Pipeline.reduce implementations are expected to maintain a few columns through the reduce operation (e.g., the geometryColumn of type jts.Geometry, the keyColumn of type Array[SpatialKey]), but we don't check if that is the case. We can catch these types of errors early in the pipeline and deliver more meaningful error messages to aid pipeline debugging.
Pipeline.reduceimplementations are expected to maintain a few columns through the reduce operation (e.g., the geometryColumn of type jts.Geometry, the keyColumn of type Array[SpatialKey]), but we don't check if that is the case. We can catch these types of errors early in the pipeline and deliver more meaningful error messages to aid pipeline debugging.