On 23Jun05, on this list there began and ended a thread concerning ULAs, multicast, and RFC 3484. I agree with the contention espoused by some that RFC 3484 should be updated. Section 3.1 clearly states that with the presence of a multicast destination address, a comparison between the multicast address scope and that of the unicast is required. Not only is there comparison, but also mapping. This 1:1 correspondence is problematic for site-local (and presumably organization-local) multicast addresses given RFC 3879 which formally deprecated unicast site-local addresses (FEC0::/10). To what type of unicast source address will the algorithm map site-local (0x5) and organization-local (0x8) now that unicast site-locals are deprecated? The last sentence in RFC 3484, page 7, section 3.1 says: “This mapping implicitly conflates unicast site boundaries and multicast site boundaries.” This part of the source address selection algorithm is fundamentally broken (for ‘new’ implementations) given RFC 3879. Similarly broken would be the part of the destination address selection algorithm that assigns a site-local scope to the RFC 1918 addresses.
Another reason for the update of RFC 3484 would seem to be that RFC 2373, 2462, 3041 (all normative references for 3484) are in the process of being updated themselves (with the very first one listed, “IP Version 6 Addressing Architecture” being in its second revision since RFC 3484 was published (the first of course being RFC 3513)).
Will RFC 3484 be updated? If ‘yes’, when? If not, why not?
Best regards,
Tim Enos
1Sam16:7
-------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
