RE: [TF-PP] Questions on Property Paths



> -----Original Message-----
> From: Steve Harris [mailto:steve.harris@garlik.com]
> Sent: 09 October 2009 17:46
> To: Seaborne, Andy
> Cc: public-rdf-dawg@w3.org Group
> Subject: Re: [TF-PP] Questions on Property Paths
> 
> On 9 Oct 2009, at 17:15, Seaborne, Andy wrote:
> >
> >> -----Original Message-----
> >> From: public-rdf-dawg-request@w3.org [mailto:public-rdf-dawg-
> request@w3.org
> >> ]
> >> On Behalf Of Steve Harris
> >> Sent: 08 October 2009 12:50
> >> To: public-rdf-dawg@w3.org Group
> >> Subject: [TF-PP] Questions on Property Paths
> >>
> >> I've just read through the document Andy posted:
> >> http://www.w3.org/2009/sparql/wiki/Design:PropertyPaths

> >>  and I have a couple of questions:
> >>
> >> Is the () construct particularly useful? Without have tried it it
> >> does
> >> seems like it might make the design much harder to implement that if
> >> it was absent. Though I'd welcome evidence to the contrary.
> >
> > With foaf:knows, the foaf:Person is connected to another foaf:Person
> > by a single property.  Sometimes, the link from one resource to the
> > next is via two of more properties so
> >
> > ?x (:p/:q)+ ?y
> 
> OK, I can see that you can't do that without that syntax.
> 
> Do you have any feel for how much effort it was to do an implementation?

My path evaluator is a simple recursive evaluator.  The implemation for "+" it's subpath and tries to repeat it.  It does not know nor care whether it's compound or not.

> 
> >> "Cycles in paths are possible and are handled." in any particular
> >> way?
> >
> > Rough wording. Cycles exist naturally so have to be handled but
> > that's an implementation issues.  It might be by tracking visited
> > nodes, or it might be just part of some algorithm. (e.g. Floyd-
> > Walshall does not have special cases for cycles).
> 
> OK, so you're not proposing any particular solution at this point?

I think it is an implementation issue and outside the spec. I hope I've shown there is more than one way of doing it and so a spec should not choose.

 Andy

> 
> - Steve

Received on Friday, 9 October 2009 17:06:51 UTC