Minutes: Washington

[Alan Emtage <[e-mail]>]

Hello all,
	I've finally compiled the minutes from the meeting in Washington
and they are included below. Also, the "old" ietf-url list that Chris
Weider was maintaining at Merit has been merged with the [e-mail]
list (one came out of the BOF one out of the WG). I have tried to purge
the duplicates but with a list that size it can be a messy business. This
message is only being sent out to the uri list and so if you receive
duplicates it's because you are listed twice here... please drop me a
line and I'll fix it. Additionally, some people may have asked to be
dropped from the uri list not realizing that they were still on the
ietf-url list. If you don't want to be on the new purged one just write
and tell me. 

Please note that these minutes are a bit more verbose than normal due to
the interest that was expressed before the meeting by people who couldn't
be there. My thanks to George Brett for helping me out with this.


-Alan

--------------------

IETF             Uniform Resource Identifiers(URI)           		Minutes



Co-Chairs:  Alan Emtage/Bunyip  ([e-mail])
	    Jim Fullton/CNIDR   ([e-mail])


1) The agenda for the first meeting of the URI WG was approved.

2) The charter for the group was reviewed and approved. It was noted that
   the "Goals and Milestones" may need to be changed in the future
   depending on the progress in this very new area.

3) Peter Deutsch/Bunyip ([e-mail]) who was initially named to
   co-chair the group resigned the position in order to follow a more
   activist role and avoid any potential conflict of interest. Jim
   Fullton/CNIDR was installed as new co-chair. However before stepping
   down Peter took the opportunity to make a few personal observations
   and commitments:

   	- Peter has offered to co-author an overview paper along with
	  Chris Weider ([e-mail]) This paper would propose a
	  possible architecture to the group describing the use and the
	  form of the various Uniform Resource objects such as URI's
	  (Uniform Resource Identifiers), URL's (Uniform Resource
	  Locators) and URSN (Uniform Resource Serial Numbers) and how
	  the would interoperate.

	- Peter gave a basic overview of his ideas about what the UR
	  objects looked liked. By his definitions:

	  - A URL identifies a particular object on the network and is
	    composed of a named scheme (Eg, FTP, WAIS, Gopher) and
	    information specific to that scheme. It was noted that this
	    idea already exists in a similar form in the World Wide Web
	    (WWW) system, and has been codified in a paper by Tim
	    Berners-Lee/CERN ([e-mail])

	  - A URSN can be broken down into a "virtual user" and an actual
	    serial number. Related topics were the issue of the
	    "producer" of an network object and the "owner"; some
	    possible schemes for implementation of the virtual user
	    (whois++ handle, X.500) ; and what the serial number would
	    looked like (possibly an MD5 checksum and other methods).

   It was decided in the interests of time that further discussions
   should be carried out on the mailing list.


4) The paper currently titled "Universal Resource Locators" by Tim
   Berners-Lee was reviewed and the following comments were made:

   - The use of the term "protocol" in the document is ambiguous given
     the context of the IETF and should be replaced or more specifically
     defined.

   - The use of the term "name" was considered to be unclear and again
     should be clarified. It was suggested that it be removed and another
     term used in its place.

   - The document should be written as a "standalone" unit. However, the
     objects described therein should be viewed as part of a larger
     architecture and an explicit description of their purpose should be
     added. It was suggested that the document could be further
     generalized from a _perceived_ WWW bias.

   - The question of the "partial form" of the URL brought heated
     discussion between two factions: one which wanted the removal of the
     form altogether and one which suggested their continued existence
     with restrictions. Some consensus developed around the idea that
     partial forms could be used internally for individual information
     systems but should not be used when exchanged externally. It was
     decided that further discussion should occur on the mailing list.

   - Consensus was reached that the document should specifically state
     URLs are to be considered transient and should not be used in static
     objects (hardcopy documents etc). Their use as references should be
     specifically discouraged. Such references was considered to be in
     the domain of the URSN, whatever they ultimately look like.


   - The paper should describe the general scheme being proposed without
     reference to particular systems (other than as examples). All
     detailed descriptions of individual systems should be put in an
     appendix. It was decided that the most likely repository for the
     individual definitions would ultimately be the Internet Assigned
     Numbers Authority (IANA) but that the original document may propose
     the definitions for a basic range of services (such as FTP).

   - It was suggested by Thomas Hacker/UMich ([e-mail]) that
     to the OSF DCE DFS (Open Software Foundation Distributed Computing
     Environment Distributed File System). 


   - Mitra/Pandora ([e-mail]) proposed a "fragment
     specifier" scheme to be incorporated into the URL document. It was
     decided that detailed discussion of this was best left to the
     mailing list.

   - Other points were:

   	- Some of the text and examples did not agree
	- The use of percentage signs should be reviewed on the mailing
	  list
	- Use of blank characters was again questioned.

     All were referred back to the mailing list for further discussion


5) A discussion about URI's followed. The questions that were raised were:

	- Given the current definitions what _exactly_ does URI mean? 

	   - Alan Emtage suggested that they may be defined as

	   	URI = URL + URSN + "Uniform Resource Representator" (URR)

	     since the current definitions of URL and URSN do not give
	     sufficient information for a user/client to determine if in
	     fact the information available is useful and that such
	     things as filename extensions are not a reliable method of
	     determining content format (and in the case of processes is
	     meaningless). However he declined to be committed on what
	     exactly these URR's would look like.

	   - It was suggested that the concept of the "URI" may be defunct
	     now since it as been decomposed into several constituent
	     parts.

	- The proposal that John Kunze/UCBerkeley ([e-mail])
	  had made on the mailing list previously was briefly discussed
	  and it was suggested that he and Clifford Lynch/UC
	  ([e-mail]) co-author an alternate document to that
	  produced by Peter Deutsch and Chris Weider, more from the
	  perspective of the library community. John's proposal for
	  access lists, descriptive fields, functional types and a "UR
	  Citation" were suggested as being better handled in detail on
	  the mailing list.

	- In addition to the document describing the general UR system,
	  Peter Deutsch and Chris Weider have agreed to co-author a paper
	  proposing the structure of URSN's.