- From: Perry A. Caro <caro@Adobe.COM>
- Date: Wed, 03 May 2000 09:52:43 -0700
- To: www-rdf-interest@w3.org
> <?xml version="1.0" encoding="UTF-8"?> > > <ImageDataset> > <ImageClass> > <ImageClass> <!--1--> > <Image name="foo"/> > </ImageClass> > <ImageClass> <!--2--> > <Image name="bar"/> > </ImageClass> > </ImageClass> > </ImageDataset> I agree that this is not well-formed. Production [6.17] says "obj", not "obj*". > <ImageDataset> > <ImageClass> > <ImageClass> <!--1--> > <Image name="foo"/> > </ImageClass> > </ImageClass> > </ImageDataset> > > Now the parser agrees, producing: > > {'Description__1', rdf:type, 'ImageDataset'} > {'Description__1', 'ImageClass', 'Description__2'} > {'Description__2', rdf:type, 'ImageClass'} > {'Description__2', 'Image', 'Description__3'} > {'Description__3', name, literal(foo)} I believe this is correct. When in doubt, I rewrite typed nodes as their basic non-abbreviated syntax equivalents: <rdf:Description rdf:type="ImageDataset"> <ImageClass> <rdf:Description rdf:type="ImageClass"> <Image> <rdf:Description> <name>foo</name> </rdf:Description> </Image> </rdf:Description> </ImageClass> </rdf:Description> Your parser should produce the same triples (apart from anonymous ID generation) for both serializations. Actually, I like your {type}_N convention for anonymous ids, so in your case, the triples should be exactly identical. Perry
Received on Wednesday, 3 May 2000 12:54:08 UTC