It appears that it would be cleaner to make the schematree unaware of types, and instead only have the input source and the final serving code take care of these. this would give some issues with the recommendation strategies,though, because they make their decisions on cardinality, which would then also include types. Related to #26
It appears that it would be cleaner to make the schematree unaware of types, and instead only have the input source and the final serving code take care of these. this would give some issues with the recommendation strategies,though, because they make their decisions on cardinality, which would then also include types. Related to #26