Release 440 (8 Mar 2024)#

General#

  • Add Snowflake connector. (#17909)

  • Add support for sub-queries inside UNNEST clauses. (#17953)

  • Improve performance of arrays_overlap(). (#20900)

  • Export JMX statistics for resource groups by default. This can be disabled with the jmxExport resource group property. (#20810)

  • ⚠️ Breaking change: Remove the defunct *.http-client.max-connections properties. (#20966)

  • Fix query failure when a check constraint is null. (#20906)

  • Fix query failure for aggregations over CASE expressions when the input evaluation could throw an error. (#20652)

  • Fix incorrect behavior of the else clause in a SQL routines with a single if/end condition. (#20926)

  • Fix the ALTER TABLE EXECUTE optimize queries failing due to exceeding the open writer limit. (#20871)

  • Fix certain INSERT and CREATE TABLE AS .. SELECT queries failing due to exceeding the of open writer limit on partitioned tables. (#20871)

  • Fix “multiple entries with same key” query failure for queries with joins on partitioned tables. (#20917)

  • Fix incorrect results when using GRANT, DENY, and REVOKE clauses on views and materialized views. (#20812)

Security#

  • Add support for row filtering and column masking in Open Policy Agent access control. (#20921)

Web UI#

  • Fix error when using authentication tokens larger than 4 kB. (#20787)

Delta Lake connector#

  • Add support for concurrent INSERT queries. (#18506)

  • Improve latency for queries with file system caching enabled. (#20851)

  • Improve latency for queries on tables with checkpoints. (#20901)

  • Fix query failure due to “corrupted statistics” when reading Parquet files with a predicate on a long decimal column. (#20981)

Hive connector#

  • Add support for bearer token authentication for a Thrift metastore connection. (#20371)

  • Add support for commenting on partitioned columns in the Thrift metastore. (#20264)

  • Add support for changing a column’s type from varchar to float. (#20719)

  • Add support for changing a column’s type from varchar to char. (#20723)

  • Add support for changing a column’s type from varchar to boolean. (#20741)

  • Add support for configuring a region and endpoint for S3 security mapping. (#18838)

  • Improve performance when reading JSON files. (#19396)

  • Fix incorrect truncation when decoding varchar(n) and char(n) in TEXTFILE and SEQUENCEFILE formats. (#20731)

  • Fix query failure when hive.file-status-cache-tables is enabled for a table and new manifest files have been added but not cached yet. (#20344)

  • Fix error when trying to INSERT into a transactional table that does not have partitions. (#19407)

  • Fix query failure due to “corrupted statistics” when reading Parquet files with a predicate on a long decimal column. (#20981)

Hudi connector#

  • Fix query failure due to “corrupted statistics” when reading Parquet files with a predicate on a long decimal column. (#20981)

Iceberg connector#

Warning

This release has a major regression which is fixed in Trino 442.

  • Improve latency of queries when file system caching is enabled. (#20803)

  • Disallow setting the materialized view owner when using system security with the Glue catalog. (#20647)

  • Rename the orc.bloom.filter.columns and orc.bloom.filter.fpp table properties to write.orc.bloom.filter.columns and write.orc.bloom.filter.fpp, respectively. (#20432)

  • Fix query failure due to “corrupted statistics” when reading Parquet files with a predicate on a long decimal column. (#20981)

SPI#

  • Add reset to position method to BlockBuilder. (#19577)

  • Remove the getChildren method from Block. (#19577)

  • Remove the get{Type} methods from Block. Callers must unwrap a Block and downcast the ValueBlock to Type.getValueBlockType() implementation. (#19577)