I ran into an issue tonight with parsing log lines whereby I had to generate a schema in a user defined function. Part of that involved converting various values into their associated data types, but I couldn't see a way to do it via Pig. Enclosed is a patch to convert org.apache.pig.data.DataType into an associated Java datatype falling back to a String by default - should go into CastUtils. Any comments/criticism, please let me know.
Cheers, Jon.