Release 0.162

Warning

The xxhash64() function introduced in this release will return a varbinary instead of a bigint in the next release.

General Changes

  • Fix correctness issue when the type of the value in the IN predicate does not match the type of the elements in the subquery.

  • Fix correctness issue when the value on the left-hand side of an IN expression or a quantified comparison is NULL.

  • Fix correctness issue when the subquery of a quantified comparison produces no rows.

  • Fix correctness issue due to improper inlining of TRY arguments.

  • Fix correctness issue when the right side of a JOIN produces a very large number of rows.

  • Fix correctness issue for expressions with multiple nested AND and OR conditions.

  • Improve performance of window functions with similar PARTITION BY clauses.

  • Improve performance of certain multi-way JOINs by automatically choosing the best evaluation order. This feature is turned off by default and can be enabled via the reorder-joins config option or reorder_joins session property.

  • Add xxhash64() and to_big_endian_64() functions.

  • Add aggregated operator statistics to final query statistics.

  • Allow specifying column comments for CREATE TABLE.

Hive Changes

  • Fix performance regression when querying Hive tables with large numbers of partitions.

SPI Changes

  • Connectors can now return optional output metadata for write operations.

  • Add ability for event listeners to get connector-specific output metadata.

  • Add client-supplied payload field X-Presto-Client-Info to EventListener.