Default Prefix for Turtle Documents

[tobiaswjohn]

The TurtleParser sets a default IRI (namely http://www.semanticweb.org/owl/owlapi/turtle#) for the empty prefix before parsing a turtle document. Doing so allows the OWL API to parse (incorrect) turtle files where the empty prefix is not defined, while other tools, i.e. the JENA API, troughs an exception because an undefined prefix is used. An example is the following:

:j  a  <i> .

I consulted the turtle standard and think that setting a default IRI for the empty prefix is not according to the standard (please correct me if I am wrong).

There are probably some users that depend on this handling of an undefined empty prefix, so changing it might be hard. But I think that the current implementation can have unexpected side effects, e.g. if a user, for whatever reason, uses the same prefix somewhere in their ontology. Maybe one can output a warning when the default IRI is used?