Release 0.184

    • 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 queries with gathering exchanges.

    • Require to have at least two arguments, as mandated by the SQL standard.

    JDBC Driver Changes

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

    • 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.

    • Fix display of table statistics when running .

    SPI Changes

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