Details
-
Bug
-
Status: Resolved
-
Critical
-
Resolution: Fixed
-
None
-
None
-
None
Description
https://github.com/apache/incubator-parquet-mr/pull/36
The ParquetLoader was not storing the pig schema into the udfcontext for the full load case which causes a schema reload on the task side, erases the requested schema, and causes problems with column index access.
This fix stores the pig schema to both the udfcontext (for task side init) and jobcontext (for TupleReadSupport) along with other properties that should be set in the loader context (required field list and column index access toggle).