21.37. Release 0.184

General Changes

  • Fix query execution failure for split_to_map(...)[...].
  • Fix issue that caused queries containing CROSS JOIN to continue using CPU resources even after they were killed.
  • Fix planning failure for some query shapes containing count(*) and a non-empty GROUP BY clause.
  • Fix communication failures caused by lock contention in the local scheduler.
  • Improve performance of element_at() for maps to be constant time rather than proportional to the size of the map.
  • Improve performance of queries with gathering exchanges.
  • Require coalesce() to have at least two arguments, as mandated by the SQL standard.
  • Add hamming_distance() function.

JDBC Driver Changes

  • Always invoke the progress callback with the final stats at query completion.

Web UI Changes

  • Add worker status page with information about currently running threads and resource utilization (CPU, heap, memory pools). This page is accessible by clicking a hostname on a query task list.

Hive Changes

  • Fix partition filtering for keys of CHAR, DECIMAL, or DATE type.
  • Reduce system memory usage when reading table columns containing string values from ORC or DWRF files. This can prevent high GC overhead or out-of-memory crashes.

TPCDS Changes

  • Fix display of table statistics when running SHOW STATS FOR ....

SPI Changes

  • Row columns or values represented with ArrayBlock and InterleavedBlock are no longer supported. They must be represented as RowBlock or SingleRowBlock.
  • Add source field to ConnectorSession.