[ 
https://issues.apache.org/jira/browse/HIVE-4073?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13589999#comment-13589999
 ] 

Harish Butani commented on HIVE-4073:
-------------------------------------

The ptf_negative_AggrFuncsWithNoGBYNoPartDef query is

select p_mfgr, 
sum(p_retailprice) as s1  
from part

Which should fail, right?

Changing ensureWindowingSourceHasPartitioning may not work. Its tied to how we 
infer windowing for aggregation invocation that don't have an explicit over 
clause.

The more direct fix maybe to do the partition on constant idea in 
SemanticAnalyzer::processWindowSpec.
The ASTTree for over() is ^(TOK_WINDOWSPEC) so this function should be invoked.
If hasPartSpec is false add the PartitionSpec on '0'
Does this make sense? 

Changing ensureWindowingSourceHasPartitioning will probably also affect how we 
handle multiple partitions. So i have another reason to try to avoid changing 
its logic :)
                
> Make partition by optional in over clause
> -----------------------------------------
>
>                 Key: HIVE-4073
>                 URL: https://issues.apache.org/jira/browse/HIVE-4073
>             Project: Hive
>          Issue Type: Bug
>          Components: PTF-Windowing
>            Reporter: Ashutosh Chauhan
>            Assignee: Brock Noland
>         Attachments: HIVE-4073-0.patch, HIVE-4073-1.patch
>
>
> select s, sum( i ) over() from tt; should work. 

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Reply via email to