Comments on: The Sixteen Faces of Eve

By: Bill de hOra

Bill de hOra — Thu, 29 Sep 2005 19:19:15 +0000

Respectfully; asking what’s the XPath is the wrong question. It’s dealing with symptoms.

Jimmy: the “So what” is that RDF/XML is like a box of chocolates; you never know what you’re gonna get.I know of no systems that are end to end model driven; you always have to deal with serializations, mappings, validation, form filling, presentation. The XML syntax sufficiently screws up the economic proposition of RDF that it makes sense to look at alternatives – a case of snatching defeat from the jaws of victory. It’s a nutso situation.

By: Howard Katz

Howard Katz — Thu, 29 Sep 2005 17:21:32 +0000

What I’ve got *should* work for the case where foaf:name or @foaf:name and foaf:homepage are descendants (not necessarily direct children) of different elements. The key is lots of “//” descendant specifiers to indicate that *somewhere* back up the hierarchy there’s a common element for both end points (name and homepage). When I have some time later on I’ll try this on face 5 just to make sure …

By: iand

iand — Thu, 29 Sep 2005 05:59:16 +0000

Good try Howard. I read that XPath as:

Select the rdf:resource attribute of the foaf:homepage element that is a descendant of an element which has either a foaf:name child element or a foaf:name attribute with the value of “Eve”

I don’t think that covers the faces where there foaf:name and foaf:homepage are children of different elements e.g. face 5

By: Howard Katz

Howard Katz — Wed, 28 Sep 2005 21:37:31 +0000

I’ll take you up on your XPath challenge. I think (he says hopefully) that the following should work:

//*[ ( .//foaf:name | .//@foaf:name ) = "Eve" ]//foaf:homepage/@rdf:resource

This assumes that the top-level context is a single document containing all 16 variations.

By: Ontogon

Ontogon — Wed, 28 Sep 2005 02:48:47 +0000

RDF/XML Sucks x 16 32

Caution. Rant ahead… I’ve always thought that the XML serialization of RDF is pretty much garbage. Sorry, had to say…

By: iand

iand — Tue, 27 Sep 2005 22:26:40 +0000

Hey I never knew that you could use rdf:type as an attribute in that way, but it’s legal. That’s another 10 variations on top of the 4 suggested by Ben. So, we’re up to 32 different ways to write 3 triples in RDF/XML

By: Jimmy Cerra

Jimmy Cerra — Tue, 27 Sep 2005 21:43:24 +0000

P.S. Re: 6. rdf:type is an exception for that rule, in that it is allowed as an attribute but still points to a URIRef and not a plain literal.

By: Jimmy Cerra

Jimmy Cerra — Tue, 27 Sep 2005 21:38:41 +0000

I don’t see much wrong with the general ideas behind RDF/XML except that it is the only official format (although there are lots of flaws with the details). The idea behind RDF/XML is not that you use XML tools to read RDF, but that you use RDF tools to read XML. I.E. You mark up an existing XML document with RDF/XML attributes indicating its semantic structure (as opposed to the logical structure).

By: iand

iand — Tue, 27 Sep 2005 21:20:15 +0000

I don’t think so, attributes can only take literal values.

By: Damian

Damian — Tue, 27 Sep 2005 20:54:00 +0000

You missed another variation: rdf:type can be an attribute. I think that doubles Eve’s faces.