Release 0.217

    • Fix a bug that caused a union of a table with a statement to execute on a single machine, which may result in out of memory errors.

    • Improve performance of some queries that use window functions by eliminating redundant shuffles.

    • Add grouped execution support for window functions.

    • Add support for choosing distribution type for semi joins based on estimated cost when the session property join_distribution_type='AUTOMATIC' is set.

    • Add support for collecting table statistics on demand with the ANALYZE statement.

    • Add a config option (query.stage-count-warning-threshold) to specify a per-query threshold for the number of stages. When this threshold is exceeded, a TOO_MANY_STAGES warning is raised.

    • Add CLI support for showing the amount of data spilled during query execution.

    • Add ST_Points() function.

    • Add .

    • Remove the system memory pool and related configuration properties (, deprecated.legacy-system-pool-enabled) entirely. System memory pool was deprecated in 0.201, and it was unused by default since that release. All memory allocations will now be served from the general/user memory pool.

    • Improve live plan to show data transfer statistics as edge labels.

    • Add the ability to zoom when viewing plans for completed queries.

    • Fix an issue where a partially successful rollback of a write could cause data loss and corrupt the metastore. The hive.skip-target-cleanup-on-rollback configuration property can be used to skip deleting target directories when partition creation is rolled back.

    • Fix an issue where creating a table on S3 could fail for S3 prefixes without any associated objects (e.g., empty S3 directories).

    • Add support for ANALYZE statement in the Hive connector. It’s possible to specify a list of partitions to collect statistics for using the WITH properties of the ANALYZE statement.

    • Add configuration property hive.temporary-staging-directory-enabled and session property to control whether a temporary staging directory should be used for write operations.

    • Add configuration property hive.temporary-staging-directory-path and session property temporary_staging_directory_path to control the location of temporary staging directory that is used for write operations. The ${USER} placeholder can be used to use a different location for each user (e.g., /tmp/${USER}).

    • Add support for defining procedures.

    • Add support for providing table statistics.