Date: Mon, 17 May 93 18:15:32 -0400
From: "Chris Weider" <clw@merit.edu>
Message-Id: <9305172215.AA27814@merit.edu>
To: uri@bunyip.com
Subject: Comments on ID on URNs
Hi gang:
This is sort of a grab-bag response to all the comments that have been
circulated on the URN draft, so I will be addressing a multitude of threads
at once...
1. Encoding metainformation in the URN
While I understand Ed Krol's desire to put some type of metainformation
into the URN, I think that the URN's functionality should be much like the
functionality of the ISBN number or other identifiers. I think that Ed's point
about needing some metainformation about a resource when you have the URN but
don't have connectivity is well taken. But I guess I don't see people using
*just* URNs in their citations. I think that we will develop a UR Citation
style that will contain the URN as one component. Some subset of the meta-
information about a given resource will be additionally encoded into the
citation, but not necessarily all of it. Besides, to be completely pragmatic
about it, if Naming Authorities get enough pressure to put metainfo into
their URNs, we haven't ruled out the possibility. I just despair of
getting agreement on all possible components of the metainformation in any
sort of realistic time frame. In addition, if we start off with the
assumption that some metainformation goes into the URN, then I think that
we have the distinct possibility of naming wars where a given Authority will
say 'well, Naming Authority X isn't putting in information that my clients
think is useful, so I'll just assign a new URN for that resource'. If this
happens, then we've lost much of what we wanted for the URNs. If we can assume
that any given citation will more information than just the URN, then I think
we've addressed both Ed's concerns and my desire to keep metainfo out of the
URN.
2: Terry's Citation Proposal
Once we get the functionality of the URN hammered out :^) I think that
the next thing the URI working group should tackle is the citations. We have
several proposals for this so far in addition to Terry's (John Kunze's springs
most immediately to mind), but there are several points I'd like to make about
Terry's proposal.
The first is that I think that the URN component of the citation needs to be
highly stressed. The URL gives us a 'where do I get it' functionality, but
the URLs are not expected to be valid at all times and places. That's what the
URN does. If a URL in a citation doesn't resolve to what you expect because
the resource has been moved, it's good to submit the URN for an update. Other
than that I think the proposal was excellent, general enough to be robust,
and can provide enough information to allow humans to make decisions.
3: General comments since I'm running out of time...
The reason we split up the URLs, the URNs and the Metainformation into three
separate parts is because they all have differing functionality. Terry mentions
this in his comments. I just got back from a session with Fred Swartz (an
officemate of mine) and we went 'round about whether or not the 'citation'
and the 'urn' could be the same. Well, my personal view is that the
identification and description functions are sufficiently different to
warrant separate mechanisms; this allows us to tailor the description of
a given resource to a given audience while preserving the underlying identity
of the resource. However, humans typically mix the 'identification' and the
'description' of an object that it may not be possible to separate them...
I'll have more comments tomorrow.... hope this wasn't too incoherent.
Chris Weider