Release 310 (3 May 2019)#
General#
Reduce compilation failures for expressions over types containing an extremely large number of nested types. (#537)
Fix error reporting when query fails with due to running out of memory. (#696)
Improve performance of
JOIN
queries involving join keys of different types. (#665)Add initial and experimental support for late materialization. This feature can be enabled via
experimental.work-processor-pipelines
feature config or viawork_processor_pipelines
session config. Simple select queries of typeSELECT ... FROM table ORDER BY cols LIMIT n
can experience significant CPU and performance improvement. (#602)Add support for
FETCH FIRST
syntax. (#666)
CLI#
Make the final query time consistent with query stats. (#692)
Hive connector#
Ignore boolean column statistics when the count is
-1
. (#241)Prevent failures for
information_schema
queries when a table has an invalid storage format. (#568)Add support for assuming AWS role when accessing S3 or Glue. (#698)
Add support for coercions between
DECIMAL
,DOUBLE
, andREAL
for partition and table schema mismatch. (#352)Fix typo in Metastore recorder duration property name. (#711)
PostgreSQL connector#
Support for the
ARRAY
type has been disabled by default. (#687)
Blackhole connector#
Support having tables with same name in different Blackhole schemas. (#550)