- Fix incorrect results when queries contain multiple grouping sets that
resolve to the same set.
- Fix incorrect results when using
- Fix compile failure for outer joins that have a complex join criteria.
- Fix error messages for failures during commit.
- Fix memory accounting for simple aggregation, top N and distinct queries.
These queries may now report higher memory usage than before.
- Reduce unnecessary memory usage of
PROPERTIES non-reserved keywords.
- Remove support for the experimental feature to compute approximate queries
based on sampled tables.
- Properly account for time spent creating page source.
- Various optimizations to reduce coordinator CPU usage.
- Fix schema evolution support in new Parquet reader.
NoClassDefFoundError when using Hadoop KMS.
- Add support for Avro file format.
- Always produce dictionary blocks for DWRF dictionary encoded streams.
- Remove legacy connector API.