>> >> The code below would be straightforward if the "/64" prefix were >> accepted by getaddrinfo. >> >> Besides, I don't think the textual representation should be defined to >> make only Basic Socket API functions straightforward. IMHO, the order >> should be logical on its own. The prefix is clearly more tightly >> related to the IP address (in6_addr) and the scope zone index is >> clearly (by virtual of the fact that scope_id is in sockaddr_in6) >> related to the transport address. So the textual representation as it >> stands now mixes IP address and Transport address information. >> >> --Stephen > > No they are both properties that can be associated with a > address. If anything the scope ties closer than the prefix > as a link-local address or prefix is meanless w/o the scope. I'm not sure what you mean by "address" above. My argument is that the prefix is associated with IP addresses (represented by struct in6_addr in the API) and the scope is associated with transport addresses (represented by struct sockaddr_in6 in the API). They are both "addresses", but they are different layer addresses. If I map what textual representation to what data structure it has effect on: [ipaddr%scope/prefix]:port is mapped into [in6_addr%sockaddr_in6/in6_addr]:sockaddr_in6 >From my view the the textual representation and the Basic socket API structures are inconsistent.
> > Mark >-- >Mark Andrews, ISC >1 Seymour St., Dundas Valley, NSW 2117, Australia >PHONE: +61 2 9871 4742 INTERNET: [EMAIL PROTECTED] -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------