[Mulgara-dev] Re: JRDF and blank nodes

[Andrew Newman]

On 10/21/06, Andrae Muys <andrae at netymon.com> wrote:
>
> On 21/10/2006, at 12:15 PM, Andrew Newman wrote:
>
> > On 10/20/06, Andrae Muys <andrae at netymon.com> wrote:
> Two things.  First I obviously wasn't clear - I wasn't suggesting
> what should happen, I was describing what is currently happening.
>
> The call to finishTransactionalOperation will end the transaction,
> consequently invalidating the nodeid in the blanknode being returned.
>

So the reason it might still be working is that the node id isn't be
invalidated because node reaping isn't occuring.  I don't remember if
that's still the case.

> As far as server-side jrdf is concerned, what is wrong with identity
> comparisons?  The equals() method for BlankNodeImpl already considers
> referential-equivalence, so obtaining a BlankNodeImpl and just using
> it wherever necessary should be adequate for what you've described.
> For the client-side implementation (and this would work as an
> alternative approach to server-side as well), I can't see any reason
> why VariableNodeImpl wouldn't work fine - this is just an AST version
> of the bnode-identifiers used in rdf serialisations.  You will need
> to define some concept of 'context' to provide a scope for the
> identifiers.
>

So there are/were cases where new instances of the same blank node (by
id) are created - mainly going between systems - like RMI.