W3C home > Mailing lists > Public > www-tag@w3.org > July 2003

[httpRange-14] each HTTP URI in RDF names a ResponsePoint

From: Sandro Hawke <sandro@w3.org>
Date: Mon, 21 Jul 2003 17:09:31 -0400
Message-Id: <200307212109.h6LL9WmO005789@roke.hawke.org>
To: pat hayes <phayes@ihmc.us>
cc: www-tag@w3.org


> >My current thinking is that HTTP URIs most directly denote
> >ResponsePoints [1].
> 
> That is the best idea Ive heard so far.  Is that consistent with 
> Tim's idea of an 'information resource'?

I think:

  sandro:ResponsePoint rdfs:subClassOf timbl:InformationResource.

So it's consistent but not synonymous.  My impression is that
Information Resource covers documents, books, libraries, databases,
etc, as well.

[ Digression over denotes/identifies/names moved into a different
thread [2] ]

How's this:

    Each (non-fragment) HTTP URI can be used as a name for a
    ResponsePoint.  (It can potentially also be used as a name
    for a document, and company, etc.)

... okay, obvious, so what, ... but I'll jump in and claim:

    Readers of RDF MAY assume the author intended each HTTP URIRef
    used as a name to name a ResponsePoint as long as (1) it is not a
    fragment URIRef and (2) a "200 OK" response may be obtained in a
    "GET" operation with the URIRef.  Because of this, authors of RDF
    SHOULD NOT use non-fragment HTTP URIs which answer "200 OK" to
    name anything other than a ResponsePoint.

There are several angles of argument about this.  I think the
ResponsePoint page [1] addresses why the assumed class should be
ResponsePoint instead of Document or anything else.   I could go into
the motivation for saying anything on this subject, ... but I'll skip
it for now. 

I realize I'm kind of ignoring WebArch here, framing things quite
differently from httpRange-14.  I'm trying to focus on the real
problem, which is how to use URIs effectively in KR languages.

The notion is that:
   1.  You use HTTP URIs to name ResponsePoints, which you often
       want to reason about
   2.  When you want to name something else and include as part of the
       name a reference to an associated authoritative ResponsePoint,
       you use indirection, and name it with either
            a.  an HTTP URI offering a "303 See Other" redirect, or 
            b.  a URI-References (with a hash) which has  a non-hash
	        part naming a ResponsePoint
   3.  You can use uuids, tag: URIs, and bNodes the rest of the time.

(There's a fuzziness here where I'm using "HTTP URI" to mean something
a little broader, including HTTPS at least, but probably not all
URIs.  I don't know the right term for this.   URL is awfully
tempting....) 

      -- sandro

[1] http://esw.w3.org/topic/ResponsePoint
[2] http://lists.w3.org/Archives/Public/www-tag/2003Jul/0262.html
Received on Monday, 21 July 2003 17:09:35 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 22:56:00 UTC