Re: Requirements for a possible "RDF 2.0"

On Fri, 2010-01-15 at 22:06 +0000, Steve Harris wrote:
> There is a cost to opening up what can go where in triples. My SPARQL
> engines (and I expect those of others) use this as a source of  
> optimisations. E.g. if some variable appears in the predicate slot,  
> then you know it can only bind to a URI, and you can limit your search
> space. 

I seem to remember that in Virtuoso TP1, the following query:

	SELECT *
	WHERE { 
		?s ?p ?o .
		OPTIONAL { ?o ?q ?v . }
	}

"optimised" things by making the assumption that ?o was not a literal.

:-(

-- 
Toby A Inkster
<mailto:mail@tobyinkster.co.uk>
<http://tobyinkster.co.uk>