- From: Roy T. Fielding <fielding@apache.org>
- Date: Fri, 4 Oct 2002 17:32:21 -0700
- To: "Simon St.Laurent" <simonstl@simonstl.com>
- Cc: www-tag@w3.org
> Processors for RDF statements and XML namespaces as well as humans > looking for information are quite likely to have to deal with these > problems, and treating them as mere representation issues isn't likely > to go over well with people who (knowingly or not) prefer Hume's valuing > of experience to Plato's valuing of abstract madness. I am not glossing over the issues at all. There are resources and there are representations, and any system that fails to recognize the difference is doomed. That does not prevent anyone from restricting themselves to identifiers with a specific meaning within specific uses, nor is their any confusion when I say that http://www.w3.org/1999/xhtml identifies the namespace and only the namespace. The fact that someone can also perform a GET on that URI and retrieve a web page says absolutely nothing about the identity of the resource unless the server chooses to tell me something about the resource via metadata or RDF. Performing a GET is a method on an abstract interface. A series of GETs over time may give an observer some hint as to the meaning of that resource, but the only way to know for sure is to obtain guarantees from the namespace owner. If someone makes RDF assertions about a resource, using a URI to target those assertions, then they are making assertions about the resource. Unless they specifically target the assertions to the representations of the resource over some time scale and/or request variance, they cannot assert anything about what a person might see when they do a GET. Even a resource that is defined to be static and have a single bit-sequence representation for all time must be described as such using an assertion that talks about the resource, its representation, and time as separate variables. Moreover, it cannot imply that it is talking about GET because there are dozens of other HTTP methods (or FTP commands, or Gopher queries, etc.) that one might want to make assertions about for that resource. > The disconnect between URI theory and practice seems itself insoluble. What I described is practice and is the only standardized theory that I know about. I see no disconnect there. The disconnect is with how people interpret RDF assertions on URI, and that doesn't appear to be a syntax issue but is rather a disagreement among various people in regard to what a URI identifies, and hence the definition of resource is the key to solving the disconnect. Cheers, Roy T. Fielding, Chief Scientist, Day Software (roy.fielding@day.com) <http://www.day.com/> Co-founder, The Apache Software Foundation (fielding@apache.org) <http://www.apache.org/>
Received on Friday, 4 October 2002 20:33:09 UTC