11.155. Release 0.86

General Changes

  • Add support for inequality INNER JOIN when each term of the condition refers to only one side of the join.
  • Add ntile() function.
  • Add map() function to create a map from arrays of keys and values.
  • Add min_by() aggregation function.
  • Add support for concatenating arrays with the || operator.
  • Add support for = and != to JSON type.
  • Improve error message when DISTINCT is applied to types that are not comparable.
  • Perform type validation for IN expression where the right-hand side is a subquery expression.
  • Improve error message when ORDER BY … LIMIT query exceeds its maximum memory allocation.
  • Improve error message when types that are not orderable are used in an ORDER BY clause.
  • Improve error message when the types of the columns for subqueries of a UNION query don’t match.
  • Fix a regression where queries could be expired too soon on a highly loaded cluster.
  • Fix scheduling issue for queries involving tables from information_schema, which could result ininconsistent metadata.
  • Fix an issue with min_by() and max_by() that could result in an error when used witha variable-length type (e.g., VARCHAR) in a GROUP BY query.
  • Fix rendering of array attributes in JMX connector.
  • Input rows/bytes are now tracked properly for JOIN queries.
  • Fix case-sensitivity issue when resolving names of constant table expressions.
  • Fix unnesting arrays and maps that contain the ROW type.