Release 356 (30 Apr 2021)#
General#
- Add support for MATCH_RECOGNIZE. (#6111) 
- Introduce - system.metadata.materialized_view_propertiestable for listing available materialized view properties. (#7615)
- Add support for limiting the maximum planning time via the - query.max-planning-timeconfiguration property. (#7213)
- Allow redirecting clients to an alternative location to fetch query information. This can be configured via the - query.info-url-templateconfiguration property. (#7678)
- Allow cancellation of queries during planning phase. (#7213) 
- Improve performance of - ORDER BY ... LIMITqueries over a- LEFT JOIN. (#7028)
- Improve performance of queries with predicates on boolean columns. (#7263) 
- Improve planning time for queries with large - INpredicates. (#7556)
- Improve performance of queries that contain joins on - varcharkeys of different length. (#7644)
- Improve performance of queries when late materialization is enabled. (#7695) 
- Reduce coordinator network overhead when scheduling queries. (#7351) 
- Fix possible deadlock for - JOINqueries when spilling is enabled. (#7455)
- Fix incorrect results for queries containing full outer join with an input that is known to produce one row. (#7629) 
- Fix failure when quantified comparison expressions contain scalar subqueries. (#7792) 
Security#
- Materialized views require - UPDATEprivilege to be refreshed. (#7707)
- Add dedicated access control for creating and dropping materialized views. (#7645) 
- Add dedicated access control for refreshing materialized views. Insert privilege on storage table is no longer required. (#7707) 
- Fix authentication failure when providing multiple scope values for - http-server.authentication.oauth2.scopes. (#7706)
JDBC driver#
- Add support for caching OAuth2 credentials in memory to avoid unnecessary authentication flows. (#7309) 
BigQuery connector#
Cassandra connector#
- Fix - NullPointerExceptionwhen reading an empty timestamp value. (#7433)
Hive connector#
- Improve performance when reading dictionary-encoded Parquet files. (#7754) 
- Fix incorrect results when referencing nested fields with non-lowercase names from ORC files. (#7350) 
- Always use row-by-row deletes for ACID tables rather than dropping partitions. (#7621) 
- Allow reading from ORC ACID transactional tables when - _orc_acid_versionmetadata files are missing. (#7579)
- Add - UPDATEsupport for ACID tables that were originally created as non-transactional. (#7622)
- Add support for connection proxying for Azure ADLS endpoints. (#7509) 
Iceberg connector#
- Show Iceberg tables created by other engines in - SHOW TABLESoutput. (#1592)
- Improve performance when reading dictionary-encoded Parquet files. (#7754) 
- Improve query planning through table metadata caching. (#7336) 
- Fix failure querying materialized views that were created using the session catalog and schema. (#7711) 
- Fix listing of materialized views when using - SHOW TABLESquery. (#7790)
Kafka connector#
- Add support for TLS security protocol. (#6929) 
MemSQL connector#
- Improve metadata caching hit rate. (#7039) 
MongoDB connector#
- Fix handling of non-lowercase MongoDB views. (#7491) 
MySQL connector#
Oracle connector#
- Improve metadata caching hit rate. (#7039) 
Phoenix connector#
- Improve metadata caching hit rate. (#7039) 
PostgreSQL connector#
Redshift connector#
- Improve metadata caching hit rate. (#7039) 
SQL Server connector#
SPI#
- Require that - ConnectorMaterializedViewDefinitionprovides a view owner. (#7489)
- Add - Connector#getMaterializedViewPropertyManagerfor specifying materialized view properties. (#7615)
- Add - ConnectorAccessControl.checkCanCreateMaterializedView()and- ConnectorAccessControl.checkCanDropMaterializedView()for authorizing creation and removal of materialized views. (#7645)
- Allow a materialized view to return a storage table in a different catalog or schema. (#7638) 
- Add - ConnectorAccessControl.checkCanRefreshMaterializedView()for authorizing refresh of materialized views. (#7707)