Release 397 (21 Sep 2022)#
General#
Security#
JDBC driver#
- Fix potential memory leak when cancelling statements. (#14176) 
Delta Lake connector#
- Rename the - parquet.experimental-optimized-writer.enabledconfiguration property and- experimental_parquet_optimized_writer_enabledsession property to- parquet.optimized-writer.enabledand- parquet_optimized_writer_enabled, respectively. (#14137)
Hive connector#
- Rename the - parquet.experimental-optimized-writer.enabledconfiguration property and- experimental_parquet_optimized_writer_enabledsession property to- parquet.optimized-writer.enabledand- parquet_optimized_writer_enabled, respectively. (#14137)
- Improve performance when querying JSON data and Hive S3 Select pushdown is enabled. (#14040) 
- Improve planning performance when querying tables in the Glue catalog that contain a large number of columns. (#14206) 
- Allow reading from a partitioned table after a column’s data type was changed from - decimalto- varcharor- string. (#2817)
- Fix query failure when reading from a Hive view and - hive.hive-views.run-as-invokerand- hive.hive-views.legacy-translationare both enabled. (#14077)
Iceberg connector#
- Improve performance of queries that contain predicates involving - date_truncwith an- hourunit on- date,- timestamp, or- timestamp with time zonepartition columns. (#14161)
- Improve performance of reads after a - DELETEremoves all rows from a file. (#14198)
- Reduce query latency when using a Glue catalog for metadata. (#13875) 
- Fix materialized views temporarily appearing empty when a refresh is about to complete. (#14145) 
- Fix potential table corruption when changing a table before it is known if committing to the Hive metastore has failed or succeeded. (#14174)