RE: Using fragment identifiers with URNs

Al Gilman wrote:
> Better quote.
> 
> [quote]
> 
>    The semantics of a fragment identifier is a property of the data
>    resulting from a retrieval action, regardless of the type 
> of URI used
>    in the reference.  Therefore, the format and interpretation of
>    fragment identifiers is dependent on the media type 
> [RFC2046] of the
>    retrieval result.  [snip]
> 
> [end quote]
> 
> I would recommend that you read this as substantiating Roy's 
> claim that the
> document asserts scheme independence for what it says about #fragment
> interpretation.  And URN vs. URL independence.

I'm aware of this paragraph, but it doesn't say what you claim it says.
In particular, it doesn't say that retrieval is always meaningful for
a URI scheme.  It just says how the meaning of a fragment identifier
is determined *after* a URI + fragment identifier has been used for a
retrieval operation.  If it were possible to have a URN scheme for
which retrieval was not meaningful (and I still haven't seen any
official document that answers this one way or another) then the
above paragraph wouldn't apply to URI references using that scheme
(and my interpretation would be that such URI references are not
meaningful).

> The upshot of looking at generic URN religion likewise should 
> be that "URNs on the whole don't tell you about the means of retrieval" 
> principle should also be interpreted as "URNs on the whole don't tell
> you about the feasibility of retrieval" as well.

I'd prefer the word "applicability" to "feasibility" as the latter
suggests that retrieval might be meaningful but just not possible with
current tools.

> 'Resources' are by common usage expected to be of some use.  The
> generic 'retrieval' notion is just whatever you have to do in order to
> be ready to use the Resource. This is not required to be universal, but
> it is just so endemic to the many ways of providing value added that it
> is assumed by default.
> A URN scheme that denies all possibility of retrieval is not 
> impossible by the URI generic rules, it is just bad market positioning of 
> technology, a "market loser in waiting."

In RDF schemas, resources can be abstract concepts, and their URIs are
names.
To use them, you just have to be able to refer to those names.  What is the
notion of retrieval here?  I don't think it's retrieving the definition of
the concepts because you don't have to do that to use them in an RDF
document, and besides, RDF Schema defines another mechanism for that:
rdfs:isDefinedBy.

> Don't look for a more computer-understandable prior notion 
> than the above description that 'resource' is something of potential 
> utility; and a URI as a string which assists authors in recording
> a reference to a resource and consumers of the recorded reference in
> isolating the indicated resource.  The only thing that _all_ URIs share
> is some rules about how you put characters in the string.  This
> discipline allows strings written in a distributed fashion to
> avoid accidental string-compare collisions.  End of URI-universal story.

Thanks for your comments.  So it sounds as if it *is* OK to define a URN
scheme that states specifically that retrieval is not meaningful.  In that
case, the URI reference syntax should not be used with URIs from that
scheme,
as they would be meaningless.  Would you agree with that?

- Stephen

Received on Thursday, 27 September 2001 17:47:16 UTC