- From: Keith Moore <moore@cs.utk.edu>
- Date: Sun, 26 Nov 1995 11:35:06 -0500
- To: Jon Knight <J.P.Knight@lut.ac.uk>
- Cc: Keith Moore <moore@cs.utk.edu>, Daniel LaLiberte <liberte@ncsa.uiuc.edu>, fielding@avron.ics.uci.edu, uri@bunyip.com, urn@mordred.gatech.edu
> In a parallel discussion going on over in the IETF mailing list at the > moment, Paul Vixie and other DNS gurus have pointed out that the DNS is > not a true directory service. I think this is something that the URN design > should bear in mind. I can see two scenarios here: > > 1) You have a URN given to which you want to resolve to a single URC. > This is just like the DNS FQDN -> IP address mapping and so DNS may be > appropriate. Yes, this is the scenario for which I would use DNS. > 2) You want to know what URN for a certain document written by a > particular person that has been registered by a particular organisation. > Now you could start guessing the DNS-like elements in the URN and in the > early days when the namespace is small this will probably work most of the > time. If you're assigning URNs and you want them to be persistent over several decades, you don't organize your URNs as a hierarchy of human-meaningful elements. So DNS wouldn't even work in the short term for this case. Even a directory may not make an effective resource discovery tool for many subject areas. Effective resource discovery often requires tools which are tuned to a particular discipline or subject domain. It might turn out that the most important function that URNs provide is to serve as a substrate for good resource discovery tools. Keith
Received on Sunday, 26 November 1995 11:35:54 UTC