The semantics of WKST is not well specified. The best we have right now is a set of examples. Would it make sense for us to include a specification for precisely how WKST affects the set of instances described by an RRULE? Better yet, would it be possible to have a compact reference implementation for the semantics of a recurrence rule?