Release 363 (6 Oct 2021)#
General#
- Add HTTP event listener implementation which sends JSON serialized events to a remote HTTP endpoint. (#8821) 
- Improve performance of queries that group by - bigintcolumns. (#9510)
- Improve performance of queries that process row or array data. (#9402) 
- Improve query planning performance. (#9462) 
- Reduce query memory usage when spilling occurs. (#9270, #9275) 
- Reduce memory usage for processing - JOINclauses. (#9327)
- Fix potential data loss in query results when clients retry requests to the coordinator. (#9453) 
- Fix incorrect result for comparisons between zero-valued decimals. (#8985) 
- Fix - SHOW ROLESfailure when there are no roles to display. (#9319)
- Fix - EXPLAIN ANALYZEto show estimates. (#9396)
- Fix incorrect result for - round()with precision set to 0. (#9371)
- Respect - deprecated.legacy-catalog-roles=trueconfiguration property in- SHOW ROLES,- SHOW CURRENT ROLESand- SHOW ROLE GRANTSstatements. (#9496)
Python client#
- Fix column type reported in - cursor.descriptionfor- time with time zonecolumn. (#9460)
BigQuery connector#
- Fix failure for queries where predicates on - geography,- arrayor- structcolumn are pushed down to BigQuery. (#9391)
Cassandra connector#
- Add support for Cassandra - tupletype. (#8570)
Elasticsearch connector#
- Add support for - scaled_floattype. (#9358)
Hive connector#
- Support usage of - avro_schema_urltable property in partitioned tables. (#9370}
- Add support for insert overwrite operations on S3-backed tables. (#9234) 
- Improve query performance when reading Parquet data with predicate on a - decimalcolumn. (#9338)
- Fix - Failed reading parquet data: Socket is closed by peerquery failure when reading from Parquet table with a predicate. (#9097)
- Fix query failure when updating or deleting from an ORC ACID transactional table that has some rows deleted since the last major compaction. (#9354) 
- Fix failure when reading large Parquet files. (#9469) 
- Fix failures for some - UPDATEqueries, such as those where the- SETclause contains the same constant more than once. (#9295)
- Fix incorrect results when filtering on Parquet columns containing a dot in their name. (#9516) 
Iceberg connector#
- Improve query performance when reading Parquet data with predicate on a - decimalcolumn. (#9338)
- Fix support for comments when adding a new column. Previously, they were silently ignored. (#9123) 
- Fix page and block sizing when writing Parquet data. (#9326) 
- Fix failure when reading large Parquet files. (#9469) 
MySQL connector#
Phoenix connector#
- Allowing forcing the mapping of certain types to - varchar. This can be enabled by setting the- jdbc-types-mapped-to-varcharconfiguration property to a comma-separated list of type names. (#2084)
Pinot connector#
- Fix failure when a column name is a reserved keyword. (#9373)