- From: Axel Polleres <axel.polleres@deri.org>
- Date: Sat, 11 Feb 2012 20:02:08 +0100
- To: "Gregory Williams" <greg@evilfunhouse.com>
- Cc: "Lee Feigenbaum" <lee@thefigtrees.net>, "SPARQL Working Group" <public-rdf-dawg@w3.org>
On 11 Feb 2012, at 19:21, Gregory Williams wrote: > On Feb 11, 2012, at 1:14 PM, Axel Polleres wrote: > > > To make this more tangible, what I was thinking is the following: > > > > Take > > > > Q: SELECT ?S WHERE { ?s :p/:q [] } > > > > and the following graph > > > > :s1 :p [ :q :o1 ]. > > :s1 :p [ :q :o1 ]. > > :s1 :p [ :q :o2 ]. > > I think this is a particularly strange query to use to base this discussion, because: > > a) (as Andy noted) things get particularly strange when you don't have a term on one end of the path > b) it's a fixed length path, which isn't the underlying problem as I understand it. I certainly hope we're not discussing changing the cardinality semantics of fixed-length paths… Sure, the query was kept simple just to show the problem why I think that it's not trivial to find an equivalent way of expressing Jorge's semantics, So the path expression wasn't my point, but the blank node... I simply don't know a way how to write an equivalent DISTINCT subquery where these things wouldn't intermingle. So, I am unsure how exactly we "implement" Option 1: > Option 1... keep everything as it is in the grammar, and explain which DISTINCT path subqueries can be optimized What I am missing is an understanding/definition of which DISTINCT path subquerise could be optimized and how. Jorge's paper outlines how this can be done for their semantics, if we don't have a way to express their semantics, what are the queries that can be optimized and how? cheers, Axel > I'm having a hard time coming up with a real query like this that used arbitrary-length paths, that didn't involve counting, and that couldn't be done with an ASK query. > > .greg > >
Received on Saturday, 11 February 2012 19:02:39 UTC