Release 350 (28 Dec 2020)#
General#
- Add HTTP client JMX metrics. (#6453) 
- Improve query performance by reducing worker to worker communication overhead. (#6283, #6349) 
- Improve performance of queries that contain - IS NOT DISTINCT FROMjoin predicates. (#6404)
- Fix failure when restricted columns have column masks. (#6017) 
- Fix failure when - tryexpressions reference columns that contain- @or- :in their names. (#6380)
- Fix memory management config handling to use - query.max-total-memory-per-noderather than only using- query.max-memory-per-nodefor both values. (#6349)
Web UI#
- Fix truncation of query text in cluster overview page. (#6216) 
JDBC driver#
- Accept - java.time.OffsetTimein- PreparedStatement.setObject(int, Object). (#6352)
- Extend - PreparedStatement.setObject(int, Object, int)to allow setting- time with time zoneand- timestamp with time zonevalues with precision higher than nanoseconds. This can be done via providing a- Stringvalue representing a valid SQL literal. (#6352)
BigQuery connector#
- Fix incorrect results for - count(*)queries with views. (#5635)
Cassandra connector#
- Support - DELETEstatement with primary key or partition key. (#4059)
Elasticsearch connector#
- Improve query analysis performance when Elasticsearch contains many index mappings. (#6368) 
Kafka connector#
- Support Kafka Schema Registry for Avro topics. (#6137) 
SQL Server connector#
- Add - data_compressiontable property to control the target compression in SQL Server. The allowed values are- NONE,- ROWor- PAGE. (#4693)
Other connectors#
This change applies to the MySQL, Oracle, PostgreSQL, Redshift, and SQL Server connectors.
- Send shorter and potentially more performant queries to remote database when a Presto query has a - NOT INpredicate eligible for pushdown into the connector. (#6075)
SPI#
- Rename - LongTimeWithTimeZone.getPicoSeconds()to- LongTimeWithTimeZone.getPicoseconds(). (#6354)