Details
-
Improvement
-
Status: Open
-
Major
-
Resolution: Unresolved
-
None
-
None
-
None
-
None
Description
Currently --as-parquetfile option is only parsed in case of ImportTool.
This option is checked in CodeGentTool as well, to ensure Sqoop class generation is not clashing with the Parquet lib AvroIndexedRecordConverter class mechanism. With the help of this option the class name is altered with a "codegen_" prefix.
However in case of ExportTool this small fix is not applied, thus if the user directly tries to export a directory containing Parquet files to the very same table from it was imported, then it will cause a Parquet error, which is only avoidable when the user uses explicitly the --class-name option, and resolves the name clashing. As the exception thrown in Parquet is very much not straightforward (it's a ClassCastException) it would be better to make the option available to ExportTool thus users would know that the usage of this option is very much suggested (also the documentation could be improved in a parallel JIRA).