[DRILL-2777] CTAS, order by and flatten of repeated list result in ExpandConversionRule error

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.2.0
• Component/s: Query Planning & Optimization
• Assignee: Jinfeng Ni
• Reporter: Rahul Kumar Challapalli
• Created: 14/Apr/15 00:44
• Updated: 15/Sep/15 17:56
• Resolved: 02/Jul/15 21:18

Description

git.commit.id.abbrev=f066786

The below query results in an error :

create table dfs.ctas_flatten.`test` as select uid, flatten(lst_lst) lst, flatten(lst_lst[0]) from `data1.json` order by flatten(lst_lst[0]);
Query failed: IndexOutOfBoundsException: index (2) must be less than size (2)

Error: exception while executing query: Failure while executing query. (state=,code=0)

The logs indicate an assertion error :

 Caused by: java.lang.AssertionError: Internal error: Error while applying rule ExpandConversionRule, args [rel#1757858:AbstractConverter.PHYSICAL.SINGLETON([]).[2](child=rel#1757856:Subset#8.PHYSICAL.ANY([]).[2],convention=PHYSICAL,DrillDistributionTraitDef=SINGLETON([]),sort=[2])]

The query itself does not have any issues. Also the below variants work :

1. No flatten of repeated list in the select clause
create table dfs.ctas_flatten.`test` as select uid, flatten(lst_lst[0]) from `data1.json` order by flatten(lst_lst[0]);
2. Removing the flatten from the order by. (Ordering by a scalar has no issues) 
create table dfs.ctas_flatten.`test` as select uid, flatten(lst_lst) lst, flatten(lst_lst[0]) from `data1.json`;

Attached the data set and the complete error log file