Lei Sun created ORC-613: --------------------------- Summary: OrcMapredRecordReader mis-reuse struct object when actual children schema differs Key: ORC-613 URL: https://issues.apache.org/jira/browse/ORC-613 Project: ORC Issue Type: Bug Components: Java Reporter: Lei Sun
When reading from schema like following: {{uniontype <struct<field0, field1, ..., fieldN>, struct<>> }} `org.apache.orc.mapreduce.OrcMapreduceRecordReader#nextStruct` will determine if previous object's schema can be reused or not. The determination of this is problematic, since it only checks the top-level type (OrcStruct) but not the schema of OrcStruct. Therefore, if encountering schema like above, and when struct at tag_0 is processed followed with a struct at tag_1, it will reuse the tag_0's struct schema which results in in correct result. -- This message was sent by Atlassian Jira (v8.3.4#803005)