12.9. Release 325 (14 Nov 2019)#
Warning
There is a performance regression in this release.
General Changes#
- Fix incorrect results for certain queries involving
FULLorRIGHTjoins andLATERAL. (#1952) - Fix incorrect results when using
IS DISTINCT FROMon columns ofDECIMALtype with precision larger than 18. (#1985) - Fix query failure when row types contain a field named after a reserved SQL keyword. (#1963)
- Add support for
LIKEpredicate toSHOW SESSIONandSHOW FUNCTIONS. (#1688, #1692) - Add support for late materialization to join operations. (#1256)
- Reduce number of metadata queries during planning.
This change disables stats collection for non-
EXPLAINqueries. If you want to have access to such stats and cost in query completion events, you need to re-enable stats collection using thecollect-plan-statistics-for-all-queriesconfiguration property. (#1866) - Add variant of
strpos()that returns the Nth occurrence of a substring. (#1811) - Add
to_encoded_polyline()andfrom_encoded_polyline()geospatial functions. (#1827)
Hive Changes#
- Fix incorrect behavior of
CREATE TABLEwhen Hive metastore is configured withmetastore.create.as.acidset totrue. (#1958) - Fix query failure when reading Parquet files that contain character data without statistics. (#1955)
- Allow analyzing a subset of table columns (rather than all columns). (#1907)
- Support overwriting unpartitioned tables for insert queries when using AWS Glue. (#1243)
- Add support for reading Parquet files where the declared precision of decimal columns does not match the precision in the table or partition schema. (#1949)
- Improve performance when reading Parquet files with small row groups. (#1925)