[HIVE-23561] FIX Arrow Decimal serialization for native VectorRowBatches

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 4.0.0-alpha-1
• Labels: pull-request-available
• Assignee: Panagiotis Garefalakis
• Reporter: Panagiotis Garefalakis
• Created: 28/May/20 12:40
• Updated: 17/Nov/22 08:48
• Resolved: 09/Jun/20 22:21

Description

Arrow Serializer does not properly handle Decimal primitive values when selected array is used.
In more detail, decimalValueSetter should be setting the value at arrowIndex[i] as the value at hiveIndex[j], however currently its using the same index!
https://github.com/apache/hive/blob/eac25e711ea750bc52f41da7ed3c32bfe36d4f67/ql/src/java/org/apache/hadoop/hive/ql/io/arrow/Serializer.java#L926

This works fine for cases where i == j (selected is not used) but returns wrong decimal row values when i != j.
This ticket fixes this inconsistency and adds tests with selected indexes for all supported types

Issue Links

• links to: Github PR
• links to: GitHub Pull Request #1038