[jira] [Commented] (HIVE-4320) Consider extending max limit for precision to 38

["Gunther Hagleitner (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HIVE-4320?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13628542#comment-13628542 ] 

Gunther Hagleitner commented on HIVE-4320:
------------------------------------------

Thought about this some more and did some testing. Comparing storing two's complement in two longs v packing 9 digits shouldn't have a performance impact. The latter method does seem make implementation easier (add/sub will not overflow, multiplication won't overflow in long). So, all in all it's better to be aligned with other DBs than gain a little more leeway in implementing these operations.
                
> Consider extending max limit for precision to 38
> ------------------------------------------------
>
>                 Key: HIVE-4320
>                 URL: https://issues.apache.org/jira/browse/HIVE-4320
>             Project: Hive
>          Issue Type: Bug
>            Reporter: Gunther Hagleitner
>            Assignee: Gunther Hagleitner
>
> Max precision of 38 still fits in 128. It changes the way you do math on these numbers though. Need to see if there will be perf implications, but there's a strong case to support 38 (instead of 36) to comply with other DBs. (Oracle, SQL Server, Teradata).

--
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