[DRILL-4902] nested aggregate query does not complain about missing GROUP BY clause

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Cannot Reproduce
• Affects Version/s: 1.9.0
• Fix Version/s: 1.16.0
• Component/s: Execution - Flow
• Labels: window_function
• Assignee: Gautam Parai
• Reporter: Khurram Faraaz
• Created: 23/Sep/16 06:10
• Updated: 03/Mar/19 14:10
• Resolved: 03/Mar/19 14:10

Description

A nested aggregate windowed query does not report an error when the partitioning column is not used in the GROUP BY clause.
Drill 1.9.0

This is the correct expected behavior.

0: jdbc:drill:schema=dfs.tmp> select count(max(c7)) over (partition by c8) from `DRILL_4589`;
Error: VALIDATION ERROR: From line 1, column 42 to line 1, column 43: Expression 'c8' is not being grouped

SQL Query null

[Error Id: 09c837b9-7a66-4a1f-9fbc-522160947274 on centos-01.qa.lab:31010] (state=,code=0)

The below query too should report above error, as the GROUP BY on partitioning column is missing.

0: jdbc:drill:schema=dfs.tmp> select count(max(c7)) over (partition by c8) from (select * from `DRILL_4589`);
+---------+
| EXPR$0  |
+---------+
| 1       |
+---------+
1 row selected (193.71 seconds)

Postgres 9.3 also reports an error for a similar query

postgres=# select count(max(c1)) over (partition by c2) from (select * from t222) sub_query;
ERROR:  column "sub_query.c2" must appear in the GROUP BY clause or be used in an aggregate function
LINE 1: select count(max(c1)) over (partition by c2) from (select * ...