- From: Jacek Kopecky <jacek.kopecky@deri.org>
- Date: Tue, 10 Jul 2007 13:46:49 +0200
- To: Tim Berners-Lee <timbl@w3.org>
- Cc: Keith Alexander <k.j.w.alexander@gmail.com>, Danny Ayers <danny.ayers@gmail.com>, al@jku.at, semantic-web@w3.org
Dear Tim, I could see a number of reasons for combining the variable-binding and result-forming functionalities between XQuery and SPARQL: 1) creating XML results from RDF sources 2) creating RDF results from XML sources 3) creating XML results from combined RDF and XML sources 4) creating RDF results from combined RDF and XML sources I was only discussing 1) before in this thread. 2) is easy with XSLT. I was proposing a recursive SPARQL CONSTRUCTXML for 1). Combining RDF and XML sources for either 3) or 4) could be nifty, but I can't see a good justification for the effort to create such a beast yet. I don't think it would be the easiest way to achieve 1). 8-) Anyway, I feel like I must be misunderstanding something in your email. Best regards, Jacek On Wed, 2007-07-04 at 12:28 -0400, Tim Berners-Lee wrote: > It seems to me that we have two query clauses > > - FOR ... LET .. WHERE from XQuery and > > - WHERE .. FILTER ... from SPARQL > > These basically generate bindings of named variables to things with > XSD data types. We now have various ways of consuming these bindings: > > - RETURN from XQuery, and > > - CONSTRUCT or SELECT or DESCRIBE (or INSERT or DELETE ... oops not > yet :-) from SPARQL. > > If I were making some mamoth software product I might just throw > those together as they seem to be, in a mixed XML/RDF world, or on > the XML edges of an all-RDF world, useful in an combination. Would > the SPARAXQL spec (SPARX for short?) be obvious combination of the > two specs? The transitions between ordered and unordered should work > out. > > The frightening thing would be the test suite! > > This seems to be a good idea against my better judgement. My better > judgement says one should keep languages simple and clean and avoid > RDF systems having to know about the XML data model and vice-versa. > > Tim > > On 2007-07 -04, at 10:01, Jacek Kopecky wrote: > > > > > Keith, yes, SPARQL CONSTRUCTXML would be very similar to XQuery, but > > using an RDF graph matching instead of XPath expressions. [...] > >
Received on Tuesday, 10 July 2007 11:47:22 UTC