Date: Wed, 2 Mar 94 10:47:34 +0100
From: Tim Berners-Lee <timbl@ptpc00.cern.ch>
Message-Id: <9403020947.AA06251@ptpc00.cern.ch>
To: uri@bunyip.com
Subject: URN and DNS namespaces
John curran said a few weeks back,
> Implementation via DNS may be a major mistake... DNS administrative
> authority realms do not overlap with URN authoirty realms, and DNS is
> not exactly dynamic at this time.
If we propose using DNS for URLs then DNS administrative realms
will be URN realms. You mean that the current domain hierarchy is
not what you want for URNs in general. You suggest that ONLY
dns names ending with .URN should be allowed in URNs.
Now suppose that we do allow a subtree of the URN space to in
fact be the existing DNS space. This means that publishers
who already have a DNS domain could just use that. This is
in keeping with allowing different standarsd bodies different
subtrees. No suppose that when we actually do the URN
convertion we build in a special case that URNs of the form
urn:/internetdns/com/pubcom/mkt/98123/182346
instead of being resolved by
pubcom.com.internetdns.urn
would be resolved by
pubcom.com
as a special case which only applies when the DNS itself happens
to be used as the resolution mechanism for a name which is also
in theurn DNS naming subtree.
The abstract namespace should be kept separate from the resolution
mechanism, but that doesn't stop an existing DNS space being adopted
as part of the abstract URN tree.
This would also mean that we could actually try the mechanism out.
Tim BL