Skip to main content

Spark - Handling Optional Values

Optional values are handled differently in SQL and Morphir. In SQL every value can potentially be NULL and there are implicit rules about how each operator should behave when NULLs are passed to them. Spark follows the same approach as documented in the NULL semantics section. Morphir on the other hand requires the modeler to handle missing values explicitly.

If the goal was simply to accurately map the Morphir code to Spark we could translate the Maybe API calls directly to case statements. While that would work it would generate unnecesarily complex logic when the default NULL semantics in Spark are already in-line with what the Morphir code expresses. For example, given the Elm code below:

|> (\x -> x.aField == baseRelation.aField)
|> Maybe.withDefault False

The corresponding SQL can be much simpler since comparing anything to missing values will return false anyway:

joinedRelation.a_field == baseRelation.a_field

TODO: Add more details