What is the use case? 'Normally' this knowledge should not be 'necessary' for the algorithm and both node types should be transparent. (Of course in real world you cannot always ignore this, e.g. like we needed the difference in the new landmarks algorithm)
