jnthn clarifies some key things about unbound names.
TimToady says "jnthn++ has said that we might be able to find a bit in the captures to tell whether a given argument has ever been bound, and warn on unbounds at the end".
TimToady says "we really gotta get a warning for" unbound names.
FROGGS suggests masak codes FROGGS' suggested solution.
Yesterday masak says '"interface consistency" prevents us from showing a warning'. But that seems wrong (see above comments).