>>>>> On Mon, 28 Mar 2005 18:20:19 -0800, >>>>> Bill Fenner <[email protected]> said:
> 0. Should we solve this problem at all? > The problem is of reaching [fe80::cafe:f00d] via a URI from a > system attached to multiple links. (note that loopback counts > as a link on some implementations.) The URI literal format (RFC > 2732) didn't address this issue, so when RFC 3986 integrated this > format it didn't address it either. The current proposal is to > use the expansion literal format from RFC 3986, to allow a URI like > http://[v6.fe80::cafe:f00d_de0]/ . > If the answer is yes, then the question of a delimiter comes up. > Percent, as the scoping architecture uses, is problematic because > percent is such a special character in URIs. I've not yet convinced that we really need usage like http://[v6.fe80::cafe:f00d_de0] or http://[v6.fe80::cafe:f00d%de0] or .. in a reasonable scenario. If an SNMP example that Margaret mentioned the other day is (one of) the intended case, could someone provide a more detailed scenario so that we can be sure about the feasibility? The scenario would include network topology, the relationship between implementation components (e.g. SNMP software), and specific examples of the usage (e.g, sample command-line usage, data flow, etc)? Meanwhile, regarding the point that the loopback interface can make many nodes multi-link, RFC4007 mentions a possible solution: if the node has only one physical network interface other than the loopback interface, we can avoid explicitlty specifying the link by introducing the notion of the "default link". JINMEI, Tatuya Communication Platform Lab. Corporate R&D Center, Toshiba Corp. [EMAIL PROTECTED] -------------------------------------------------------------------- IETF IPv6 working group mailing list [email protected] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
