6 working group mailing list [EMAIL PROTECTED] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------