[
https://issues.apache.org/jira/browse/HIVE-25?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12663560#action_12663560
]
Prasad Chakka commented on HIVE-25:
-----------------------------------
I will change Hive.g so that sortOrder in DDL can only have identifiers and not
column refs.
The way I want to do this is to allow select columns to be referenced in
cluster/distributed/sort clauses according to following rules
1) column has an alias in the select clause. this column can only be referred
with the alias in the above 3 clauses
2) column doesn't have an alias but it may have table alias. this column can be
referred with the columnName or tabAlias.columnName in the above 3 clauses.
where clause can only contain tabAlias.columnName and is different from above
rules.
I will upload modified patch with tests for all the above conditions. How about
it?
> [Hive] cluster by does not work with column aliasing
> ----------------------------------------------------
>
> Key: HIVE-25
> URL: https://issues.apache.org/jira/browse/HIVE-25
> Project: Hadoop Hive
> Issue Type: Bug
> Components: Query Processor
> Reporter: Namit Jain
> Assignee: Prasad Chakka
> Priority: Critical
> Fix For: 0.2.0
>
> Attachments: hive-25-code-changes.patch, hive-25.patch
>
>
> SELECT b.*
> FROM
> (
> SELECT x.col1, x.col2
> FROM tmp x
> WHERE x.col2= 10
> CLUSTER BY col2
> ) b"
> The above query works, but if I change the cluster by to:
> CLUSTER BY x.col2,
> it croaks.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
