Some comments inline. On Tue, Oct 12, 2010 at 8:40 AM, Brian Rosen <[email protected]> wrote: >>> Section 12.3 >>> Something I am clueless on. The user name looks like a FQDN, like >> [email protected]. If it looks like a FQDN, can one resolve >> the address from outside the peer network? If one can, how? If one >> cannot, why do we have a string of characters that looks like a >> FQDN? This is a clarifying question, in that I can get it, but my >> brain hurts. >> >> Hmm - not sure what to do. The FQDN has to do with how the namespace is >> allocated not if it is resolvable for or not. Even with email, >> [email protected] is not going to >> be resolvable outside the split DNS. I think the important things is .net >> allocated example to someone that allocated dht to someone that allocated >> alice to someone to form a unique name. There e is also the question about >> what resolution means related to a separate conversation on if the reload >> URI should have a // in it or not. The question comes down to does resolve >> mean resolve using DNS or resolve using some combination of protocols >> including DNS and RELOAD. Anyway - I have no idea what we should do here. >> If you think some text change is needed ... let me know. > <as individual> > What we want is a unique userid. We don't need an FQDN. Would a URN be more > appropriate? > We could even create a namespace for a URN that allowed a "username" and a > "domain name" as components: > urn:reload:userid:dht.example.net:alice > > Brian
I don't think you really want URN, which requires a single namespace authority for the whole namespace (the UUID URN being the one counter example permitted so far). A URI scheme for it would be easier. That would require adjusting the format to include the URI scheme as a prepend(it could re-use existing schemes that were used as an identifier or mint a new one); I think the same should be done with the node-ids, frankly, as they otherwise are fairly opaque values. The node-id mechanism in the now expired p2p-overlay draft described one way to do that, but there are others. Just my two cents, regards, Ted _______________________________________________ P2PSIP mailing list [email protected] https://www.ietf.org/mailman/listinfo/p2psip
