At 6:31 PM -0800 2/28/07, Hallam-Baker, Phillip wrote: > >This is a design choice in the protocol, one that I would see as a layering violation. Application layer protocols should not be talking about IP addresses. Network management protocols are arguably at the application layer (and they certainly feed into applications in the traditional sense). Many of those need to talk about interface addresses and the addresses of the other ends of flows. This is, naturally, just one example of an application layer protocol that is an exception to the preference you express. Once you get past all of the exceptions, though, things look to me more and more like engineering judgement calls based on trade-offs rather than hard-and-fast rules. With a different architecture, we would have gotten different trade-offs (and we might yet), but from where we are now, I think we have to acknowledge that this engineering choice is not going away. Ted Hardie _______________________________________________ Ietf@xxxxxxxx https://www1.ietf.org/mailman/listinfo/ietf