- From: Dan Connolly <connolly@w3.org>
- Date: Mon, 04 Jun 2001 18:02:01 -0500
- To: Aaron Swartz <aswartz@upclink.com>
- CC: w3c rdfcore wg <w3c-rdfcore-wg@w3.org>
Aaron Swartz wrote: > > http://www.w3.org/2000/03/rdf-tracking/#rdfms-difference-between-ID-and-abou > t > > What is the difference between using and ID attribute to 'create' a new > resource and an about attribute to refer to it? As an aside, the DAML spec takes the approach that rdf:ID="foo" is exactly the same as rdf:about="#foo" I think that's worth saying in the spec. (perhaps we can say that you can't have the same ID attribute value on two different rdf:ID elements; easy to do with XPath/schematron/xml-schema. It might help catch some errors.) [...] > ** The heart of the issue: > > - ID claims to "create" a new resource, but it is assumed by some that > referring to a resource implies its existence.[1] > > - ID claims to define an anchor ID, but it is not defined as an XML ID (and > possibly cannot, because of interaction with bagID[2]) and the editor of the > spec feels that it is not usable as such[3]. > > - While ID is in the syntax, it is not specified in the model, which means > that an RDF document cannot be properly "round-tripped" which I feel is an > important goal. I don't understand why any of these is a problem that needs solving. How about a test case to show why it matters? At what level do you expect to "round-trip" RDF? I expect it at the level of n-triples. rdf:ID is indistinguishable from rdf:about at that level. re creating resources: that's an editorial nit. Resources just are. re anchor ID: how is XML ID-ness relevant? anchor IDs are a MIME type thingy, not an XML thingy. > ** Proposals > > 1) Remove/Deprecate ID from the syntax. > > The Working Group could decide to remove ID from the syntax due to > incompatibility/mistaken compatibility with XML IDs. What incompatibility? > Cons: Breaks backwards compatibility. > > 2) Allow ID to create triples in the model. > > Sergey Melnik has suggested[4] that ID create an isDefinedBy triple. > > Cons: Requires processors to change their interpretation. > > I am willing to hear other proposals, but personally, I currently agree with > proposal 2. I'm not against proposal 2, but I find it hard to justify as a correction to the spec. It also creates a dependency from RDF M&S to RDFS, where there is none now. -- Dan Connolly, W3C http://www.w3.org/People/Connolly/
Received on Monday, 4 June 2001 19:02:11 UTC