20.38. Release 0.202

General Changes

  • Fix correctness issue for queries involving aggregations over the result of an outer join (#10592).
  • Fix map() to raise an error on duplicate keys rather than silently producing a corrupted map.
  • Fix map_from_entries() to raise an error when input array contains a null entry.
  • Fix out-of-memory error for bucketed execution by scheduling new splits on the same worker as the recently finished one.
  • Fix query failure when performing a GROUP BY on json or ipaddress types.
  • Fix correctness issue in line_locate_point(), ST_IsValid(), and geometry_invalid_reason() functions to not return values outside of the expected range.
  • Fix failure in geometry_to_bing_tiles() and ST_NumPoints() functions when processing geometry collections.
  • Fix query failure in aggregation spilling (#10587).
  • Remove support for SHOW PARTITIONS statement.
  • Improve support for correlated subqueries containing equality predicates.
  • Improve performance of correlated EXISTS subqueries.
  • Limit the number of grouping sets in a GROUP BY clause. The default limit is 2048 and can be set via the analyzer.max-grouping-sets configuration property or the max_grouping_sets session property.
  • Allow coercion between row types regardless of field names. Previously, a row type is coercible to another only if the field name in the source type matches the target type, or when target type has anonymous field name.
  • Increase default value for experimental.filter-and-project-min-output-page-size to 500kB.
  • Improve performance of equals operator on array(bigint) and array(double) types.
  • Respect X-Forwarded-Proto header in client protocol responses.
  • Add support for column-level access control. Connectors have not yet been updated to take advantage of this support.
  • Add support for correlated subqueries with correlated OR predicates.
  • Add multimap_from_entries() function.
  • Add bing_tiles_around(), ST_NumGeometries(), ST_GeometryN(), and ST_ConvexHull() geospatial functions.
  • Add wilson_interval_lower() and wilson_interval_upper() functions.
  • Add IS DISTINCT FROM for json and ipaddress type.

Hive Changes

  • Fix optimized ORC writer encoding of TIMESTAMP before 1970-01-01. Previously, the written value was off by one second.
  • Fix query failure when a Hive bucket has no splits. This commonly happens when a predicate filters some buckets out entirely.
  • Remove the hive.bucket-writing config property.
  • Add support for creating and writing bucketed sorted tables. The list of sorting columns may be specified using the sorted_by table property. Writing to sorted tables can be disabled using the hive.sorted-writing config property or the sorted_writing_enabled session property. The maximum number of temporary files for can be controlled using the hive.max-sort-files-per-bucket property.
  • Collect and store basic table statistics (rowCount, fileCount, rawDataSize, totalSize) when writing.
  • Add hive.orc.tiny-stripe-threshold config property and orc_tiny_stripe_threshold session property to control the stripe/file size threshold when ORC reader decides to read multiple consecutive stripes or entire fires at once. Previously, this feature piggybacks on other properties.

CLI Changes

  • Add peak memory usage to --debug output.

SPI Changes

  • Make PageSorter and PageIndexer supported interfaces.