Add the query.dispatcher-query-pool-size configuration property to prevent
the coordinator from hanging when too many queries are being executed at once. (#20817)
Improve performance of queries selecting only catalog, schema, and name from
the system.metadata.materialized_views table. (#21448)
⚠️ Breaking change: Remove the deprecated legacy.materialized-view-grace-period
configuration property. (#21474)
Increase the number of columns supported by MERGE queries before failing
with a MethodTooLargeException error. (#21299)
Fix potential query hang when there is an error processing data. (#21397)
Fix possible worker crashes when running aggregation queries due to
out-of-memory error. (#21425)
Fix incorrect results when performing aggregations over null values. (#21457)
Fix failure for queries containing expressions involving types that do
not support the = operator (e.g., HyperLogLog, Geometry, etc.). (#21508)
Fix incorrect results for distinct count aggregations over a constant value. (#18562)
Fix sporadic query failure when filesystem caching is enabled. (#21342)
Fix unexpected failure for join queries containing predicates that might raise
an error for some inputs. (#21521)
Add support for creating views with custom properties. (#21401)
Add support for writing Bloom filters in Parquet files. (#20662)
⚠️ Breaking change: Remove the deprecated PARTITION_COLUMN and PARTITION_VALUE
arguments from the flush_metadata_cache procedure in favor of
PARTITION_COLUMNS and PARTITION_VALUES. (#21410)
Add support for pushing down execution of the variance, var_pop,
var_samp,covar_pop, covar_samp, corr, regr_intercept, and
regr_slope functions to the underlying database. (#21384)