> -----Original Message----- > From: Thomas Narten [mailto:[EMAIL PROTECTED] > Sent: Tuesday, March 21, 2006 7:48 PM > To: Durand, Alain > Cc: Bob Hinden; IPv6 Mailing List > Subject: Re: Proposed M&O bits text for RFC2461bis > > No, it means that if you don't invoke DHC to get addresses, > you won't be getting all the addresses the network is telling > you it thinks you need. That is, DHC is providing addresses > that are (presumably) not available via stateless addrconf, > hence you should use DHC to get them.
So, that would imply that, if you need to get more addresses, the ones you get from stateless autoconf are not enough to operate. In other words, addresses are not equals and some may be "better" (for some definition of better) than the others or at least have different purposes... But how does the node knows wich ones are "better", or which ones to use? In particular, if the node sources a datagram, which address is it suppose to use? If we leave it to the default address selection rules, this ends up being defined by a longest prefix match with the destination, which may or may not correspond to this notion of "better" addresses to use. This would seem to imply that when we revisit the default address selection rules, we may want to add an extra rule to deal with that case. - Alain. -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
