Release 339 (21 Jul 2020)#
General#
Add
approx_most_frequent()
. (#3425)Physical bytes scan limit for queries can be configured via
query.max-scan-physical-bytes
configuration property andquery_max_scan_physical_bytes
session property. (#4075)Remove support for addition and subtraction between
TIME
andINTERVAL YEAR TO MONTH
types. (#4308)Fix planning failure when join criteria contains subqueries. (#4380)
Fix failure when subquery appear in window function arguments. (#4127)
Fix failure when subquery in
WITH
clause contains hidden columns. (#4423)Fix failure when referring to type names with different case in a
GROUP BY
clause. (#2960)Fix failure for queries involving
DISTINCT
when expressions inORDER BY
clause differ by case from expressions inSELECT
clause. (#4233)Fix incorrect type reporting for
TIMESTAMP
andTIMESTAMP WITH TIME ZONE
for legacy clients. (#4408)Fix failure when querying nested
TIMESTAMP
orTIMESTAMP WITH TIME ZONE
for legacy clients. (#4475, #4425)Fix failure when parsing timestamps with time zone with an offset of the form
+NNNN
. (#4490)
JDBC driver#
Fix reading
TIMESTAMP
andTIMESTAMP WITH TIME ZONE
values with a negative year or a year higher than 9999. (#4364)Fix incorrect column size metadata for
TIMESTAMP
andTIMESTAMP WITH TIME ZONE
types. (#4411)Return correct value from
ResultSet.getDate()
,ResultSet.getTime()
andResultSet.getTimestamp()
methods when session zone is set to a different zone than the default zone of the JVM the JDBC is run in. The previous behavior can temporarily be restored usinguseSessionTimeZone
JDBC connection parameter. (#4017)
Druid connector#
Fix handling of table and column names containing non-ASCII characters. (#4312)
Hive connector#
Make
location
parameter optional for thesystem.register_partition
procedure. (#4443)Avoid creating tiny splits at the end of block boundaries. (#4485)
Remove requirement to configure
metastore.storage.schema.reader.impl
in Hive 3.x metastore to let Presto access CSV tables. (#4457)Fail query if there are bucket files outside of the bucket range. Previously, these extra files were skipped. (#4378)
Fix a query failure when reading from Parquet file containing
real
ordouble
NaN
values, if the file was written by a non-conforming writer. (#4267)
Kafka connector#
Kudu connector#
Add support for grouped execution. It can be enabled with the
kudu.grouped-execution.enabled
configuration property or thegrouped_execution
session property. (#3715)
MongoDB connector#
Allow querying Azure Cosmos DB. (#4415)
Oracle connector#
Allow providing credentials via the
connection-user
andconnection-password
configuration properties. These properties were previously ignored if connection pooling was enabled. (#4430)
Phoenix connector#
Fix handling of row key definition with white space. (#3251)
SPI#
Allow connectors to wait for dynamic filters before splits are generated via the new
DynamicFilter
object passed toConnectorSplitManager.getSplits()
. (#4224)