Release 483 (17 Jul 2026)#
General#
Add support for the PIVOT clause to turn row values into columns. (#29404)
Add support for reading values from a
jsoncolumn using the JSON simplified accessor syntax, with dotted and subscripted notation such asSELECT j.customer.name FROM ordersorSELECT j.items[0].price.decimal(18,2) FROM orders. (#29565)Add support for
SELECT j.*over ajsoncolumn to return its top-level members as a single JSON array. (#29565)Add support for the
OVERLAPSpredicate to test whether two time periods share any instant. See Date and time functions and operators. (#29652)Add support for the
datetime()method in SQL/JSON path expressions. (#29603)Add the
title_case()function. (#2942)Add the
ST_Transform()andST_TransformXY()functions for transforminggeometryvalues between coordinate reference systems. (#29737)Add support for constructing and inspecting three-dimensional points with the new three- and four-argument
ST_Point()overloads and theST_Z(),ST_Force2D(), andST_Force3D()functions. (#29737)Add the
ST_GeomFromEWKT()function for parsing extended well-known text. (#29737)Add the
ST_PointOnSurface(),ST_MakePolygon(),ST_Multi(),ST_LineMerge(),ST_Normalize(),ST_ReducePrecision(),ST_VoronoiPolygons(),ST_MinimumBoundingCircle(), andST_OrientedEnvelope()functions. (#29737)Add the
ST_Collect(),ST_MakeLine(), andST_Polygonize()functions and thegeometry_collect_agg()aggregate function for combininggeometryvalues. (#29737)Add support for tuning the HTTP connection pool used for Azure exchange spooling storage with the
exchange.azure.max-connections,exchange.azure.pending-acquire-max-count, andexchange.azure.connection-acquisition-timeoutconfiguration properties. (#30083)Preserve
geometrySRID and Z-coordinate metadata across serialization, format conversions, and supported geospatial operations. (#29737)Improve performance of queries with a
BETWEENpredicate that compares a cast expression against a range of literals, such asCAST(ts AS date) BETWEEN DATE '2020-01-01' AND DATE '2020-01-02'. (#14648)Improve performance of joins whose condition contains an equality shared by every branch of an
IForCASEexpression, such asIF(k IS NULL, a = b, a = b AND c = d). (#30288)Reduce worker memory usage for queries that use dynamic filtering on
bigint,integer, ordatejoin keys. (#30321)Fix incorrect results for
NOT BETWEENpredicates that compare a cast expression against a range of literals. (#30073)Fix potential incorrect results when comparing a non-deterministic expression cast to
dateagainst adateliteral, such asCAST(f() AS date) = DATE '2020-01-01'. (#30286)Fix incorrect results and
MERGEfailures for queries containing a right or full outer join. (#30278)Fix query failure when a filter contains a row constructor with more than 64 fields referencing input columns. (#30279)
Fix query failure when spooling query results that contain rows with highly variable sizes. (#30112)
Fix query failure during fault-tolerant execution when processing small data pages. (#23531)
Fix query failure caused by an optimizer timeout when comparing a cast or a date and time function against a literal. (#30347)
Fix failure of
SHOW SESSIONand similar metadata queries when a catalog fails to load. (#30086)Fix query failure when calling
theta_sketch_union()with explicit nominal entries and seed arguments. (#29725)
Security#
Add support for restricting OAuth 2.0 single sign-on to accounts from a specific domain, configurable with the
http-server.authentication.oauth2.domain-hintconfiguration property. (#30272)
Web UI#
Add support for sorting the query list by progress. (#30350)
⚠️ Breaking change: Make the redesigned Web UI, previously served at
/ui/preview, the default at/ui, and move the previous Web UI to/ui/legacy. The previous Web UI is deactivated by default and can be restored by settingweb-ui.legacy.enabledtotrue. (#22697)Fix query state filter checkboxes not toggling when clicking the item text in the preview UI. (#30078)
JDBC driver#
Prevent JDBC queries from hanging indefinitely when an error occurs while reading query results. (#30007)
CLI#
Delta Lake connector#
Add support for configuring the S3 authentication type with the
s3.auth-typeconfiguration property. UseANONYMOUSfor reading public buckets. (#27512)⚠️ Breaking change: Require
s3.auth-type=IAM_ROLEwhens3.iam-roleis set. (#27512)⚠️ Breaking change: Remove the
s3.use-web-identity-token-credentials-providerconfiguration property. Uses3.auth-type=WEB_IDENTITYinstead. (#27512)Reduce memory usage when writing Parquet files. (#30372)
Prevent orphaned deletion vector files after a failed write. (#30342)
Prevent deletion vectors written by Trino from being rejected by other Delta Lake readers. (#30355)
Druid connector#
Hive connector#
Add support for configuring the S3 authentication type with the
s3.auth-typeconfiguration property. UseANONYMOUSfor reading public buckets. (#27512)⚠️ Breaking change: Require
s3.auth-type=IAM_ROLEwhens3.iam-roleis set. (#27512)⚠️ Breaking change: Remove the
s3.use-web-identity-token-credentials-providerconfiguration property. Uses3.auth-type=WEB_IDENTITYinstead. (#27512)Improve performance of reading TEXTFILE and CSV data. (#30291)
Reduce memory usage when writing Parquet files. (#30372)
Fix incorrect results for queries with a filter on
charcolumns in ORC files. (#30187)
Hudi connector#
Add support for limiting the maximum size of splits when reading Hudi tables with the
hudi.max-split-sizeconfiguration property. (#29842)Add support for configuring the S3 authentication type with the
s3.auth-typeconfiguration property. UseANONYMOUSfor reading public buckets. (#27512)⚠️ Breaking change: Require
s3.auth-type=IAM_ROLEwhens3.iam-roleis set. (#27512)⚠️ Breaking change: Remove the
s3.use-web-identity-token-credentials-providerconfiguration property. Uses3.auth-type=WEB_IDENTITYinstead. (#27512)Improve performance of reading Hudi tables. (#29842)
Iceberg connector#
Add support for reading Iceberg tables with encrypted Parquet files. (#28204)
Add support for querying metadata tables, such as
$files,$snapshots, and$partitions, for materialized views. (#30218)Add support for configuring the S3 authentication type with the
s3.auth-typeconfiguration property. UseANONYMOUSfor reading public buckets. (#27512)Allow using Application Default Credentials with the Iceberg REST catalog by making the
gcs.json-key-file-pathconfiguration property optional wheniceberg.rest-catalog.securityis set toGOOGLE. (#29084)⚠️ Breaking change: Require
s3.auth-type=IAM_ROLEwhens3.iam-roleis set. (#27512)⚠️ Breaking change: Remove the
s3.use-web-identity-token-credentials-providerconfiguration property. Uses3.auth-type=WEB_IDENTITYinstead. (#27512)Reduce memory usage when writing position delete files in Iceberg tables. (#30081)
Reduce memory usage when writing Parquet files. (#30372)
Fix incorrect results for queries with a filter on
charcolumns in ORC files. (#30187)Prevent table corruption by failing
OPTIMIZEwhen aNOT NULLcolumn containsNULLvalues. (#30062)Fix potential loss of table metadata when a
CREATE TABLEcommit to the Hive metastore catalog fails after the table was created. (#29973)Fix incorrect listing of views from unrelated schemas when using an Iceberg JDBC catalog. (#29738)
Fix
CREATE TABLEfailure when using a catalog that generates the table location, such as an Iceberg REST catalog. (#26742)Fix failure when setting a column comment containing characters rejected by the Glue catalog, such as newlines. (#30353)
Fix query failure when using an Iceberg REST catalog with SigV4 request signing. (#30163)
Ignite connector#
⚠️ Breaking change: Require the additional
--add-opens=java.base/java.util=ALL-UNNAMEDJVM configuration option when using the Ignite connector. (#30255)
Lakehouse connector#
Add support for reading Iceberg tables with encrypted Parquet files. (#28204)
Add support for configuring the S3 authentication type with the
s3.auth-typeconfiguration property. UseANONYMOUSfor reading public buckets. (#27512)⚠️ Breaking change: Require
s3.auth-type=IAM_ROLEwhens3.iam-roleis set. (#27512)⚠️ Breaking change: Remove the
s3.use-web-identity-token-credentials-providerconfiguration property. Uses3.auth-type=WEB_IDENTITYinstead. (#27512)Reduce memory usage when writing Parquet files. (#30372)
Fix incorrect results for queries with a filter on
charcolumns in ORC files. (#30187)
SQL Server connector#
Fix query failure when reading a table with hidden columns using a
LIMIT. (#30084)
SPI#
Consistently expose raw array and offset accessors on all block types. (#30128)
⚠️ Breaking change: Change the block null representation to bit-packed validity bitmaps instead of boolean null arrays. Plugins that construct blocks directly instead of through the block builder API must be updated. (#30185)