Release 481 (11 May 2026)#
General#
Add support for casting
booleantonumber. (#28879)Add support for the
numbertype in Python UDFs. (#28921)Add support for casting between
numberandjson. (#28394)Improve performance of json_value and json_table by evaluating
ON EMPTYandON ERRORclauses lazily. (#28969)Add support for
DESCRIBE OUTPUTwith inline queries, allowing direct description of query results without requiring aPREPAREstatement. For example,DESCRIBE OUTPUT (SELECT * FROM nation). (#28002)Add support for the
NEARESTclause for approximate matches in joins. (#21759)Add support for binding parameters in
WITH SESSION,SET SESSION, andCALLstatements. (#29053)Add support for persisting external authentication tokens to disk (in
~/.trino/), allowing reuse across separate JDBC client processes. SetexternalAuthenticationTokenCache=SYSTEMto enable. (#28783)Include connector split source metrics in
io.trino.spi.eventlistener.QueryInputMetadata#connectorMetrics. (#28870)Replace the Esri geometry library with JTS to improve interoperability with the broader spatial ecosystem. (#27881)
⚠️ Breaking change: Reject WKT input that does not conform to OGC standards. Some inputs that were silently accepted before will now fail. (#27881)
⚠️ Breaking change: Change
ST_Union()to return an empty geometry collection instead ofNULLfor empty inputs. (#27881)⚠️ Breaking change: Stop inserting vertices at intersection points for point-on-line unions in
ST_Union(). (#27881)Improve compatibility of file system exchange with Azure containers that have hierarchical namespaces enabled. (#29042)
Improve performance of queries with simple
ANDandORpredicates that have highly selective terms. (#24336)Reduce excessive memory usage caused by suboptimal join ordering for queries on columns with unknown statistics. (#29157)
Fix query failures and transaction errors caused by race conditions when dynamic catalogs are dropped concurrently with
system.jdbcorsystem.metadataqueries. (#28017)Fix incorrect results when using
json_parse()orJSONtype literals on documents containing numbers with more than 16 significant digits in the decimal portion. (#28867)Fix failure when executing table procedures on tables with uppercase column names. (#28970)
Fix failure when executing
ALTER TABLE EXECUTE OPTIMIZEwithORpredicates on partitionedtimestamp with time zonecolumns. (#27136)Fix incorrect freshness check for materialized views whose definition contains non-deterministic functions. (#28682)
Fix failure when executing
DESCRIBE OUTPUTwith[VERSION | TIMESTAMP] AS OFclauses. (#29077)
JDBC driver#
Add support for the
varianttype. (#29046)Add support for transparent OAuth2 token refresh when using the JDBC
accessTokenconnection property against a server configured withhttp-server.authentication.oauth2.refresh-tokens=true. Previously, such connections failed with401 Unauthorizedonce the embedded access token expired. (#29264)Fix failure when calling
setBigDecimal()with aBigDecimalvalue whosetoString()representation uses scientific notation, such as0E-10. (#23523)
CLI#
Add support for the
varianttype. (#29046)
ClickHouse connector#
Add support for reading all ClickHouse
DECIMALcolumns. (#28873)
Delta Lake connector#
⚠️ Breaking change: Remove legacy object storage support for Azure Storage, Google Cloud Storage, IBM Cloud Object Storage, S3, and S3-compatible systems. Use native file system support for object storage.
fs.hadoop.enabledapplies only to HDFS. See legacy file system support for migration details. (#24878)Add support for configuring the Azure connection-pool idle time and HTTP request timeout via the
azure.connection-pool-max-idle-timeandazure.http-request-timeoutconfiguration properties. (#29284)Reduce memory usage when writing files to S3. (#28488)
Reduce query planning time by reading metadata and protocol information from Delta checksum files. This optimization can be disabled by setting the
delta.load-metadata-from-checksum-fileconfiguration property orload_metadata_from_checksum_filesession property tofalse. (#28381)Fix
DELETEdeleting incorrect rows on Delta Lake tables backed by Parquet files with column indexes, particularly when the table was written by Apache Spark. (#28885)
Hive connector#
⚠️ Breaking change: Remove legacy object storage support for Azure Storage, Google Cloud Storage, IBM Cloud Object Storage, S3, and S3-compatible systems. Use native file system support for object storage.
fs.hadoop.enabledapplies only to HDFS. See legacy file system support for migration details. (#24878)Add support for Esri GeoJSON. (#28859)
Add support for configuring the Azure connection-pool idle time and HTTP request timeout via the
azure.connection-pool-max-idle-timeandazure.http-request-timeoutconfiguration properties. (#29284)Improve performance of queries with date partition predicates on Hive tables in the Glue catalog. (#28817)
Reduce memory usage when writing files to S3. (#28488)
Fix failure when creating tables in the Hive metastore version 3.1. (#28798)
Fix failure when parsing bucket numbers on tables backed by files without a bucket name. (#28632)
Hudi connector#
Add support for configuring the Azure connection-pool idle time and HTTP request timeout via the
azure.connection-pool-max-idle-timeandazure.http-request-timeoutconfiguration properties. (#29284)
Iceberg connector#
⚠️ Breaking change: Remove legacy object storage support for Azure Storage, Google Cloud Storage, IBM Cloud Object Storage, S3, and S3-compatible systems. Use native file system support for object storage.
fs.hadoop.enabledapplies only to HDFS. See legacy file system support for migration details. (#24878)Add experimental support for the
varianttype for Iceberg v3 tables. Older CLI versions will render the value as JSON unless updated. (#24538)Add support for reading and writing
timestamp(9)andtimestamp(9) with time zonein Iceberg v3 tables. (#27835)Add support for Azure vended credentials in Iceberg REST catalog. (#23238)
Add support for specifying HTTP headers in REST catalog. (#24236)
Add support for execution metrics while running the
add_filesandadd_files_from_tableprocedures. (#28996)Add support for execution metrics while running the
optimizeprocedure. (#28992)Add more columns to
$filessystem table. (#29044)Add
added_snapshot_idcolumn to$filessystem table. (#28911)Add support for configuring the Azure connection-pool idle time and HTTP request timeout via the
azure.connection-pool-max-idle-timeandazure.http-request-timeoutconfiguration properties. (#29284)Add support for refreshable vended credentials for the Iceberg REST catalog (S3, GCS, Azure). (#28998)
Reduce memory usage when writing files to S3. (#28488)
Fix failure when reading the
$entriesor$all_entriesmetadata tables. (#29211)
Lakehouse connector#
⚠️ Breaking change: Remove legacy object storage support for Azure Storage, Google Cloud Storage, IBM Cloud Object Storage, S3, and S3-compatible systems. Use native file system support for object storage.
fs.hadoop.enabledapplies only to HDFS. See legacy file system support for migration details. (#24878)Add support for configuring the Azure connection-pool idle time and HTTP request timeout via the
azure.connection-pool-max-idle-timeandazure.http-request-timeoutconfiguration properties. (#29284)Reduce memory usage when writing files to S3. (#28488)
MongoDB connector#
Fix incorrect results when reading JSON columns containing numbers with more than 16 significant digits in the decimal portion. (#28867)
MySQL connector#
Fix incorrect results when reading JSON columns containing numbers with more than 16 significant digits in the decimal portion. (#28867)
Pinot connector#
Fix incorrect results when reading JSON columns containing numbers with more than 16 significant digits in the decimal portion. (#28867)
PostgreSQL connector#
SingleStore connector#
Fix incorrect results when reading JSON columns containing numbers with more than 16 significant digits in the decimal portion. (#28867)
SQL Server connector#
SPI#
Add
varianttype defined by the Iceberg specification. (#24538)Add a
tableBranchparameter to theConnectorAccessControl.checkCanXxxtable-level methods. (#29179)Add support for pushing down
COALESCEexpressions in connectors. (#28984)⚠️ Breaking change: Remove the deprecated
getObjectandappendTomethods from theTypeinterface. (#29003)