Re: No "TOP" of the docuverse [Was: URC usage scenarios ]

Owen Rees (rtor@ansa.co.uk)
Tue, 04 Oct 1994 13:10:43 BST

Message-Id: <9410041210.AA16411@plato.ansa.co.uk>
To: uri@bunyip.com
Subject: Re: No "TOP" of the docuverse [Was: URC usage scenarios ]
In-Reply-To: Message from avatar@jolt.mpx.com.au of Tue, 04 Oct 1994 08:34:43
<m0qrvxz-0005tkC@jolt.mpx.com.au>
Date: Tue, 04 Oct 1994 13:10:43 BST
From: Owen Rees <rtor@ansa.co.uk>

avatar@jolt.mpx.com.au (Andrew Pam) writes:
> My proposal is to make the URNs relative to the current server at all times.
> Each host can follow a straightforward algorithm to make each URN relative
> to itself whenever one is received, requiring only the name of the host
> from which each URN was transferred. [...]

This sort of totally context-relative strategy works fine for names that can
only ever be transmitted through an infrastructure that does all the necessary
name generation and resolution. It does not satisfy the "bar napkin"
requirement; you cannot write those sort of names on a bar napkin, or print
them in the newspaper and have people type them in so that they refer to the
intended resource. As I understand it, this is the intent behind the "Global
scope" and "Global uniqueness" functional requirements, and also some of the
encoding requirements.

The "bar napkin" requirement is such an important one that we have to find a
good enough approximation to global naming in this case, despite the strength
of the arguments against global naming in general. (The ANSA Naming Model
which I wrote about yesterday includes an explanation of why global naming
fails.)

Regards,
Owen Rees <rtor@ansa.co.uk>
Information about ANSA is at <URL:http://www.ansa.co.uk/>.