What a URI reference identifies

It is fair to say that there is no consensus.

Although RFC 2396 supposedly folded URLs and URNs into URIs, it is quite
apparent that there are at least two communities of people who use URIs and
URIreferences in two different ways. Each community has strong opinions
about what a URI ought identify and what a URI reference ought identify, yet
I hear no compelling argument to suade me either way.

Perhaps we should state the facts and move on. I propose something along the
lines of:

"
The use and meaning of URIs and URI references is application dependent.
URIs and URI references are generally used in two different fashions. When a
representation is returned on dereferencing a URI, a fragment identifier
locates a particular part of the resource representation. In this model, the
resource identified by a URI might be anything with identity, but the
fragment identifier references a piece of the resource representation. The
URI is transmitted to the server for resolution, the fragment identifier is
used solely by the user-agent.

In other applications, URIs tend not to be dereferenced, used rather as
names, or identifiers for resources. In such cases, URI references are also
used as opaque names for objects in the domain of discourse.

HTML/HTTP is a canonical example of the first form  of URI reference use.
RDF is a canonical example of the second form of URI reference use.
"

Jonathan

Received on Wednesday, 27 March 2002 15:09:18 UTC