Fix planning failure for queries that use GROUPING and contain aggregation expressions
that require implicit coercions.
Fix planning failure for queries that contains a non-equi left join that is semantically
equivalent to an inner join.
Fix issue where a query may have a reported memory that is higher than actual usage when
an aggregation is followed by other non-trivial work in the same stage. This can lead to failures
due to query memory limit, or lower cluster throughput due to perceived insufficient memory.
Fix query failure for CHAR functions trim(), rtrim(), and substr() when
the return value would have trailing spaces under VARCHAR semantics.
Fix formatting in EXPLAINANALYZE output.
Improve error message when a query contains an unsupported form of correlated subquery.
Improve performance of CAST(json_parse(...)AS...).
Change spilling for aggregations to only occur when the cluster runs out of memory.
Remove the experimental.operator-memory-limit-before-spill config property
and the operator_memory_limit_before_spill session property.
Allow configuring the amount of memory that can be used for merging spilled aggregation data
from disk using the experimental.aggregation-operator-unspill-memory-limit config
property or the aggregation_operator_unspill_memory_limit session property.
Work around ORC-222 which results in
invalid summary statistics in ORC or DWRF files when the input data contains invalid string data.
Previously, this would usually cause the query to fail, but in rare cases it could
cause wrong results by incorrectly skipping data based on the invalid statistics.
Fix issue where reported memory is lower than actual usage for table columns containing
string values read from ORC or DWRF files. This can lead to high GC overhead or out-of-memory crash.
Improve error message for small ORC files that are completely corrupt or not actually ORC.
Add predicate pushdown for the hidden column "$path".