[ 
https://issues.apache.org/jira/browse/HIVE-621?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12729376#action_12729376
 ] 

Adam Kramer commented on HIVE-621:
----------------------------------

Actually, this behavior appears to have just changed also--I did an INSERT 
OVERWRITE LOCAL DIRECTORY, and the result was delimited by ^A *and* \t 
characters (like, a key/value pairing...first column, ^A, other columns).

Inserting into a local directory should DEFINITELY use the same delimiter to 
separate all columns. Failure to do this is likely a bug. The "wish" is that it 
also be consistent with SELECT output.

> Inconsistent reporting of "null" and delimiter across access types
> ------------------------------------------------------------------
>
>                 Key: HIVE-621
>                 URL: https://issues.apache.org/jira/browse/HIVE-621
>             Project: Hadoop Hive
>          Issue Type: Wish
>            Reporter: Adam Kramer
>
> If you INSERT OVERWRITE LOCAL DIRECTORY, you get a bunch of \N fields in your 
> output...If you hive -e "SELECT ..." you get a bunch of "null" fields in your 
> output. I also recall having seen NULL (in caps) before somewhere, too, but 
> maybe I'm dreaming that.
> In any case, it would be nice if this were consistent! Preference: \N 
> everywhere.
> Ditto, when you I O L D, your fields are ^A-delimited, but when you SELECT, 
> they're ^V (tab)-delimited. It would be nice if this, too, were consistent. 
> Preference: tabs everywhere.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.

Reply via email to