Add support for a GRACE PERIOD clause in the CREATE MATERIALIZED VIEW
task. For backwards compatibility, the existing materialized views are
interpreted as having a GRACE PERIOD of zero, however, new materialized
views have an unlimited grace period by default. This is a backwards
incompatible change, and the previous behavior can be restored with the
legacy.materialized-view-grace-period configuration property or the
legacy_materialized_view_grace_period session property. (#15842)
Fix potential incorrect query stats when tasks are waiting on running drivers
to fully terminate. (#15478)
Add support for specifying the number of nodes that will write data during
INSERT, CREATE TABLE ... AS SELECT, or EXECUTE queries with the
query.max-writer-tasks-count configuration property. (#16238)
Improve performance of queries that contain predicates involving the year
function. (#14078)
Improve performance of queries that contain a sum aggregation. (#16624)
Improve performance of filter function on arrays. (#16681)
Add a native OpenX JSON file format reader and writer. These can be disabled
with the openx_json_native_reader_enabled and
openx_json_native_writer_enabled session properties or the
openx-json.native-reader.enabled and openx-json.native-writer.enabled
configuration properties. (#16073)
Add support for implicit coercions between char types of different lengths. (#16402)
Improve performance of queries with joins where both sides of a join have keys
with the same table bucketing definition. (#16381)
Improve query planning performance for queries scanning tables with a large
number of columns. (#16203)
Improve scan performance for COUNT(*) queries on row-oriented formats. (#16595)
Ensure the value of the hive.metastore-stats-cache-ttl configuration
property always is greater than or equal to the value specified in the
hive.metastore-cache-ttl configuration property. (#16625)
Skip listing Glue metastore tables with invalid column types. (#16677)
Fix query failure when a file that is using a text file format with a
single header row that is large enough to be split into multiple files. (#16492)
Fix potential query failure when Kerberos is enabled and the query execution
takes longer than a Kerberos ticket's lifetime. (#16680)