[IMPALA-3861] Cannot resolve correlated column reference inside a BETWEEN predicate.

• Type: Bug
• Status: Resolved
• Priority: Blocker
• Resolution: Fixed
• Affects Version/s: Impala 2.3.0, Impala 2.5.0, Impala 2.4.0, Impala 2.6.0
• Fix Version/s: Impala 2.7.0, Impala 2.6.2
• Component/s: Frontend
• Labels: ramp-up
• Target Version: Impala 2.7.0, Impala 2.6.2
• Assignee: Alexander Behm
• Reporter: Alexander Behm
• Created: 15/Jul/16 05:46
• Updated: 04/Jan/17 23:58
• Resolved: 30/Jul/16 18:25

Description

Impala fails to properly resolve correlated column references in a BETWEEN predicate.
This is a regression from Impala 2.2, introduced in

Consider the following query with a correlated BETWEEN predicate in an IN subquery:

select 1 from functional.alltypes t1
where t1.id in
  (select test_id from functional.jointbl t2
   where t2.test_id between t1.tinyint_col and t1.int_col)
ERROR: AnalysisException: Could not resolve column/field reference: 't1.tinyint_col'

select 1 from functional.alltypes t1
where t1.id in
  (select test_id from functional.jointbl t2
   where t2.test_id between tinyint_col and int_col)
ERROR: AnalysisException: Could not resolve column/field reference: 'tinyint_col'

Workaround
Rewrite the BETWEEN predicate as an explicit range condition:

select 1 from functional.alltypes t1
where t1.id in
  (select test_id from functional.jointbl t2
   where t2.test_id >= t1.tinyint_col and t2.test_id <= t1.int_col)