Re: Requirement 3.6: Optional Match

Rob Shearer wrote:
> 
> I'm afraid I can't really support this as a requirement, although I
> think features which allow this result (albeit not necessary directly)
> would almost certainly be extremely useful.
> 
> The thing is, "optional bindings" are really nothing more than unions
> between independent queries: union the results of the query with the
> binding with the results of the query without it. I'd say a more
> general-purpose "union" operator is quite a bit more important than this
> one particular form, and even then I'm not sure whether it rises to the
> level of requirement.

maybe. i guess it depends on what is meant by union.

regardless, our goal is not a *minimal* query language
but a useful one. as andy noted, RDQL users have requested
optional triples.

and, implementing optional triples as multiple union
queries could strain the capabilities of query
optimizers. seems like it's more efficient to
make optional triples explicit. SQL has both
union and outer join, for example.

kevin



> > -----Original Message-----
> > From: public-rdf-dawg-request@w3.org
> > [mailto:public-rdf-dawg-request@w3.org] On Behalf Of Seaborne, Andy
> > Sent: 06 May 2004 08:40
> > To: RDF Data Access Working Group
> > Subject: Requirement 3.6: Optional Match
> >
> >
> >
> > I suggest what I think are small changes:
> >
> > + including the term "graph pattern" as is in the suggested
> > text for 3.1.
> > + tweaking the title
> > + Be neutral to "optional triples" and "may-bind" variables.
> >
> > ------------------------
> >
> > 3.6 Optional Matching
> >
> > It must be possible to express a query graph pattern that
> > does not fail to
> > match when some nominated part of the graph pattern fails to
> > match.  Any
> > such triples matched by this optional part, or variable
> > bindings caused by
> > this optional part, can be requested in the results.
> >
> > ------------------------
> >
> >       Andy
> >
> >
> >
> >
> > For reference, the text in v1.39 is:
> > """
> > 3.6 Optional Match
> > It must be possible to express a query that does not fail when some
> > nominated part of the query specifying "optional triples"
> > fails to match.
> > Any such triples matched by this optional part, or variable
> > bindings caused
> > by this optional part, can be returned in the results, if requested.
> > """
> >
> >

Received on Friday, 7 May 2004 20:41:32 UTC