Skip to main content

Release 480 (24 Mar 2026)

General​

  • Add the number type. (#28319)
  • Add coordinator and worker counts to /metrics endpoint. (#27408)
  • Allow configuring the maximum amount of memory to use while writing tables through the task.scale-writers.max-writer-memory-percentage configuration property. (#27874)
  • Add variant of array_first() for finding the first element that matches a predicate. (#27706)
  • Add DataSketches functions. (#27563)
  • ⚠️ Breaking change: Remove enable-large-dynamic-filters configuration property and the corresponding system session property enable_large_dynamic_filters. (#27637)
  • ⚠️ Breaking change: Remove the dynamic-filtering.small* and dynamic-filtering.large-broadcast* configuration properties. (#27637)
  • ⚠️ Breaking change: Remove deprecated.http-server.authentication.oauth2.groups-field configuration property. (#28646)
  • Improve performance for remote data exchanges on newer CPU architectures and Graviton 4 CPUs. (#27586)
  • Improve performance of queries with remote data exchanges or aggregations. (#27657)
  • Reduce out-of-memory errors for queries involving window functions when spilling is enabled. (#27873)
  • Improve query performance when exchange manager is configured but query does not use task level retries. (#28698)
  • Allow query parameters within table [VERSION | TIMESTAMP] AS OF clause. (#28681)
  • Fix incorrect results when using localtimestamp() with precision 3. (#27806)
  • Fix localtimestamp() failure for precisions 7 and 8. (#27807)
  • Fix spurious query failures when querying the system catalog during catalog drop operations. (#28017)
  • Fix failure when executing date_add() function with a value greater than Integer.MAX_VALUE. (#27899)
  • Fix incorrect results when the result of casting json, time, boolean or interval values to varchar(n) doesn't fit in the target type. (#552)
  • Fix query failures and increased latency when using Azure exchange manager. (#28058)

Web UI​

  • Add cluster status info to the header in the preview UI. (#27712)
  • Sort stages in the query details page numerically rather than alphabetically. (#27655)

JDBC driver​

  • Return the correct class name for map, row, time with time zone, timestamp with time zone, varbinary and null values when calling ResultSetMetaData.getColumnClassName. (#28314)

ClickHouse connector​

  • Fix failure when creating a table if a prior CREATE TABLE ... AS SELECT operation for the same table failed. (#27702)

Delta Lake connector​

  • ⚠️ Breaking change: Remove live files table metadata cache. The configuration properties metadata.live-files.cache-size, metadata.live-files.cache-ttl and checkpoint-filtering.enabled are now defunct and must be removed from server configurations. (#27618)
  • ⚠️ Breaking change: Remove the hive.write-validation-threads configuration property. (#27729)
  • ⚠️ Breaking change: Remove the parquet.optimized-writer.validation-percentage configuration property, use parquet.writer.validation-percentage, instead. (#27729)
  • ⚠️ Breaking change: Remove the hive.parquet.writer.block-size configuration property, use parquet.writer.block-size, instead. (#27729)
  • ⚠️ Breaking change: Remove the hive.parquet.writer.page-size configuration property, use parquet.writer.page-size, instead. (#27729)
  • ⚠️ Breaking change: Remove the gcs.use-access-token configuration property. (#26941)
  • ⚠️ Breaking change: Remove the hive.fs.new-file-inherit-ownership configuration property. (#28029)
  • Improve the effectiveness of Bloom filters for high-cardinality columns in Parquet files. (#27656)
  • Remove the requirement for the PutObjectTagging AWS S3 permission when writing to Delta Lake tables on S3. (#27701)
  • Fix potential table corruption when executing CREATE OR REPLACE with table definition changes. (#27805)
  • Fix query failures and increased latency when using Azure file system. (#28058)
  • Fix failure when the file path contains # in GCS. (#28292)
  • Fix NPE when loading parquet column index with non stats supported column. (#28560)

DuckDB connector​

  • Fix failure when creating a table if a prior CREATE TABLE ... AS SELECT operation for the same table failed. (#27702)

Hive connector​

  • Add support for reading nanosecond-precision timestamps from Parquet files into timestamp(p) with time zone columns. (#27861)
  • ⚠️ Breaking change: Remove the hive.write-validation-threads configuration property. (#27729)
  • ⚠️ Breaking change: Remove the parquet.optimized-writer.validation-percentage configuration property, use the parquet.writer.validation-percentage, instead. (#27729)
  • ⚠️ Breaking change: Remove the hive.parquet.writer.block-size configuration property, use parquet.writer.block-size, instead. (#27729)
  • ⚠️ Breaking change: Remove the hive.parquet.writer.page-size configuration property, use parquet.writer.page-size, instead. (#27729)
  • ⚠️ Breaking change: Remove the deprecated gcs.use-access-token configuration property. (#26941)
  • ⚠️ Breaking change: Remove the hive.fs.new-file-inherit-ownership configuration property. (#28029)
  • Improve the effectiveness of Bloom filters for high-cardinality columns in Parquet files. (#27656)
  • Fix query failures and increased latency when using Azure file system. (#28058)
  • Fix incorrect memory accounting for INSERT queries targeting bucketed and sorted tables. (#28315)
  • Fix INSERT into and ANALYZE on tables with timestamp columns when using Hive metastore version 4. (#26214, #28330)

Hudi connector​

Iceberg connector​

  • Add support for the BigLake metastore in Iceberg REST catalog. (#26219)
  • Add delete_after_commit_enabled and max_previous_versions table properties. (#14128)
  • Add support for column default values in Iceberg v3 tables. (#27837)
  • Add support for creating, writing to or deleting from Iceberg v3 tables. (#27786, #27788)
  • Add support for Iceberg v3 tables in optimize, expire_snapshots and remove_orphan_files table procedures. (#27836)
  • Add support for Iceberg v3 row lineage. (#27836)
  • Add content column to $manifests and $all_manifests metadata tables. (#27975)
  • Add support for changing map and array nested types through ALTER ... SET DATA TYPE. (#27998)
  • Add support for creating materialized views with columns of number type. (#28399)
  • Clean up unused files from materialized views when they are refreshed. (#28008)
  • ⚠️ Breaking change: Remove the hive.write-validation-threads configuration property. (#27729)
  • ⚠️ Breaking change: Remove the parquet.optimized-writer.validation-percentage configuration property, use parquet.writer.validation-percentage, instead. (#27729)
  • ⚠️ Breaking change: Remove the hive.parquet.writer.block-size configuration property, use parquet.writer.block-size, instead. (#27729)
  • ⚠️ Breaking change: Remove the hive.parquet.writer.page-size configuration property, use parquet.writer.page-size, instead. (#27729)
  • ⚠️ Breaking change: Remove the deprecated gcs.use-access-token configuration property. (#26941)
  • ⚠️ Breaking change: Remove the hive.fs.new-file-inherit-ownership configuration property. (#28029)
  • ⚠️ Breaking change: Remove support for the iceberg.extended-statistics.enabled configuration option and extended_statistics_enabled session property. (#27914)
  • Improve the effectiveness of Bloom filters for high-cardinality columns in Parquet files. (#27656)
  • Improve query performance when querying a fresh materialized view. (#27608)
  • Improve optimize to clean up partition scoped equality delete files when a partition filter is used. (#28371)
  • Enhance optimize_manifests to cluster manifests by partition, improving read performance for queries that apply partition filters. (#27358)
  • Reduce planning time of queries on tables containing delete files. (#27955)
  • Reduce planning time for queries involving simple FROM and WHERE clauses. (#27973)
  • Reduce query planning time on large tables. (#28068)
  • Add support for temporary GCS credentials provided by REST catalog. (#24518)
  • Fix failure when reading $files metadata tables when scheme involving bucket or truncate changes. (#26109)
  • Fix failure when reading $file_modified_time metadata column on tables with equality deletes. (#27850)
  • Avoid large footers in Parquet files from certain rare string inputs. (#27903)
  • Fix failures for queries with joins on metadata columns. (#27984)
  • Fix query failures and increased latency when using Azure file system. (#28058)
  • Fix incorrect memory accounting for INSERT queries targeting bucketed and sorted tables. (#28315)
  • Fix an issue where using ALTER TABLE ... SET PROPERTIES to set partition spec unintentionally removed existing partition columns from the partition spec. (#26492)
  • Fix failures when reading from tables with write.parquet.compression-codec property set to LZ4. (#28291)
  • Fix value of compression-codec table property written by Trino to be compliant with Iceberg spec. (#28293)
  • Fix failure when the file path contains # in GCS. (#28292)
  • Fix failure when reading tables with iceberg.jdbc-catalog.schema-version=V0. (#28419)
  • Avoid worker crashes when reading from tables with a larger number of equality deletes. (#28468)
  • Fix failure when querying $files table after CREATE OR REPLACE TABLE with a different partitioning schema. (#25339)

Ignite connector​

  • Fix failure when creating a table if a prior CREATE TABLE ... AS SELECT operation for the same table failed. (#27702)

Lakehouse connector​

  • Improved performance and memory usage when Equality Delete files are used (#28507)
  • Fix failure when reading Iceberg $files tables. (#26751)

MariaDB connector​

  • Add support for reading MariaDB DECIMAL(p, s) columns when p > 38. (#28744)
  • Fix failure when creating a table if a prior CREATE TABLE ... AS SELECT operation for the same table failed. (#27702)

MySQL connector​

  • Add support for reading MySQL DECIMAL(p, s) columns when p > 38. (#28744)
  • ⚠️ Breaking change: Remove incorrect support for reading MySQL BIT(n) columns when n > 1. (#28744)
  • Fix failure when creating a table if a prior CREATE TABLE ... AS SELECT operation for the same table failed. (#27702)

Oracle connector​

  • Add support for configuring the connection wait timeout through the oracle.connection-pool.wait-timeout catalog property. (#27744)
  • Add support for reading all Oracle NUMBER columns. (#28747, #28401)
  • Fix failure when creating a table if a prior CREATE TABLE ... AS SELECT operation for the same table failed. (#27702)
  • Fix failure when reading float type in query table function. (#27880)

PostgreSQL connector​

  • Add support for reading all PostgreSQL NUMERIC and DECIMAL columns. (#28141)
  • Fix failure when creating a table if a prior CREATE TABLE ... AS SELECT operation for the same table failed. (#27702)

Redshift connector​

  • Fix failure when creating a table if a prior CREATE TABLE ... AS SELECT operation for the same table failed. (#27702)

SingleStore connector​

  • Remove incorrect support for reading SingleStore BIT(n) columns when n > 1. (#28744)
  • Fix failure when creating a table if a prior CREATE TABLE ... AS SELECT operation for the same table failed. (#27702)

Snowflake connector​

  • Fix failure when creating a table if a prior CREATE TABLE ... AS SELECT operation for the same table failed. (#27702)

SQL Server connector​

  • Fix failure when creating a table if a prior CREATE TABLE ... AS SELECT operation for the same table failed. (#27702)

Vertica connector​

SPI​

  • Remove support for TypeSignatureParameter. Use TypeParameter instead. (#27574)
  • Remove support for ParameterKind. Use TypeParameter.Type, TypeParameter.Numeric, and TypeParameter.Variable instead. (#27574)
  • Remove support for NamedType, NamedTypeSignature and NamedTypeParameter. Use TypeParameter.Type instead. (#27574)
  • Deprecate MaterializedViewFreshness#getLastFreshTime. Use getLastKnownFreshTime instead. (#27803)
  • Change ColumnMetadata.comment and ColumnMetadata.extraInfo to Optional<String>. (#28151)