I wonder if we can do one of the following two, opposite strategies:
a) Be really opinionated on which columns need to be included in the measure config file, and just make all potential columns compulsory
b) Only make essential columns (ref, measure_name, unit, improvement_direction?) required, and make everything else optional. It strikes me that things like target and owner are good to have but strictly speaking optional, in that the report and charts can be successfully generated without them. And then the package will just add on any missing optional cols (as NAs), and we then just have to ensure that NA values are handled appropriately by all the package processes.
I wonder if we can do one of the following two, opposite strategies:
a) Be really opinionated on which columns need to be included in the measure config file, and just make all potential columns compulsory b) Only make essential columns (ref, measure_name, unit, improvement_direction?) required, and make everything else optional. It strikes me that things like target and owner are good to have but strictly speaking optional, in that the report and charts can be successfully generated without them. And then the package will just add on any missing optional cols (as NAs), and we then just have to ensure that NA values are handled appropriately by all the package processes.