[HIVE-11459] LEFT SEMI JOIN with SELECT * should not include right-side non-key columns

• Type: Bug
• Status: Resolved
• Priority: Critical
• Resolution: Fixed
• Fix Version/s: 2.3.0
• Component/s: Hive
• Assignee: Jesús Camacho Rodríguez
• Reporter: Matt McCline
• Created: 04/Aug/15 22:02
• Updated: 27/Feb/24 22:24
• Resolved: 17/Aug/16 16:06

Description

select * from tmp2 left semi join tmp1 where c1 = id and c0 = q
tmp2.id	tmp2.d	tmp2.q	tmp1.c0	tmp1.d	tmp1.c1
1	^	20134	20134	^	1
3	^	20134	20134	^	3
1	^	20135	20135	^	1
1	^	20138	20138	^	1
2	^	20139	20139	^	2

tmp1 column "d" is a non-key and should not be included.

The LEFT SEMI JOIN clause is something Hive added. We ought to define the semantics here

Also see https://issues.apache.org/jira/browse/HIVE-11431.