General Changes

• Fix correctness issue for certain correlated join queries when the correlated subquery on the right produces no rows. (#1969)
• Fix incorrect handling of multi-byte characters for Regular Expression Functions when the pattern is empty. (#2313)
• Fix failure when join criteria contains columns of different types. (#2320)
• Fix failure for complex outer join queries when dynamic filtering is enabled. (#2363)
• Improve support for correlated queries. (#1969)
• Allow inserting values of a larger type into as smaller type when the values fit. For example, BIGINT into SMALLINT, or VARCHAR(10) into VARCHAR(3). Values that don’t fit will cause an error at runtime. (#2061)
• Add regexp_count() and regexp_position() functions. (#2136)

Security Changes

• Fix a security issue allowing users to gain unauthorized access to Presto cluster when using password authenticator with LDAP. (#2356)
• Add support for LDAP referrals in LDAP password authenticator. (#2354)

JDBC Driver

• Fix behavior of java.sql.Connection#commit() and java.sql.Connection#rollback() methods when no statements performed in a transaction. Previously, these methods would fail. (#2339)
• Fix failure when restoring autocommit mode with java.sql.Connection#setAutocommit() (#2338)

Hive Connector Changes

• Reduce query latency and Hive metastore load when using the AUTOMATIC join reordering strategy. (#2184)
• Allow configuring hive.max-outstanding-splits-size to values larger than 2GB. (#2395)
• Avoid redundant file system stat call when writing Parquet files. (#1746)
• Avoid retrying permanent errors for S3-related services such as STS. (#2331)

Kafka Connector Changes

• Remove internal columns: _segment_start, _segment_end and _segment_count. (#2303)
• Add new configuration property kafka.messages-per-split to control how many Kafka messages will be processed by a single Presto split. (#2303)

Elasticsearch Connector Changes

• Fix query failure when an object in an Elasticsearch document does not have any fields. (#2217)
• Add support for querying index aliases. (#2324)

Phoenix Connector Changes

• Add support for mapping unsupported data types to VARCHAR. This can be enabled by setting the unsupported-type-handling configuration property or the unsupported_type_handling session property to CONVERT_TO_VARCHAR. (#2427)

Other Connector Changes

These changes apply to the MySQL, PostgreSQL, Redshift and SQL Server connectors:

• Add support for creating schemas. (#1874)
• Add support for caching metadata. The configuration property metadata.cache-ttl controls how long to cache data (it defaults to 0ms which disables caching), and metadata.cache-missing controls whether or not missing tables are cached. (#2290)

This change applies to the MySQL and PostgreSQL connectors:

• Add support for mapping DECIMAL types with precision larger than 38 to Presto DECIMAL. (#2088)