Re: Registrar

Larry Masinter <masinter@parc.xerox.com>
To: e-krol@uiuc.edu
Cc: raisch@ora.com, rodgers@billings.ora.com, krol@ux1.cso.uiuc.edu,
        timbl@nxoc01.cern.ch, ccoprmm@oit.gatech.edu, terry@ora.com,
        uri@bunyip.com, www-talk@nxoc01.cern.ch
In-Reply-To: e-krol@uiuc.edu's message of Wed, 14 Jul 1993 16:44:40 -0700 <93Jul14.164458pdt.2797@golden.parc.xerox.com>
Subject: Re: Registrar
From: Larry Masinter <masinter@parc.xerox.com>
Sender: Larry Masinter <masinter@parc.xerox.com>
Fake-Sender: masinter@parc.xerox.com
Message-id: <93Jul14.172502pdt.2791@golden.parc.xerox.com>
Date: 	Wed, 14 Jul 1993 17:24:49 PDT
Status: RO
To take your points in reverse order:

> I think we don't particularly care whether a URN points to a changing
> resource or a static resource. What we care about is that a person
> has some means to find out what a symbol is for.  The question
> is far bigger than static/dyanmic, draft/non-draft, all you want to do
> is know what you got.

Maybe I miss-parsed your sentence, but I think that the systems that
we build for caching, replication, distribution, location of URLs from
URNs will care quite a bit whether the URN is a changing resource or a
static one.

> Second, there has been some talk about delegation of registration 
> authority.  How are you going to get the registrar for a particular
> area to conform.  If McGraw-Hill decides to use their pool of URN's
> in a contrary way do you think you have the clout to say I'm sorry
> you can't publish.

URNs are not like ISBN numbers, in that we're talking about using them
in systems and protocols; if someone builds a FTP server that doesn't
really talk the FTP protocol, then people won't be able to use it
effectively. If McGraw-Hill were to proclaim that some of their
variable URNs were static, people who maintained caches of McGraw-Hill
documents would get false cache hits. The system wouldn't work.

Secondly, we *could* very well use some scheme for URNs that couldn't
be subverted, e.g., use a one-way hash of the original document data
as the immutable URN, or use a URL with a time-stamp, rather than some
more subjective measure of 'identity' for the implementation.