W3C home > Mailing lists > Public > www-rdf-interest@w3.org > December 1999

Re: I-D ACTION:draft-ietf-webdav-binding-protocol-02.txt (fwd)

From: Godfrey Rust <godfreyrust@dds.netkonect.co.uk>
Date: Tue, 28 Dec 1999 18:48:25 +0000
Message-Id: <3.0.6.32.19991228184825.007e1230@mail.netkonect.co.uk>
To: www-rdf-interest@w3.org
Pierre-Antoine CHAMPIN wrote

>the U in URI doesn't mean 'unique' but 'uniform'!
>A resource can be defined as anything which has AN identifier - it may be
more than one.

At 10:53 AM 12/28/99 +0100, 
>Godfrey Rust wrote:
>> If a resource can be defined as anything which has a URI (which seems to be
>> its only required characteristic) it seems to rather pull the rug away to
>> say that something which is defined only by its unique identity can have
>> multiple unique identities.

It's hardly worth pursuing, but as Eric Hellman has pointed out that the U
in URI means Universal and Pierre-Antoine that it means Uniform, I would be
happy to hear any other suggestions. As far as I know W3C has tended to use
"Universal" since Tim's early-90s RFC, and IETF uses "Uniform" in its wg,
and in practise it makes no difference, but I would welcome clarification
from those more steeped in its development than I.

I wasn't, in fact, ever suggesting that it did mean unique, but that the
characteristic of uniqueness in a URI (as opposed to its universality or
uniformity) may often be the only thing which distinguishes one resource
(and its attendant metadata) from another (i.e. x is a resource because it
is identified as a resource - an increasingly common occurrence in a
virtual world which allows more or less infinite granularity).

This has been a basic issue underlying some of the discussions surrounding
DOI metadata as it becomes commercially significant when distinguishing
entities for purposes of intellectual property rights in resources, many of
which are generic or abstract. Much of my own work is concerned with the
relationship between non-Web identifiers (of the ISxx variety and others)
and URL/URN/URI, and the principle that Dan suggests of a many-to-one
relationship  between URIs and a resource significantly complicates this,
and also probably makes URIs within RDF less useful as a focus for the
production of canonical metadata about resources and their attendant
rights. In other standardised areas of identification, multiple IDs of the
same type (as opposed to multiple IDs of different types like UPC and EAN)
generally only occur by error and are an irritant rather than a major
problem. It is clearly impossible to outlaw multiple URIs for any resource
(as it is in any other sphere of identification), so I am sure RDF has to
allow them, but recognising and dealing with duplication as an aberattion
and encouraging it as a principle are, I think, rather different things. 

Happy &  bugfree New Year to the list, by the way, and thanks for some
fascinating stuff.


Godfrey Rust
..........................................
Data Definitions
14 Gloucester Road
London W5 4JB
T (44) 181 579 8655
F (44) 181 579 0938
Mobile  07775 908398
Received on Tuesday, 28 December 1999 13:47:37 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:51:42 GMT