Release 364 (1 Nov 2021)#

General#

Add support for ALTER MATERIALIZED VIEW ... RENAME TO .... (#9492)
Improve performance of GROUP BY with single grouping column. (#9514)
Improve performance of decimal aggregations. (#9640)
Improve performance when evaluating the WHERE and SELECT clause. (#9610)
Improve performance when computing the product of decimal values with precision larger than 19. (#9744)
Improve CPU coordinator utilization. (#8650)
Remove support for the unwrap_casts session property and optimizer.unwrap-casts configuration option. (#9550)
Fix incorrect results for queries with nested joins and IS NOT DISTINCT FROM join clauses. (#9805)
Fix displaying character type dynamic filter values in EXPLAIN ANALYZE. (#9673)
Fix query failure for update operation if it has a correlated subquery. (#8286)
Fix decimal division when result is between -1 and 0. (#9696)
Fix SHOW STATS failure for a query projecting a boolean column. (#9710)

Improve responsiveness of Web UI when query history contains queries with long query text. (#8892)

Allow using token from existing Kerberos context. This allows the client to perform Kerberos authentication without passing the Keytab or credential cache to the driver. (#4826)

Allow to skip setting permissions on new directories by setting hive.fs.new-directory-permissions=skip in connector properties file. (#9539)
Allow translating Hive views which cast timestamp columns to decimal. (#9530)
Add optimize table procedure for merging small files in non-transactional Hive table. Procedure can be executed using ALTER TABLE <table> EXECUTE optimize(file_size_threshold => ...) syntax. (#9665)
Restrict partition overwrite on insert to auto-commit context only. (#9559)
Reject execution of CREATE TABLE when bucketing is requested on columns with unsupported column types. Previously CREATE was allowed but it was not possible to insert data to such a table. (#9793)
Improve performance of querying Parquet data for files containing column indexes. (#9633)
Fix Hive 1 and Hive 3 incompatibility with Parquet files containing char or varchar data produced by the experimental Parquet writer. Hive 2 or newer should now read such files correctly, while Hive 1.x is still known not to read them. (#9515, (#6377))
Fix ArrayIndexOutOfBoundsException when inserting into a partitioned table with hive.target-max-file-size set. (#9557)
Fix reading Avro schema written by Avro 1.8.2 with non-spec-compliant default values. (#9243)
Fix failure when querying nested Parquet data if column indexes are enabled. (#9587)
Fix incorrect results when querying Parquet data. (#9587)
Fix query failure when writing to a partitioned table with target max file size set. (#9557)

Add support for renaming materialized views. (#9492)
Create Parquet files that can be read more efficiently. (#9569)
Improve query performance when dynamic filtering can be leveraged. (#4115)
Return value with UTC zone for table partitioned on timestamp with time zone. (#9704)
Fix data loss in case of concurrent inserts to a table. (#9583)
Fix query failure when reading from $partitions table for a table partitioned on timestamp with time zone or uuid (#9703, #9757)
Fix query failure when reading Iceberg table statistics. (#9714)

Support reading and writing MemSQL datetime(6) and timestamp(6) types as Trino timestamp(6). (#9725)

Fix query failure when count(*) aggregation is pushed down to SQL Server database and the table has more than 2147483647 rows. (#9549)

Expose which columns are covered by dynamic filters. (#9644)
Add SPI for table procedures that can process table data in a distributed manner. Table procedures can be run via ALTER TABLE ... EXECUTE ... syntax. (#9665)