I just implemented the destination address ordering described in draft-ietf-ipngwg-default-addr-select-06.txt, and have another comment about the longest matching.
The draft has the following text: Rule 9: Use longest matching prefix. If CommonPrefixLen(DA, Source(DA)) > CommonPrefixLen(DB, Source(DB)), then sort DA before DB. Similarly, if CommonPrefixLen(DA, Source(DA)) < CommonPrefixLen(DB, Source(DB)), then sort DB before DA. It is not clear to me how we should compare the match length when the address family of A and the family of B. If we converted IPv4 addresses to IPv4-mapped IPv6 addresses before the comparison, and the policy table did not prefer a particular address family, then the combination of IPv4 src and dst would tend to match longer than the IPv6 src and dst (because the former would match at least 96 bits). In any event, I don't see it meaningful to compare match lengths between different families of addresses. So, I'd propose to revise the text like this: Rule 9: Use longest matching prefix. When the address families of DA and DB are same, if CommonPrefixLen(DA, Source(DA)) > CommonPrefixLen(DB, Source(DB)), then sort DA before DB. Similarly, if CommonPrefixLen(DA, Source(DA)) < CommonPrefixLen(DB, Source(DB)), then sort DB before DA. JINMEI, Tatuya Communication Platform Lab. Corporate R&D Center, Toshiba Corp. [EMAIL PROTECTED] -------------------------------------------------------------------- IETF IPng Working Group Mailing List IPng Home Page: http://playground.sun.com/ipng FTP archive: ftp://playground.sun.com/pub/ipng Direct all administrative requests to [EMAIL PROTECTED] --------------------------------------------------------------------