Release 305 (7 Mar 2019)#

General#

  • Fix failure of Regular expression functions for certain patterns and inputs when using the default JONI library. (#350)

  • Fix a rare ClassLoader related problem for plugins providing an EventListenerFactory. (#299)

  • Expose join_max_broadcast_table_size session property, which was previously hidden. (#346)

  • Improve performance of queries when spill is enabled but not triggered. (#315)

  • Consider estimated query peak memory when making cost based decisions. (#247)

  • Include revocable memory in total memory stats. (#273)

  • Add peak revocable memory to operator stats. (#273)

  • Add ST_Points() function to access vertices of a linestring. (#316)

  • Add a system table system.metadata.analyze_properties to list all ANALYZE properties. (#376)

Resource groups#

  • Fix resource group selection when selector uses regular expression variables. (#373)

Web UI#

  • Display peak revocable memory, current total memory, and peak total memory in detailed query view. (#273)

CLI#

  • Add option to output CSV without quotes. (#319)

Hive connector#

  • Fix handling of updated credentials for Google Cloud Storage (GCS). (#398)

  • Fix calculation of bucket number for timestamps that contain a non-zero milliseconds value. Previously, data would be written into the wrong bucket, or could be incorrectly skipped on read. (#366)

  • Allow writing ORC files compatible with Hive 2.0.0 to 2.2.0 by identifying the writer as an old version of Hive (rather than Presto) in the files. This can be enabled using the hive.orc.writer.use-legacy-version-number configuration property. (#353)

  • Support dictionary filtering for Parquet v2 files using RLE_DICTIONARY encoding. (#251)

  • Remove legacy writers for ORC and RCFile. (#353)

  • Remove support for the DWRF file format. (#353)

Base-JDBC connector library#

  • Allow access to extra credentials when opening a JDBC connection. (#281)