Currently there are two ways to return type information in EvalFunc. Type getReturnType() Schema outputSchema(Schema input)
Does anyone really understand why we need both of them ? Seems like there can be two things:- 1) getReturnType() is a deprecated one. Then we should mark it as @deprecated. 2) getReturnType() is for returning the actual type. outputSchema() is for returning schema of complex type only in case that getReturnType() happens to return a complex type. This, however, poses a problem that in getReturnType() refers to Java types but types in outputSchema() are our types in DataType class. Pi
