> Well, *I* hope so :-) Not everything in the extensions in ARQ
> should be candidates for standardisation but property paths are one
> I think would be useful.

Yes absolutely. IMHO by far the most useful aspect here is the *
operator that allows users to get transitive sub/super classes and
properties at query time, e.g.

?p rdfs:subPropertyOf* rdfs:label .
?ex ?p ?label .

to get all values of rdfs:label (or sub-properties thereof) at ?ex. Or

?type rdfs:subClassOf* ex:Person .
?instance a ?type .

to get all instances of ex:Person (or sub-classes thereof).

This extension basically means that a huge fraction of use cases of
RDFS/OWL inferencing can be covered by SPARQL alone. No need for any
other (potentially slow) inference engines in the stack here!

I hope the working group will consider this.

Holger

PS: Having said this, I also believe that SPARQL 2 should contain LET,
but that's another story.