Travis Woodruff created PIG-3649:
------------------------------------
Summary: POPartialAgg incorrectly calculates size reduction when
multiple values aggregated
Key: PIG-3649
URL: https://issues.apache.org/jira/browse/PIG-3649
Project: Pig
Issue Type: Bug
Affects Versions: 0.11.1, 0.12.0, 0.11
Reporter: Travis Woodruff
{{POPartialAgg.aggregate()}} counts the number of output columns
({{valueTuple.size() - 1}}), but {{checkSizeReduction()}} compares this to the
number of input tuples.
When multiple columns are aggregated, this causes the reduction factor to be
calculated as too high by a factor of the number of columns being aggregated,
which causes in-memory aggregation to be disabled when it should not be,
adversely affecting performance,
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
