CURRENT MEETING REPORT


Minutes of the Uniform Resource Names BOF (urn)

Reported by Sally Hambridge, Intel Corporation


Chris Weider chaired the BOF, and asked Larry Masinter to open the 
session.  Larry gave a brief history of the URI Working Group and also 
listed some caveats about what the group should address and what it 
should not address.

Keith Moore gave a presentation on the University of Tennessee, 
Knoxville (UTK) meeting from October 30-31, 1995.  This meeting was 
attended by most of the implementors of URN schemes, and was an 
attempt to reach agreement on what URNs should be.  The Knoxville 
proposal called for a URN to be ÒURN:NA[:LUI],Ó where NA is the 
Naming Authority and LUI is Local Unique Identifier.  The Naming 
Authority is resolved to records which describe current resolution 
services for that Naming Authority.  The protocol used to resolve the 
name is independent of the name itself.

Subsequent discussion on the mailing list evolved this proposal to 
ÒURN:NID:NSS,Ó where NID is the Namespace Identifier and NSS is 
the Namespace Specific String.  The Namespace ID governs syntatic 
interpretation of the Namespace Specific String.  The NSS includes the 
Naming Authority, if available.  The NID[+NA] governs location of 
resolution services.  The protocol used to resolve the name is 
independent of the name itself.  This construct allows the  incorporation 
of different name structures.

Examples include:

o URN:ISBN:914256789x (where 9142 indicates a naming authority)
o URN:MID:199512020720@muddcs.cs.hmc.edu (where 
  muddcs.cs.hmc.edu indicates a naming authority)
o URN:INET:acme.com:welcome (where acme.com indicates a naming 
  authority)

Keith showed a concept diagram for resolution in which he proposed 
that a URN would fetch a description of the NSS in an NID registry.  If 
a NA exists, the resolution would lookup the NID+NA, identify 
resolution services, pass the URN to one (or more) and the URN 
resolution service(s) would return the result.  Without a NA, the 
resolution would look up the NID which would pass the URN to one (or 
more) resolution service(s) which would return the result.

The framework for URNs appears to be compliant with the 
requirements for URNs in RFC 1737.  It provides global scope, 
persistence, independence, and legacy support.  Participating 
Namespaces must provide global uniqueness and extensibility and may 
vary in how they provide persistence and scalability.

There was intense discussion of this proposal.  Bill Arms was hopeful, 
claiming that there is at least a basis for agreement.  Roy Fielding still 
has a real issue with the URN:, claiming it will not work.  We need 
something firm enough to code to but flexible  enough to allow change in 
the future.  Discussion continued to say that the proposal does not force 
agreement on names or retention (which the UTK group considered out-
of-scope), or on to what a URN resolves.

Keith reminded the group that we were not a working group and that to 
be one we needed a Charter, problem statements (what problem are we 
trying to solve), drafts, a chair, and a running code.  Chris Weider said 
the proposal allows all schemes to work within browsers, and allow 
widespread deployment.  Ron Daniel said the issue of scalable 
resolution was outside the bounds of naming.  He thought we should let 
people implement and compete.  Larry Masinter remarked that the 
word "authority" was charged, and that it could mean a body which 
assigns names OR one that was responsible for resolution.  Larry also 
brought up the issues which were in RFC 1737 not mentioned: Lexical 
Equivalents and Internationalization.  Chris W. asked the group if it 
was better to work on issues or get parties together to work them.  Peter 
Deutsch said it would be nice to focus on what we agreed on for a 
change.  Bill Arms said it would be useful to identify issues but not work 
on them in this meeting.  Chris then declared the meeting finished, and 
those who wished to work on issues could stay.