On Mar 31, 2008, at 10:10 AM, Bob Hinden wrote: > Fred, > >>> I think I am in general agreement of the consequences, but it isn't >>> about scope, it is about reachability. Anytime there is a a >>> choice of >>> addresses to use, the same issue come to play. It's hard to know >>> about >>> reachability with out prior knowledge or trying it out and seeing >>> what >>> works. This applies to ULAs and other global scope unicast >>> addresses, >>> IPv4 and IPv6 addresses, etc., etc. >> >> By this line of reasoning, scope is irrelevant. The only thing that >> the Robustness Principle would guide towards is the ability to >> exchange datagrams and open state with one's peer. > > Yes (with the obvious exception of IPv6 link-local addresses). I > think that without some sort of active confirmation, anything else > we can do is a hint as to the best order of trying out a source and > destination address pair.
That is of course what I'm getting at. A host doesn't participate in routing, and any assertion that "well, he could use prefix foo with prefix bar if he knows that one can reach the other" presumes either configuration or routing knowledge. Absent either configuration or routing knowledge of more specific reachability, two interfaces that are using the same prefix (could be link-local prefix, ULA prefix, or global prefix) can talk with each other using the common prefix. Two devices that don't share a prefix shouldn't know about addresses the peer has that they can't use. Note, however, that DNS purists will tell you that there should be no question of address domains - they simply report the records they have on file, which may yield some addresses not reachable via ambient routing. Therefore, especially given the DNS purist model, such systems need to assume that either they connect from a global address to global address or try the cross product of their two address sets to see what works. If you don't like the latter - and yes, there are good reasons to not like it - you're stuck with the former. -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------