Release 446 (1 May 2024)#
General#
Improve performance of
INSERT
statements into partitioned tables when theretry_policy
configuration property is set toTASK
. (#21661)Improve performance of queries with complex grouping operations. (#21726)
Reduce delay before killing queries when the cluster runs out of memory. (#21719)
Prevent assigning null values to non-null columns as part of a
MERGE
statement. (#21619)Fix
CREATE CATALOG
statements including quotes in catalog names. (#21399)Fix potential query failure when a column name ends with a
:
. (#21676)Fix potential query failure when a SQL routine contains a label reference in a
LEAVE
,ITERATE
,REPEAT
, orWHILE
statement. (#21682)Fix query failure when SQL routines use the
NULLIF
orBETWEEN
functions. (#19820)Fix potential query failure due to worker nodes running out of memory in concurrent scenarios. (#21706)
BigQuery connector#
Improve performance when listing table comments. (#21581)
⚠️ Breaking change: Enable
bigquery.arrow-serialization.enabled
by default. This requires--add-opens=java.base/java.nio=ALL-UNNAMED
injvm-config
. (#21580)
Delta Lake connector#
Fix failure when reading from Azure file storage and the schema, table, or column name contains non-alphanumeric characters. (#21586)
Fix incorrect results when reading a partitioned table with a deletion vector. (#21737)
Hive connector#
Iceberg connector#
Add support for the Snowflake catalog. (#19362)
Automatically use
varchar
as a type during table creation whenchar
is specified. (#19336, #21515)Deprecate the
schema
andtable
arguments for thetable_changes
function in favor ofschema_name
andtable_name
, respectively. (#21698)Fix failure when executing the
migrate
procedure with partitioned Hive tables on Glue. (#21391)Fix failure when reading from Azure file storage and the schema, table, or column name contains non-alphanumeric characters. (#21586)
Pinot connector#
Fix query failure when a predicate contains a
'
. (#21681)
Snowflake connector#
Add support for the
unsupported-type-handling
andjdbc-types-mapped-to-varchar
type mapping configuration properties. (#21528)
SPI#
Remove support for
@RemoveInput
as an annotation for aggregation functions. AWindowAggregation
can be declared in@AggregationFunction
instead, which supports input removal. (#21349)Extend
QueryCompletionEvent
with various aggregated, per-stage, per-task distribution statistics. New information is available inQueryCompletedEvent.statistics.taskStatistics
. (#21694)