Release 413 (12 Apr 2023)

General

• Improve performance of queries involving window operations or row pattern recognition on small partitions. (#16748)

• Improve performance of queries with the row_number() and rank() window functions. (#16753)

• Fix potential failure when cancelling a query. (#16960)

Delta Lake connector

• Add support for nested timestamp with time zone values in structural data types. (#16826)

• Disallow using _change_type, _commit_version, and _commit_timestamp as column names when creating a table or adding a column with change data feed. (#16913)

• Disallow enabling change data feed when the table contains _change_type, _commit_version and _commit_timestamp columns. (#16913)

• Fix incorrect results when reading INT32 values without a decimal logical annotation in Parquet files. (#16938)

Hive connector

• Fix incorrect results when reading INT32 values without a decimal logical annotation in Parquet files. (#16938)

• Fix incorrect results when the file path contains hidden characters. (#16386)

Hudi connector

• Fix incorrect results when reading INT32 values without a decimal logical annotation in Parquet files. (#16938)

Iceberg connector

• Fix incorrect results when reading INT32 values without a decimal logical annotation in Parquet files. (#16938)

• Fix failure when creating a schema with a username containing uppercase characters in the Iceberg Glue catalog. (#16116)

Oracle connector

• Add support for table comments and creating tables with comments. (#16898)

Phoenix connector

• Add support for MERGE. (#16661)

SPI

• Deprecate the getSchemaProperties() and getSchemaOwner() methods in ConnectorMetadata in favor of versions that accept a String for the schema name rather than CatalogSchemaName. (#16862)