[DRILL-6417] Project is not pushed into scan if use subquery with UNION operator

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Won't Fix
• Affects Version/s: 1.14.0
• Assignee: Vitalii Diravka
• Reporter: Anton Gozhiy
• Created: 15/May/18 11:46
• Updated: 11/Jul/18 16:23
• Resolved: 11/Jul/18 15:14

Description

Data:
Use attached dataset

Query:

explain plan for select id
from 
  (select id, part_col, int_col, bool_col, date_col, float_col, time_col, ts_col
   from dfs.tmp.`DRILL_3855_test_data`
   where part_col = 'Partition_one' or part_col = 'Partition_two'
   union
   select id, part_col, int_col, bool_col, date_col, float_col, time_col, ts_col
   from dfs.tmp.`DRILL_3855_test_data`
   where part_col = 'Partition_two' or part_col = 'Partition_three')
where int_col = 0

Expected plan:

Scan ... columns=[`part_col`, `id`, `int_col`]

Actual plan:

Scan ... columns=[`part_col`, `id`, `int_col`, `bool_col`, `date_col`, `float_col`, `time_col`, `ts_col`]

Notes:
Works as expected if change "union" to "union all"

Issue Links

• relates to: DRILL-3130 Project can be pushed below union all / union to improve performance
• relates to: CALCITE-2410 Enable ProjectSetOpTransposeRule for UNION (distinct) SetOp operator