Release 479 (14 Dec 2025)#

General#

  • Generate TLS certificates for internal cluster communications automatically when node discovery type is set to ANNOUNCE. (#27030)

  • Add support for specifying the location of the Exchange Manager configuration file. (#26611)

  • Publish time taken by query in FINISHING state to event listener. (#27202)

  • Add lineage support for output columns of SELECT queries. (#26241)

  • Add support for setting and dropping column defaults via ALTER TABLE ... ALTER COLUMN statement. (#26162)

  • Add array_first() and array_last() functions. (#27295)

  • Add GRACE PERIOD to SHOW CREATE MATERIALIZED VIEW output. (#27529)

  • Allow field name declaration in row literals. For example, row(1 as a, 2 as b) is now legal. (#25261)

  • Add queryText as a regular expression in resource group selector. (#27129)

  • Require JDK 25 to build and run Trino. (#27153)

  • ⚠️ Breaking change: The configuration property task.statistics-cpu-timer-enabled is now defunct and must be removed. (#27504)

  • Deprecate EXPLAIN type LOGICAL and DISTRIBUTED. Use EXPLAIN without a type clause, instead. (#27434)

  • Remove prefer_streaming_operators session property. (#27506)

  • Add experimental performance improvements for remote data exchanges on newer CPU architectures, such as Graviton 3, Skylake, Icelake, or Zen 4+. This can be disabled by setting exchange.experimental.vectorized-serde.enabled=false. (#27426, #26919)

  • Improve performance of array_sort() function. (#27272)

  • Improve performance of repeat() function. (#27369)

  • Improve performance of data exchanges involving variable width data. (#27377)

  • Improve performance of queries referencing fresh materialized views. (#27551)

  • Fix query failure when one of the branches of a UNION is known to produce zero rows during query planning. (#21506)

  • Fix configuring partitioned layout for spooling protocol. (#27247)

  • Fix EXPLAIN (TYPE IO) failure when the WHERE clause involves a type which cannot be cast to varchar. (#27433)

Web UI#

  • Render long single-line query string correctly in preview UI query details page. (#27328)

  • Render catalog properties correctly in preview UI. (#27327)

JDBC driver#

  • Add a extraHeaders option to support sending arbitrary HTTP headers. (#15826)

Docker image#

  • Update JDK to 25.0.1. (#27117)

CLI#

  • Add a --extra-header option to support sending arbitrary HTTP headers. (#15826)

  • Fix TLS connection failures when using unqualified (single-label, without “.”) hostnames. ({issue} 27478)

BigQuery connector#

  • Fix query failure when reusing query() table function result. (#27573)

Delta Lake connector#

  • Add APPLICATION_DEFAULT authentication type for GCS. (#26984)

  • ⚠️ Breaking change: Remove support for unauthenticated access when GCS authentication type is set to SERVICE_ACCOUNT. (#26984)

  • Rename s3.exclusive-create config to delta.s3.transaction-log-conditional-writes.enabled. (#27372)

  • Fix incorrect results for queries involving IS NOT DISTINCT FROM. (#27213)

  • Fix failure when writing to tables created by Databricks 17.3. (#27100)

  • Fix failure when checking Azure hierarchical namespaces. (#27278)

  • Avoid worker crashes by failing queries attempting to read columns with huge values in parquet files. (#27148)

  • Fix potential FileAlreadyExistsException failure when network failures occur during writes to S3. (#27330, #27388)

  • Fix potential failure when reading cloned tables. (#27098)

Hive connector#

  • Add APPLICATION_DEFAULT authentication type for GCS. (#26984)

  • ⚠️ Breaking change: Remove support for unauthenticated access when GCS authentication type is set to SERVICE_ACCOUNT. (#26984)

  • Rename s3.exclusive-create config to delta.s3.transaction-log-conditional-writes.enabled. (#27372)

  • Improve accuracy of table statistics written by INSERT queries with OVERWRITE behaviour. (#26517)

  • Fix failure when checking Azure hierarchical namespaces. (#27278)

  • Avoid worker crashes by failing queries attempting to read columns with huge values in parquet files. (#27148)

  • Fix potential FileAlreadyExistsException failure when network failures occur during writes to S3. (#27330)

  • Fix failure when listing tables with invalid table metadata in AWS Glue. (#27525)

Hudi connector#

  • Add APPLICATION_DEFAULT authentication type for GCS. (#26984)

  • ⚠️ Breaking change: Remove support for unauthenticated access when GCS authentication type is set to SERVICE_ACCOUNT. (#26984)

  • Fix failure when checking Azure hierarchical namespaces. (#27278)

  • Avoid worker crashes by failing queries attempting to read columns with huge values in parquet files. (#27148)

  • Fix potential FileAlreadyExistsException failure when network failures occur during writes to S3. (#27330)

  • Fix failure when querying HDFS that includes empty directories. (#26897)

Iceberg connector#

  • Add support for updating token-exchange-enabled via the iceberg.rest-catalog.oauth2.token-exchange-enabled config property. (#27174)

  • Add retain_last and clean_expired_metadata options to expire_snapshots command. (#27357)

  • Add APPLICATION_DEFAULT authentication type for GCS. (#26984)

  • ⚠️ Breaking change: Remove support for unauthenticated access when GCS authentication type is set to SERVICE_ACCOUNT. (#26984)

  • Reduce memory pressure when the table contains highly nested fields. (#25077)

  • Fix failures when querying $files table after changes to table partitioning. (#26746)

  • Fix incorrect results for queries involving IS NOT DISTINCT FROM. (#27213)

  • Fix failure when checking Azure hierarchical namespaces. (#27278)

  • Avoid worker crashes by failing queries attempting to read columns with huge values in parquet files. (#27148)

  • Fix potential FileAlreadyExistsException failure when network failures occur during writes to S3. (#27330)

  • Fix potential failure when dropping a schema with CASCADE option. (#27361)

  • Ignore non-existent or corrupted base Iceberg tables when querying materialized views within the grace period. (#27606)

Loki connector#

  • Fix failure when initializing the connector. (#27180)

Memory connector#

  • Add support for setting and dropping column defaults via ALTER TABLE ... ALTER COLUMN. (#26162)

  • Fix failure in RENAME SCHEMA when there is more than one table in the schema. (#27205)

MySQL connector#

  • Fix incorrect results for queries involving IS NOT DISTINCT FROM. (#27213)

PostgreSQL connector#

  • Fix incorrect results for queries involving IS NOT DISTINCT FROM. (#27213)

Redshift connector#

  • Fix failure when reading Redshift character varying type. (#27224)

SQL Server connector#

  • Fix potential failure when listing tables and columns. (#10846)

SPI#

  • Add non-callback based entry builder to RowBlockBuilder. (#27198)

  • Add non-callback based entry builder to ArrayBlockBuilder. (#27198)

  • Fix ColumnMetadata.builderFrom to retain column default value. (#27503)