Hi, It seems that specifying the input schema when reading Avro files in a MapReduce job is required. I would expect it to be optional, since the schema already exists in the file itself. It seems that specifying an input schema would only be useful to handle schema discrepancies between actual and expected schemas.
Alex.
