Description
When reading from an Orc file, SchemaEvolution evolution tries to determine if this is an Acid compliant format by comparing field names with Acid event names in SchemaEvolution.checkAcidSchema. Would be good to make this comparison case insensitive.
This requirement comes in from HIVE-20699 where a Hive query is being used to run compaction (and hence write the compacted data to the bucket files via a HiveQL query). Since hive converts all column names to lower case, the compacted files end up with lower case Acid schema columns. The change is much simpler when made in Orc.
Attachments
Attachments
Issue Links
- blocks
-
HIVE-20699 Query based compactor for full CRUD Acid tables
- Resolved
- relates to
-
ORC-553 Add test case to check that SchemaEvolution checkAcidSchema works well
- Closed
- links to