W3C home > Mailing lists > Public > public-cwm-bugs@w3.org > October 2006

Re: Bug? SPARQL query comma-joined object variables

From: Yoshio FUKUSHIGE <fukushige.yoshio@jp.panasonic.com>
Date: Wed, 04 Oct 2006 01:23:11 +0900
To: Tim Berners-Lee <timbl@w3.org>
Cc: fukushige.yoshio@jp.panasonic.com, public-cwm-bugs@w3.org, Yosi Scharf <syosi@mit.edu>
Message-Id: <20061004011104.03A7.FUKUSHIGE.YOSHIO@jp.panasonic.com>

Hello Tim,

On Tue, 03 Oct 2006 08:46:54 -0400
Tim Berners-Lee <timbl@w3.org> wrote:

> 
> Yoshio,
> 
> Looks like a bug in the SPARQL compiler.
> Yosi, do you have any time to look at that?
> 
> Tim
> 
> 
> On Oct 3, 2006, at 7:11, Yoshio Fukushige wrote:
> 
> >
> > Hi,
> >
> > With the data triples in file <gomi.n3>
> > --
> > @prefix : <#> .
> > :s :p :o1, :o2.
> > --
> >
> > and a SPARQL query
> > --
> > PREFIX  : <gomi.n3#>
> >
> > SELECT ?x ?y
> > WHERE{
> >  :s :p ?x, ?y.
> > }
> > --
> >
> > cwm (1.1.0rc2) breaks (on cygwin).
> >
> 
> Did you mean "cwm works"?  Or is there a separate bug in that test?

I wanted to say the cwm (1.1.0rc2), given the query with comma joined
variables, stopped after spitting the error messages I cited later in
my post.

Given the queries with separate triples, or  semi-colon joined variables, 
cwm stopped successfully (without error messages), spitting (A).

I'm not sure (A) is what I ought to get.
If not, there may be a separate bug.

I should read again the SPARQL language specification [2] carefully
again.

[2] http://www.w3.org/TR/rdf-sparql-query/

Best,
Yoshio
> 
> 
> > But with
> > --
> > PREFIX  : <gomi.n3#>
> >
> > SELECT ?x ?y
> > WHERE{
> >  :s :p ?x.
> >  :s :p ?y.
> > }
> > --
> > or
> > --
> > PREFIX  : <gomi.n3#>
> >
> > SELECT ?x ?y
> > WHERE{
> >  :s :p ?x; :p ?y.
> > }
> > --
> >
> > it's OK and spit
> > (A)
> > --
> >      @prefix : <http://www.w3.org/2000/10/swap/sparqlCwm#> .
> >      @prefix gomi: <#> .
> >     {
> >         gomi:o1     :bound "x",
> >                     "y" .
> >
> >         }     a :Result .
> >     {
> >         gomi:o1     :bound "x" .
> >         gomi:o2     :bound "y" .
> >
> >         }     a :Result .
> >     {
> >         gomi:o1     :bound "y" .
> >         gomi:o2     :bound "x" .
> >
> >         }     a :Result .
> >     {
> >         gomi:o2     :bound "x",
> >                     "y" .
> >
> >         }     a :Result .
> > --
> >
> > I'm afraid it's a bug.
> >
> > Here's what I got as the error message
> >
> > --
> > #Processed by Id: cwm.py,v 1.177 2005/08/28 22:53:56 syosi Exp
> >         #    using base file:/cygdrive/c/home/MEI/NEDO2006/local- 
> > test/member/fuku/sparql-test/gomi.n3
> > loaded from file  0.196000099182
> >         ['_O_QCOMMA_E____QObjectList_E__C', ('http://www.w3.org/ 
> > 2000/10/swap/grammar/sparql#GT_COMMA', u','), ('andExtra',  
> > ('objectList', [('Var', y)]), [])]
> > Traceback (most recent call last):
> >   File "/usr/local/cwm-1.1.0rc2/cwm", line 715, in ?
> >     doCommand()
> >   File "/usr/local/cwm-1.1.0rc2/cwm", line 534, in doCommand
> >     filterContext = _store.load(_uri, why=r, referer="",  
> > contentType="x-application/sparql")
> >   File "/usr/local/cwm-1.1.0rc2/swap/llyn.py", line 1329, in load
> >     F = webAccess.load(store, uri, openFormula, asIfFrom,  
> > contentType, flags, referer, why)
> >   File "/usr/local/cwm-1.1.0rc2/swap/webAccess.py", line 166, in load
> >     F = p.parse(sparql_parser.start).close()
> >   File "/usr/local/cwm-1.1.0rc2/swap/sparql/sparql_parser.py", line  
> > 100, in parse
> >     self.onFinish()
> >   File "/usr/local/cwm-1.1.0rc2/swap/sparql/sparql_parser.py", line  
> > 112, in onFinish
> >     prod = self.sink.prod(self.productions.pop())
> >   File "/usr/local/cwm-1.1.0rc2/swap/sparql2cwm.py", line 388, in prod
> >     return getattr(self, 'on_' + abbr(production[0]))(production)
> >   File "/usr/local/cwm-1.1.0rc2/swap/sparql2cwm.py", line 1409, in  
> > on__O_QCOMMA_E____QObjectList_E__C
> >     raise RuntimeError(`p`)
> > RuntimeError: ['_O_QCOMMA_E____QObjectList_E__C', ('http:// 
> > www.w3.org/2000/10/swap/grammar/sparql#GT_COMMA', u','),  
> > ('andExtra', ('objectList', [('Var', y)]), [
> > ])]
> > --
> >
> > And I'm not completely sure if it's OK
> > to have only one variable bound in the result (A) above.
> > (and always with "x" bound but not "y")
> >
> > In contrast SPARQLer [1] returns a result
> > where the two variables got always bound,
> > although there are cases where the two variables are bound to a  
> > single resource (:o1 or :o2).
> >
> > [1] http://sparql.org/sparql.html
> >
> > Best,
> > Yoshio
> >
> > -- 
> > Yoshio Fukushige <fukushige.yoshio@jp.panasonic.com>
> > Network Development Center,
> > Matsushita Electric Industrial Co., Ltd.
> >
> >
> 

-- 
Yoshio FUKUSHIGE <fukushige.yoshio@jp.panasonic.com>
Network Development Center,
Matsushita Electric Industrial Co., Ltd.
Received on Tuesday, 3 October 2006 16:22:45 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 19:52:01 UTC