Minutes of the URI WG meeting of 18 July, 1995

  I mis-addressed my first draft of these minutes, so here's the next
revision.  Sorry for the delay.

I.  SUMMARY OF THE MEETING

A.  The minutes of last meeting were approved without dissent.

B.  The majority of the meeting was devoted to reviews of material
provided electronically to the Working Group.  Specifics of these
activities may be found in the e-mail and internet drafts which were the
basis of the presentations, though some are captured later in these
notes.

1.  Leslie Daigle presented an overview of the SILK-based URA and
metadata resolver developed at Bunyip.

2.  Karen Sollins summarized her perspectives on URN resolution
standards and services.

3.  Keith Moore reviewed the key aspects of the resolution server
developed at the University of Tennessee, Knoxville.

4.  Dirk van Gulik highlighted the Harvester search and resolver tool
developed at the Center for Earth Observation.

5.  There was a brief question and answer period as follow up to these
presentations, which included discussion on the distinction between URLs
and meta-data in general, and whether meta-data or actual documents
should be returned by a resolver or search engine.  The issue of
punctuation and character sets was raised but not pursued.  Pointers to
electronic sources of information and experimental tools were given.

6.  Ron Daniels provided a brief update on the URC drafts and work
needed to complete them.

7.  Related IETF sessions were highlighted:  the "Read the Label BOF on
Wednesday afternoon, the unofficial Tuesday afternoon BOF on meta-data,
and the Tuesday evening MIME registration session.


B.  The remainder of the meeting was devoted to a discussion of the
Working Group's Charter.

1.  John Klensin started this discussion, as Area Director responsible.
He noted the perceived lack of productivity of the WG, its inability to
define its objectives, lack of consensus and difficulty in determining
consensus, and its tendency to "solve" problems by creating yet another
new type of UR identifier to be defined, all of which result in long-
term drift that must be stopped.  He suggested that the WG must
reaorganize itself, use small-group interactions to achieve focus, and
find appropriate lines to separate activities rather than reasons to add
new work to the WG.  Discussion was reasonably lively and included
concerns about coordination among multiple groups working on the various
activities that relate to UR*'s, the claim that there's working code and
rough consensus on URAs by those that need them, and a reminder that the
original URI charter was deliberately written very broadly.

2.  This was followed by a charter review presented by Leslie Daigle and
based on her proposals distributed electronically to the WG.  This began
with a quick review of the text defining the work of the WG, which, by
common consent, was not addressed in discussions in order to focus on
the Goals and Milestones section.  There was considerable discussion of
many aspects of the proposed Goals and Milestones, with the following
actions resulting therefrom:

a.  WG action on the URN syntax drafts should be moved ahead and
accomplished by (1) an interim report comparing the various URN
proposals (based on work already distributed to the WG) and recomending
action, with either a proposed standard for a unified, general URN
syntax to be adopted by the WG at the October IETF or agreement to allow
separate, competing proposals for URNs to proceed independently.  [WILL
THIS INCLUDE DEFINING THE MECHANISM FOR URN NAME ASSIGNMENT?]

b.  A similar approach should be followed for URCs, though specifics
were not determined.

c.  Revision of RFC 1738 will be postponed until after the WG action on
URN syntax.  (A new date was not agreed upon for this.)

d.  Revision of RFC 1736 will not be done by this WG.  Instead, a new WG
to be formed specifically for the purpose of determining procedures for
IANA acceptance of new URNs for registration.  This WG will be tasked to
revise RFC 1736 or create a new document for that purpose, and to revise
RFC 1737 only if that appears necessary in order to accomplish its task.

e.  The WG needs to reach consensus on the finger and mailserver URLs in
order for them to move forward.  This should be done quickly.

f.  Leslie Daigle, Chris Weider, and [WHO?] agreed to complete a draft
informational RFC on Uniform Resource Architecture for discussion in
December 1995.

g.  Klensin announced that an additional URI session could be scheduled
for Thursday afternoon and used to continue the revision of the WG
charter and possibly action on the finger and mailserver URL proposals.
(It was thereafter announced as "UR- Next generation" BOF, 1300-1500,
Carl Larsson Room.)


II.  ADDITIONAL INFORMATION ABOUT THE PRESENTATIONS


    Leslie's presentation on the SILK-based URA resolver noted that it
is designed to be used to create sophisticated user decision-making and
information presentation through use of the meta-data provided by URAs.
It is intended to simplify the invocation of URAs, to create new objects
through individually specified processing of meta-information and to
access multiple URNs in order to obtain the meta-data needed to satisfy
the users' needs, thereby going beyond the usual interaction of a client
with a single server.  It allows HTML display of information to be
bypassed in preference to a user's specifications.


    Karen reviewed the ideas for "URN Resolution Standards" from her
Internet draft, including extremes & intermediate positions about what
to standardize regarding name assignment and resolution.  She noted that
resolution services must be designed to accomodate a wide variation in
requirements (locality vs ubiquity, wide variations in the speed
required for resolution, etc.) depending on the nature of the data or
other object being retrieved, the availability and the volitility of the
information being located, policy controls, pricing, etc.  Karen
suggested that client development repesents the major cost for retrieval
services and that clients need to be able to use a large variety of
services and servers, which implies the need for stabiltiy in the client
interface to resolution services, modularity, and independence of
service implementation.  Karen concludes from these ideas that we
should:
    standardize on client-service protocols
    standardize on the form of URNs (needed quickly!)
    standardize on how client/application might learn about
       resolution service suggestions - meta-information
    not standardize on a single inter-server protocol (but perhaps
       write several server-server protocols, each of which might
       become standardized)


    Keith discussed the resolution server, "SONAR" server, which point
to "best" resolver, and Web client which he and his colleagues are
developing.  They use LIFNs, make relative URLs work correctly, are
fast, and are available now for experimentation.  See:
    http://mobile.netlib.org


    Dirk reviewed "Harvester", a "search and choice" tool.  He noted
that, from the perspective of Harvester, it doesn't really matter what
URNs are, they're just like a handle to point to metadata, unique keys
into databases.  His services are based on negotiation between client
and resolution service.  The URN looks  like the X-DNS URN.  Relevant
URLs for this work include:
    http://elect6.jrc.it/~dirux/alibrooker.html
    http://ewse.jrc.it/.
    http://www.ceo.org/.
    x-dns-2 on port 4500 @ elect6.jrc.it

    Note from the scribe:  These minutes are probably rendered irrelevant
by John Klensin's subsequent action to terminate the WG.

The scribe for the WG session was Jim Conklin.

Received on Friday, 21 July 1995 04:16:55 UTC