[SPARK-18577] Ambiguous reference with duplicate column names in aggregate

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Won't Fix
• Affects Version/s: 1.6.2
• Component/s: SQL
• Assignee: Unassigned
• Reporter: Yerui Sun
• Created: 24/Nov/16 12:27
• Updated: 28/Nov/16 10:49
• Resolved: 28/Nov/16 10:49

Description

Assuming we have a table 't' with 3 columns 'id', 'name' and 'rank', and here's the sql to re-produce issue:

select id, count(*) from t t1 join t t2 on t1.name = t2.name group by t1.id

The error message is:

Reference 'id' is ambiguous, could be: id#3, id#9.; line 1 pos 7

The sql can be parsed in Hive, since the select 'id' reference can be resolved to 't1.id', which presented in group expressions.