[
https://issues.apache.org/jira/browse/HIVE-6687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13943398#comment-13943398
]
Laljo John Pullokkaran commented on HIVE-6687:
----------------------------------------------
Apparently view schema also uses same result set schema.
Modified patch to:
1. Separate out View Schema vs Result Set Schema.
2. View Schema won't use qualified table names. View schema would also ensure
that column names are unique.
3. ResultSet schema by default would use table aliases if provided (select *,
or user provided qualified projections select r1.x..)
4. To get old behavior for result set schema, introduced a config param
"hive.resultset.use.unique.column.names"; this is set to true by default. User
will have to set this to false for old behavior.
> JDBC ResultSet fails to get value by qualified projection name
> --------------------------------------------------------------
>
> Key: HIVE-6687
> URL: https://issues.apache.org/jira/browse/HIVE-6687
> Project: Hive
> Issue Type: Bug
> Components: HiveServer2
> Affects Versions: 0.12.0
> Reporter: Laljo John Pullokkaran
> Assignee: Laljo John Pullokkaran
> Fix For: 0.12.1
>
> Attachments: HIVE-6687.2.patch
>
>
> Getting value from result set using fully qualified name would throw
> exception. Only solution today is to use position of the column as opposed to
> column label.
> {code}
> String sql = "select r1.x, r2.x from r1 join r2 on r1.y=r2.y";
> ResultSet res = stmt.executeQuery(sql);
> res.getInt("r1.x");
> {code}
> res.getInt("r1.x"); would throw exception unknown column even though sql
> specifies it.
> Fix is to fix resultsetschema in semantic analyzer.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
