Date: Mon, 17 May 93 16:31:00 CDT
From: brennan@hal.com (Dave Brennan)
Message-Id: <9305172131.AA25350@hysteria.hal.com>
To: uri@bunyip.com
Subject: Re: Internet Draft on URNs
Here's some additional comments that elaborate a bit more on my
previous mail. I wrote it in response to some private mail, but
seems worthwhile to mail out.
This is pretty much in response to the comments that URNs need
additional information that humans can understand.
What this comes down to is a fundamental question over what the purpose
of a URN is. As I mentioned in my message to the list, I feel a URN
should just be a unique document identifier. Period. The advantages
to this approach are:
1) URNs will tend to be relatively compact. This means that:
a) They can be shipped around with ease, and
b) indexed and resolved with a low overhead.
2) It places no restrictions on how we provide the kind of information
that humans are interested in (like title, author, etc). This might
be handled by a URI (if they ever get defined), with a standard
bibliographic reference, or perhaps even by querying an information
server with the URN.
The "off-line" reader example is a good one to consider. This problem
is easily solved if people get into a habit of shipping around meaningful
information with their URNs. The same is true of ISBN numbers. No one
who wants to be useful gives someone an ISBN number without some kind
of context. Conversely, if I know some general information about a
resource it should be possible to dig up a URN through some sort of
query (see below).
Once again, I don't think human readable should be a design goal for URNs.
There's better ways of obtaining information that's meaningful for a human.
Also, I definitely don't want humans trying to manually trying to construct
a URN.
It was suggested to me that it might be good if humans could re-construct
URNs. The real reason a human might try to construct a URN would be as an
attempt to locate some information.
This is where I think the separation is important. If I already know
exactly what I want because I have a valid URN, search functionality
is of no consequence and I don't want to be bogged down or bothered
with it. If do want to search for something, there's better methods
than taking a shot in the dark at creating a URN.
Like many other people, my first impression was: "What good is a URN if I
don't know what it refers to." After giving this issue much though in the
context of existing Internet services and the information browser that I'm
currently implementing, I came to these conclusions.
It's still an open question in my mind as to whether URN resolution and
queries to turn up information about a URN or a URN from information would
be handled by the same kind of resolver. In either case it's critical
(as far as I'm concerned) that information related to URNs is kept as
up-to-date as possible.
Ok, now I'll shut up too. :-)
-- Dave Brennan HaL Computer Systems brennan@hal.com (512) 794-2855