Release 436 (11 Jan 2024)

General

• ⚠️ Breaking change: Require JDK 21.0.1 to run Trino, including updated JVM config. (#20010)

• Improve performance by not generating redundant predicates. (#16520)

• Fix query failure when invoking the json_table function. (#20122)

• Fix query hang when a SQL routine dereferences a row field. (#19997).

• Fix potential incorrect results when using the ST_Centroid() and ST_Buffer() functions for tiny geometries. (#20237)

Delta Lake connector

• Add support for querying files with corrupt or incorrect statistics, which can be enabled with the parquet_ignore_statistics catalog session property. (#20228)

• Improve performance of queries with selective joins on partition columns. (#20261)

• Reduce the number of requests made to AWS Glue when listing tables, schemas, or functions. (#20189)

• Fix incorrect results when querying Parquet files containing column indexes when the query has filters on multiple columns. (#20267)

ElasticSearch connector

• ⚠️ Breaking change: Add support for ElasticSearch version 8, and remove support for ElasticSearch version 6. (#20258)

• Add OpenSearch connector. (#11377)

Hive connector

• Reduce the number of requests made to AWS Glue when listing tables, schemas, or functions. (#20189)

• Fix failure when reading certain Avro data with Union data types. (#20233)

• Fix incorrect results when querying Parquet files containing column indexes when the query has filters on multiple columns. (#20267)

Hudi connector

• Add support for enforcing that a filter on a partition key must be present in the query. This can be enabled by with the hudi.query-partition-filter-required configuration property or the query_partition_filter_required catalog session property. (#19906)

• Fix incorrect results when querying Parquet files containing column indexes when the query has filters on multiple columns. (#20267)

Iceberg connector

• Add support for querying files with corrupt or incorrect statistics, which can be enabled with the parquet_ignore_statistics catalog session property. (#20228)

• Improve performance of queries with selective joins on partition columns. (#20212)

• Reduce the number of requests made to AWS Glue when listing tables, schemas, or functions. (#20189)

• Fix potential loss of data when running multiple INSERT queries at the same time. (#20092)

• Fix incorrect results when providing a nonexistent namespace while listing namespaces. (#19980)

• Fix predicate pushdown not running for Parquet files when columns have been renamed. (#18855)

SQL Server connector

• Fix incorrect results for DATETIMEOFFSET values before the year 1400. (#16559)