12.93. Release 0.147

General Changes

  • Fix race condition that can cause queries that process data from non-columnardata sources to fail.
  • Fix incorrect formatting of dates and timestamps before year 1680.
  • Fix handling of syntax errors when parsing EXTRACT.
  • Fix potential scheduling deadlock for connectors that expose node-partitioned data.
  • Fix performance regression that increased planning time.
  • Fix incorrect results for grouping sets for some queries with filters.
  • Add SHOW CREATE VIEW and SHOW CREATE TABLE.
  • Add support for column aliases in WITH clause.
  • Support LIKE clause for SHOW CATALOGS and SHOW SCHEMAS.
  • Add support for INTERSECT.
  • Add support for casting row types.
  • Add sequence() function.
  • Add sign() function.
  • Add flatten() function.
  • Add experimental implementation of resource groups.
  • Add Local File Connector.
  • Remove experimental intermediate aggregation optimizer. The optimizer.use-intermediate-aggregationsconfig option and task_intermediate_aggregation session property are no longer supported.
  • Add support for colocated joins for connectors that expose node-partitioned data.
  • Improve the performance of array_intersect().
  • Generalize the intra-node parallel execution system to work with all query stages.The task.concurrency configuration property replaces the old task.join-concurrencyand task.default-concurrency options. Similarly, the task_concurrency sessionproperty replaces the task_join_concurrency, task_hash_build_concurrency, andtask_aggregation_concurrency properties.

Hive Changes

  • Fix reading symlinks when the target is in a different HDFS instance.
  • Fix NoClassDefFoundError for SubnetUtils in HDFS client.
  • Fix error when reading from Hive tables with inconsistent bucketing metadata.
  • Correctly report read bytes when reading Parquet data.
  • Include path in unrecoverable S3 exception messages.
  • When replacing an existing Presto view, update the view datain the Hive metastore rather than dropping and recreating it.
  • Rename table property clustered_by to bucketed_by.
  • Add support for varchar(n).

Kafka Changes

  • Fix error code 6 when reading data from Kafka.
  • Add support for varchar(n).

Redis Changes

  • Add support for varchar(n).

MySQL and PostgreSQL Changes

  • Cleanup temporary data when a CREATE TABLE AS fails.
  • Add support for varchar(n).