Release 420 (22 Jun 2023)

General

• Add support for the any_value() aggregation function. (#17777)
• Add support for underscores in numeric literals. (#17776)
• Add support for hexadecimal, binary, and octal numeric literals. (#17776)
• Deprecate the dynamic-filtering.small-broadcast.* and dynamic-filtering.large-broadcast.* configuration properties in favor of dynamic-filtering.small.* and dynamic-filtering.large.*. (#17831)

Security

• Add support for configuring authorization rules for ALTER ... SET AUTHORIZATION... statements in file-based access control. (#16691)
• Remove the deprecated legacy.allow-set-view-authorization configuration property. (#16691)

BigQuery connector

• Fix direct download of access tokens, and correctly use the proxy when it is enabled with the bigquery.rpc-proxy.enabled configuration property. (#17783)

Delta Lake connector

• Add support for comments on view columns. (#17773)
• Add support for recalculating all statistics with an ANALYZE statement. (#15968)
• Disallow using the root directory of a bucket (scheme://authority) as a table location without a trailing slash in the location name. (#17921)
• Fix Parquet writer incompatibility with Apache Spark and Databricks Runtime. (#17978)

Druid connector

• Add support for tables with uppercase characters in their names. (#7197)

Hive connector

• Add a native Avro file format reader. This can be disabled with the avro.native-reader.enabled configuration property or the avro_native_reader_enabled session property. (#17221)
• Require admin role privileges to perform ALTER ... SET AUTHORIZATION... statements when the hive-security configuration property is set to sql-standard. (#16691)
• Improve query performance on partitioned Hive tables when table statistics are not available. (#17677)
• Disallow using the root directory of a bucket (scheme://authority) as a table location without a trailing slash in the location name. (#17921)
• Fix Parquet writer incompatibility with Apache Spark and Databricks Runtime. (#17978)
• Fix reading from a Hive table when its location is the root directory of an S3 bucket. (#17848)

Hudi connector

• Disallow using the root directory of a bucket (scheme://authority) as a table location without a trailing slash in the location name. (#17921)
• Fix Parquet writer incompatibility with Apache Spark and Databricks Runtime. (#17978)
• Fix failure when fetching table metadata for views. (#17901)

Iceberg connector

• Disallow using the root directory of a bucket (scheme://authority) as a table location without a trailing slash in the location name. (#17921)
• Fix Parquet writer incompatibility with Apache Spark and Databricks Runtime. (#17978)
• Fix scheduling failure when dynamic filtering is enabled. (#17871)

Kafka connector

• Fix server startup failure when a Kafka catalog is present. (#17299)

MongoDB connector

• Add support for ALTER TABLE ... RENAME COLUMN. (#17874)
• Fix incorrect results when the order of the dbref type fields is different from databaseName, collectionName, and id. (#17883)

SPI

• Move table function infrastructure to the io.trino.spi.function.table package. (#17774)