W3C home > Mailing lists > Public > public-rdf-wg@w3.org > August 2012

Re: shared bnodes (Skolems, SPARQL)

From: Andy Seaborne <andy.seaborne@epimorphics.com>
Date: Fri, 31 Aug 2012 19:12:09 +0100
Message-ID: <5040FE79.7020609@epimorphics.com>
To: Sandro Hawke <sandro@w3.org>
CC: Steve Harris <steve.harris@garlik.com>, public-rdf-wg@w3.org


On 31/08/12 17:30, Sandro Hawke wrote:
>> and also related ...
>>
>> PREFIX : <http://example/>
>>
>> INSERT DATA
>> {
>>   GRAPH :g1 { :s1 :p1 _:a }
>>   GRAPH :g2 { :s2 :p2 _:a }
>> }
>>
>> followed by ...
>>
>> SELECT * { GRAPH ?g { ?s ?p ?o } }
>>
>> or
>>
>> SELECT ?g1 ?g2 ?o {
>>   GRAPH ?g1 { ?s2 ?p2 ?o }
>>   GRAPH ?g2 { ?s2 ?p2 ?o }
>> }
>>
>
> Indeed.
>
> Are those currently SPARQL 1.1 test cases?  [1]   Is it clear what the
> answer is in SPARQL 1.1?
>
>        -- Sandro
>
>
> [1] I don't see anything like that in
> http://www.w3.org/2009/sparql/docs/tests/data-sparql11/basic-update or
> nearby, but maybe some tests are somewhere else...?

basic-update/insert-05a.ru

It has to go about testing whether the bNode is shared in a somewhat 
convoluted way because the results are recorded in RDF graphs, not in 
TriG.  So the results can't record shared bNode as it's two RDF files 
which when parsed can't express a shared bNode.

The test copies one graph (with a bNode) into another, twice, then 
counts to see it how many triples, and records that count.  If the copy 
was introducing naming apart, the count would be different.  (Yes - the 
engine could keep a rename map and do it consistently but that is 
getting a bit far fetched.)

Not a perfect test but given the bootstrap problem of recording results, 
the SPARQL-WG thought it sufficient.

	Andy
Received on Friday, 31 August 2012 18:12:45 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 22:02:06 UTC