We have a version of the software that is ready for Mike Carbin's students to use to build a semantics around.

This means the key concepts can't change, though superficial aspects of their expression can (and the software itself can change quite a bit, including names for the library, performance characteristics, etc).

Key concepts include:

• traces and addresses,
• generative vs inference procedures,
• interventions,
• coverage of examples,
• the meta-circular interpreter's key clauses,
• opaque vs traced execution