- From: David Menendez <zednenem@psualum.com>
- Date: Mon, 20 Sep 2004 20:12:40 -0400
- To: Daniel O'Connor <daniel.oconnor@gmail.com>
- Cc: www-rdf-interest@w3.org
Daniel O'Connor writes:
>
> Mmm, I stumbled across this in my internet travels today:
>
> http://www.w3.org/DesignIssues/HTTP-URI.html
>
> And thought it was of some relevance.
It's funny. I agree with most of TBL's premises, but I reach the
opposite conclusion: I side with Roy Fielding and the "URIs can identify
anything" camp.
I think the problem with Tim's argument is that he assumes "URIs can
identify abstract things (not web pages)" is the same as "all URIs
identify abstract things (not web pages)". As I see it, some URIs
identify web pages and other identify abstract, non-web-page things.
The key point, for me, is that web pages are *themselves* abstract
things. Every time you dereference <http://www.cnn.com/>, for example,
you will get a different HTML document, but they all represent the same
thing.
You can ask, "Does <http://www.cnn.com/> identify 'CNN' or 'the front
page of CNN'?" The truth is we don't know. CNN controls the URI, and so
far as I know they haven't come down on one side or another. That just
means that we shouldn't use <http://www.cnn.com/> in our RDF data,
because we don't know what it means.
On the other hand, if we want to talk about a particular HTML document
we obtained by dereferencing <http://www.cnn.com/>, then we need to say
something like this:
[ a ex:HTMLDocument ] ex:obtainedFrom "http://www.cnn.com/".
The fun part[1] is that the document is itself a resource and could be
given its own URI (perhaps <http://example.com/pagesIveDownloaded/12345>
or <cid:123456@example.com>).
[1]. For certain definitions of "fun".
--
David Menendez <zednenem@psualum.com> <http://www.eyrie.org/~zednenem/>
Received on Tuesday, 21 September 2004 00:12:44 UTC