Release 422 (13 Jul 2023)#


  • Add support for adding nested fields with an ADD COLUMN statement. (#16248)

  • Improve performance of INSERT and CREATE TABLE AS ... SELECT queries. (#18005)

  • Prevent queries from hanging when worker nodes fail and the task.retry-policy configuration property is set to TASK. (#18175)


  • Add support for validating JWT types with OAuth 2.0 authentication. (#17640)

  • Fix error when the http-server.authentication.type configuration property is set to oauth2 or jwt and the principal-field property’s value differs. (#18210)

BigQuery connector#

  • Add support for writing to columns with a timestamp(p) with time zone type. (#17793)

Delta Lake connector#

  • Add support for renaming columns. (#15821)

  • Improve performance of reading from tables with a large number of checkpoints. (#17405)

  • Disallow using the vacuum procedure when the max writer version is above 5. (#18095)

Hive connector#

  • Add support for reading the timestamp with local time zone Hive type. (#1240)

  • Add a native Avro file format writer. This can be disabled with the avro.native-writer.enabled configuration property or the avro_native_writer_enabled session property. (#18064)

  • Fix query failure when the hive.recursive-directories configuration property is set to true and partition names contain non-alphanumeric characters. (#18167)

  • Fix incorrect results when reading text and RCTEXT files with a value that contains the character that separates fields. (#18215)

  • Fix incorrect results when reading concatenated GZIP compressed text files. (#18223)

  • Fix incorrect results when reading large text and sequence files with a single header row. (#18255)

  • Fix incorrect reporting of bytes read for compressed text files. (#1828)

Iceberg connector#

  • Add support for adding nested fields with an ADD COLUMN statement. (#16248)

  • Add support for the register_table procedure to register Hadoop tables. (#16363)

  • Change the default file format to Parquet. The iceberg.file-format catalog configuration property can be used to specify a different default file format. (#18170)

  • Improve performance of reading row types from Parquet files. (#17387)

  • Fix failure when writing to tables sorted on UUID or TIME types. (#18136)

Kudu connector#

  • Add support for table comments when creating tables. (#17945)

Redshift connector#

  • Prevent returning incorrect results by throwing an error when encountering unsupported types. Previously, the query would fall back to the legacy type mapping. (#18209)