My initial reaction is still that option 2 is excessively modeled, compared to the typical OpenMRS use case. But if you all think it’s important, I guess we can add the extra model complexity.
@jthomas, @burke, can we discuss this briefly on a Monday design call and approve one of these approaches for inclusion in openmrs-core?
@karrtikiyer, could you join a call Monday evening IST? (Perhaps even on the 29th?)
-Darius (by phone)