20.19. Release 0.195-e
Introduce cost-based plan optimizations.
distributed_joinsession property with
reorder_joinssession property with
Introduce distributed sort. It is now possible to turn on distributed sort via experimental
distributed_sortsession property (off by default). Distributed sort will help in cases when sorted data doesn’t fit on single node. Performance improvement can be expected, but partially distributed data streams are still merged on a single node.
Fix sort memory leak.
Add support for ROLE management including CREATE ROLE, DROP ROLE, GRANT ROLE, REVOKE ROLE, SET ROLE, SHOW CURRENT ROLES, SHOW ROLES and SHOW ROLE GRANTS commands.
Support prepared statements that are longer than 4K bytes.
Support predicate pushdown for the
<column> IN <values list>predicate where values in the values list require casting to match the type of column.
Improve predicate inferring/pushdown between source and filtering subquery for queries like
SELECT ... WHERE <symbol> IN <subquery>.
Hive Connector changes
Allow partitions without files for bucketed tables (via
Allow multiple files per bucket for bucketed tables (via
hive.multi-file-bucketing.enabled). There must be one or more files per bucket. File names must match the Hive naming convention.
Allow reading incompletely bucketed tables with missing files (via
Allow disabling ability to create external tables from Presto (via
Implement ROLE management support.
JMX Connector changes
Add wildcard character
*which allows to query several MBeans data with single query.
Secured internal cluster communication over HTTPS using Kerberos or LDAP authentication
Add a Kerberos principal rule for exact match to File Based System Access Control
Data type changes
TIMESTAMP semantic changes
localtimefunctions were fixed to return the correct value for non-UTC timezones.
DECIMALthe default literal for non-integral numbers.
Fix decimal precision of
Fix returned value from
dis a negative integer. Previously, no rounding was done in this case.