Release 0.142¶
General Changes¶
Fix planning bug for
JOIN
criteria that optimizes to aFALSE
expression.Fix planning bug when the output of
UNION
doesn’t match the table column order inINSERT
queries.Fix error when
ORDER BY
clause in window specification refers to the same column multiple times.Add support for complex grouping operations -
CUBE
,ROLLUP
andGROUPING SETS
.Add support for
IF NOT EXISTS
inCREATE TABLE AS
queries.Add
substring()
function.Add
http.server.authentication.krb5.keytab
config option to set the location of the Kerberos keytab file explicitly.Add
optimize_metadata_queries
session property to enable the metadata-only query optimization.Improve support for non-equality predicates in
JOIN
criteria.Add support for non-correlated subqueries in aggregation queries.
Improve performance of
json_extract()
.
Hive Changes¶
Change ORC input format to report actual bytes read as opposed to estimated bytes.
Fix cache invalidation when renaming tables.
Fix Parquet reader to handle uppercase column names.
Fix issue where the
hive.respect-table-format
config option was being ignored.Add hive.compression-codec config option to control compression used when writing. The default is now
GZIP
for all formats.Collect and expose end-to-end execution time JMX metric for requests to AWS services.