[jira] [Updated] (HIVE-6452) fix bug in UnsignedInt128.multiplyArrays4And4To8 and revert temporary fix in Decimal128.multiplyDestructive

["Eric Hanson (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/HIVE-6452?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Eric Hanson updated HIVE-6452:
------------------------------

    Attachment: multiplyArrays4And4To8-start.patch

> fix bug in UnsignedInt128.multiplyArrays4And4To8 and revert temporary fix in Decimal128.multiplyDestructive
> -----------------------------------------------------------------------------------------------------------
>
>                 Key: HIVE-6452
>                 URL: https://issues.apache.org/jira/browse/HIVE-6452
>             Project: Hive
>          Issue Type: Sub-task
>    Affects Versions: 0.13.0
>            Reporter: Eric Hanson
>         Attachments: multiplyArrays4And4To8-start.patch
>
>
> UnsignedInt128.multiplyArrays4And4To8 has a bug that causes rare multiply failures, one of which appears in TestDecimal128.testKnownPriorErrors.
> Fix the bug by finishing the TODO section in UnsignedInt128.multiplyArrays4And4To8 in the provided multiplyArrays4And4To8-start.patch. Make it fast and make it work with no per-operation storage allocations.
> Retain the rest of the work (the new tests) in multiplyArrays4And4To8-start.patch as much as possible.
> Revert the changes to Decimal128.multiplyDestructive so it doesn't use the short-term, slow fix based on HiveDecimal. I.e. use the implementation in multiplyDestructiveNativeDecimal128.



--
This message was sent by Atlassian JIRA
(v6.1.5#6160)