[redland-dev] duplicate bnode ids

[Dave Beckett]

On Wed, 15 Sep 2004, Rainer Volz wrote:
> I'm trying to use Redland for the management, query, and transformation of PIM
> data. However, I ran into a problem when I tried to query the information
> stored. 
> 
> Whenever I tried to query information that involved bnodes I got strange
> results, more data than I wanted. After some experimenting I found out that
> some of the ca. 900 bnodes share the same generated ID, although they have
> different contents and are located in different contexts.
> 
> I'm using a Redland 0.9.18 bdb storage with contexts on a Linux
> system. Currently the data (turtle files) is imported in one run via
> rdfproc - no updates, deletes etc. Is there a way to avoid the duplicate ids?

It shouldn't be doing that.  I assume you mean multiple runs
of rdfproc, one for each file like:
  for file in *.ttl; do
    contexturi=http://example.org/$file
    rdfproc db parse-stream turtle file:$file $baseuri $contexturi
  done

There was a bug in an earlier version of raptor that meant turtle
Âbnodes weren't properly rewritten, but that should be fixed with
the raptor that ships with 0.9.18.

Can you let me have a small set of data that demonstrates this,
and maybe send me the method you use to import them?

Thanks

Dave