Vuk Ercegovac created IMPALA-6392:
-------------------------------------
Summary: Explain format for parquet predicate statistics should be
consistent with predicates
Key: IMPALA-6392
URL: https://issues.apache.org/jira/browse/IMPALA-6392
Project: IMPALA
Issue Type: Bug
Components: Frontend
Affects Versions: Impala 2.11.0
Reporter: Vuk Ercegovac
Priority: Minor
In EXPLAIN_LEVEL 2, "parquet predicate statistics" combine conjuncts from
multiple tuple descriptors into one line:
parquet statistics predicates: c_custkey > 0, o.o_orderkey > 0, l.l_partkey
> 0
However, "predicates" separate the conjuncts into a line per tuple descriptor:
predicates: c_custkey > 0, !empty(c.c_orders)
predicates on o: !empty(o.o_lineitems), o.o_orderkey > 0
predicates on l: l.l_partkey > 0
These two cases should be consistent. Prefer the line per tuple descriptor
style.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
