[redland-dev] problem using model with virtuoso storage backend

Dave Beckett dave at dajobe.org
Sun Nov 21 20:25:04 CET 2010


Looks like you have two bugs there maybe.  I had a quick look at the
virtuoso storage code but that area is a bit of a mess of cryptic variables
and I couldn't see anything obviously wrong.

You'll have to take a look at BindObject() in the clause to do with literals
and the prepared statement.  It might be one of those integers is wrong (!)

https://github.com/dajobe/librdf/blob/master/src/rdf_storage_virtuoso.c

and/or file a bug or two at bugs.librdf.org so it'll get tracked.

Thanks

Dave


On 11/17/10 8:28 AM, Jasper Op de Coul wrote:
> Hi
> 
> I've been experimenting with the Virtuoso storage backend through the 
> python bindings. I ran into some problems I would like to share.
> 
> With the python bindings you can query the backend using either:
> 
> query.execute(model) or model.execute(query)
> 
> this translates to librdf_query_execute and librdf_model_query_execute 
> in the c code. A Segmentation fault occurs when using 
> librdf_query_execute with the 'vsparql' language, the 
> librdf_model_query_execute function works fine though.
> 
> Another problem I ran into is that the add_statement functions on a 
> model will assume that the object part of the statement is a resource 
> uri if there is no language or datatype specified.
> 
> This also happens when you parse directly into a model:
> 
> data = '<x:test> <x:label> "test".\n'
> parser = RDF.Parser("ntriples")
> parser.parse_string_into_model(model, data, RDF.Uri('file://'))
> for stat in model: print stat
> 
> prints:
> 
> <x:test> <x:label> <test>
> 
> (adding a language or datatype to the literal triggers the right behaviour)
> 
> When I upload data using the rdfproc utility everything works fine,
> so it seems the problem is tied up with the model code somehow.
> 
> 
> I'm using:
> 
> redland-1.0.11
> redland-bindings-1.0.11.1
> virtuoso-opensource-6.1.2
> 
> All tests of the `redland-virtuoso-test` passed.
> 
> Thanks,
> 
> Jasper Op de Coul
> 
> 



More information about the redland-dev mailing list