Release 338 (07 Jul 2020)#

General Changes#

  • Fix incorrect results when joining tables on a masked column. (#4251)

  • Fix planning failure when multiple columns have a mask. (#4322)

  • Fix incorrect comparison for TIMESTAMP WITH TIME ZONE values with precision larger than 3. (#4305)

  • Fix incorrect rounding for timestamps before 1970-01-01. (#4370)

  • Fix query failure when using VALUES with a floating point NaN value. (#4119)

  • Fix query failure when joining tables on a real or double column and one of the joined tables contains NaN value. (#4272)

  • Fix unauthorized error for internal requests to management endpoints. (#4304)

  • Fix memory leak while using dynamic filtering. (#4228)

  • Improve dynamic partition pruning for broadcast joins. (#4262)

  • Add support for setting column comments via the COMMENT ON COLUMN syntax. (#2516)

  • Add compatibility mode for legacy clients when rendering datetime type names with default precision in information_schema tables. This can be enabled via the deprecated.omit-datetime-type-precision configuration property or omit_datetime_type_precision session property. (#4349, #4377)

  • Enforce NOT NULL column declarations when writing data. (#4144)

JDBC Driver Changes#

  • Fix excessive CPU usage when reading query results. (#3928)

  • Implement DatabaseMetaData.getClientInfoProperties(). (#4318)

Elasticsearch Connector Changes#

  • Add support for reading numeric values encoded as strings. (#4341)

Hive Connector Changes#

  • Fix incorrect query results when Parquet file has no min/max statistics for an integral column. (#4200)

  • Fix query failure when reading from a table partitioned on a real or double column containing a NaN value. (#4266)

  • Fix sporadic failure when writing to bucketed sorted tables on S3. (#2296)

  • Fix handling of strings when translating Hive views. (#3266)

  • Do not require cache directories to be configured on coordinator. (#3987, #4280)

  • Fix Azure ADL caching support. (#4240)

  • Add support for setting column comments. (#2516)

  • Add hidden $partition column for partitioned tables that contains the partition name. (#3582)

Kafka Connector Changes#

  • Fix query failure when a column is projected and also referenced in a query predicate when reading from Kafka topic using RAW decoder. (#4183)

MySQL Connector Changes#

  • Fix type mapping for unsigned integer types. (#4187)

Oracle Connector Changes#

  • Exclude internal schemas (e.g., sys) from schema listings. (#3784)

  • Add support for connection pooling. (#3770)

Base-JDBC Connector Library Changes#

  • Exclude the underlying database‚Äôs information_schema from schema listings. (#3834)