Release 0.155¶
General Changes¶
Fix incorrect results when queries contain multiple grouping sets that resolve to the same set.
Fix incorrect results when using
map
withIN
predicates.Fix compile failure for outer joins that have a complex join criteria.
Fix error messages for failures during commit.
Fix memory accounting for simple aggregation, top N and distinct queries. These queries may now report higher memory usage than before.
Reduce unnecessary memory usage of
map_agg()
,multimap_agg()
andmap_union()
.Make
INCLUDING
,EXCLUDING
andPROPERTIES
non-reserved keywords.Remove support for the experimental feature to compute approximate queries based on sampled tables.
Properly account for time spent creating page source.
Various optimizations to reduce coordinator CPU usage.
Hive Changes¶
Fix schema evolution support in new Parquet reader.
Fix
NoClassDefFoundError
when using Hadoop KMS.Add support for Avro file format.
Always produce dictionary blocks for DWRF dictionary encoded streams.
SPI Changes¶
Remove legacy connector API.