[ 
https://issues.apache.org/jira/browse/DRILL-712?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13997225#comment-13997225
 ] 

Steven Phillips commented on DRILL-712:
---------------------------------------

This is because we are using the same major type as in the input vectors for 
the output vectors in both HashJoin and MergeJoin.

The correct thing to do is, in a left join, if the incoming right side vector 
is REQUIRED, the output should be OPTIONAL, (and vice verse).

This is causing the count to include all of the records that should be null.

> Right side of Left join has zero values when should be null
> -----------------------------------------------------------
>
>                 Key: DRILL-712
>                 URL: https://issues.apache.org/jira/browse/DRILL-712
>             Project: Apache Drill
>          Issue Type: Bug
>            Reporter: Zhiyong Liu
>
> git.commit.id=e7a486d784c072458d44b7692ea0262da368f001
> 2014-05-13 11:16:03 INFO  QuerySubmitter:89 - Submitting query:               
>                            
> select count(*)                                                               
>                            
>   from (select l.l_orderkey as x, c.c_custkey as y                            
>                            
>         from lineitem l                                                       
>                            
>         left outer join customer c                                            
>                            
>                         on l.l_orderkey = c.c_custkey) as foo                 
>                            
>   where y < 10000                                                             
>                            
> 2014-05-13 11:16:10 INFO  DrillTestBase:201 - Query submit end time: 
> 2014/05/13 11:16:10.0010            
> 2014-05-13 11:16:10 INFO  DrillTestBase:202 - The execution time for the 
> query: 7 seconds.               
> 2014-05-13 11:16:10 INFO  TestVerifier:203 - These rows are not expected:     
>                            
> 2014-05-13 11:16:10 INFO  TestVerifier:206 -    60175 : 1 time(s).            
>                            
> 2014-05-13 11:16:10 INFO  TestVerifier:213 - Total number of unexpected rows: 
> 1                          
> 2014-05-13 11:16:10 INFO  TestVerifier:217 - These rows are expected but are 
> not in result set:          
> 2014-05-13 11:16:10 INFO  TestVerifier:220 -    1467 : 1 time(s).             
>                            
> 2014-05-13 11:16:10 INFO  TestVerifier:227 - Total number of expected but 
> missing: 1 
> also,resultset from drill is inconsistent.



--
This message was sent by Atlassian JIRA
(v6.2#6252)

Reply via email to