[redland-dev] Parser incorrectly interprets Blank Nodes on reload...

[Brian Holdsworth]

All,

I am seeing a strange behavior on my model when I
re-parse it after using the rdfxml serializer.  My
model has lots of blank node identifiers.  When I
serialize it to a rdfxml file, I get statements such
as this related to the blank nodes:

  <rdf:Description rdf:about="r1r377">
    <rdf:value>843-218-6669</rdf:value>
  </rdf:Description>

Later on, I use the rdfxml parser to load the file
containing these statements into a model.  My parser
invocation includes a base URI specification.  What
seems to happen is that the parser prepends the base
URI to the statements about the blank nodes.  So, I
wind up with new statements like this:

<rdf:Description
rdf:about="http://my.base_uri.com/r1r377">
        <rdf:value>843-218-6669</rdf:value>
</rdf:Description>

When that happens, I seem to lose the original
association between resources in my model and the
blank nodes they use to point to using an rdf:nodeID.

Shouldn't the parser recognize statments about
resources named "r1rXXX" as being internal identifiers
for blank nodes?  If so, shouldn't it ignore the base
URI when parsing these statements, and just give it a
blank node URI like "_:r1r377", or some such thing?

Thanks,
Brian