Release 0.201#
General#
- Change grouped aggregations to use - IS NOT DISTINCT FROMsemantics rather than equality semantics. This fixes incorrect results and degraded performance when grouping on- NaNfloating point values, and adds support for grouping on structural types that contain nulls.
- Fix planning error when column names are reused in - ORDER BYquery.
- System memory pool is now unused by default and it will eventually be removed completely. All memory allocations will now be served from the general/user memory pool. The old behavior can be restored with the - deprecated.legacy-system-pool-enabledconfig option.
- Improve performance and memory usage for queries using - row_number()followed by a filter on the row numbers generated.
- Improve performance and memory usage for queries using - ORDER BYfollowed by a- LIMIT.
- Improve performance of queries that process structural types and contain joins, aggregations, or table writes. 
- Add session property - prefer-partial-aggregationto allow users to disable partial aggregations for queries that do not benefit.
- Add support for - current_user(see Session information).
Security#
- Change rules in the System access control for enforcing matches between authentication credentials and a chosen username to allow more fine-grained control and ability to define superuser-like credentials. 
Hive#
- Replace ORC writer stripe minimum row configuration - hive.orc.writer.stripe-min-rowswith stripe minimum data size- hive.orc.writer.stripe-min-size.
- Change ORC writer validation configuration - hive.orc.writer.validateto switch to a sampling percentage- hive.orc.writer.validation-percentage.
- Fix optimized ORC writer writing incorrect data of type - mapor- array.
- Fix - SHOW PARTITIONSand the- $partitionstable for tables that have null partition values.
- Fix impersonation for the simple HDFS authentication to use login user rather than current user. 
SPI#
- Support resource group selection based on resource estimates.