CsvColumn

Note that the codomain of functions produced from a CsvColumn, c, is Option[c.ColType]. This is so that functions with optional variables can be treated the same way as functions with no variables or only required variables. In order to accomplish this, we need to use Option so that if a variable is missing the feature output is None. If all variables are present, it can be a Some that wraps the user-designed feature's functionality.

We can go even further and allow the feature author to return an Option directly and then fall back to the defVal prodvided in the specification if the function returns None.

This is how the feature function compilation works. It does something like the following:

val semantics: CompiledSemantics[A] = ???
val c: CsvColumn = ???

val f = semantics.createFunction[Option[c.ColType]](c.wrappedSpec, Some(c.defVal))(c.refInfo)

f match {
  case Right(success) => success.andThenGenAggFunc(_ orElse c.defVal)
}