Re: Problem with auto-generated fragment IDs for graph names

On Wed, 20 Feb 2013 11:07:00 +0000, Steve Harris <steve.harris@garlik.com> said:

    > Why not? It's legal for other RDF serialisations to do that, via
    > skolemisation. It's logically sound, and a sensible approach to
    > situations where you don't want to require the generating
    > implementation to mint URIs.

Really? Other serialisations do that?

Skolemisation *changes* the data. It changes it in such a way that
propositions continue to make sense in the same circumstances.

Sure, it is useful to be able to transform data to an equisatisfiable
skolem form just as it is useful to transform and change data in other
ways.

Minting URIs, skolem constants or otherwise changes the data.

Changing the data is not the business of (de)serialisation.

-w