[ 
https://issues.apache.org/jira/browse/HIVE-17235?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16117875#comment-16117875
 ] 

Gopal V commented on HIVE-17235:
--------------------------------

[~owen.omalley]: the LongColumnVector clone needs re-implementation of all 
Decimal operators to be useful.

Right now, that change adds an extra interface to test/debug with no actual 
operators to use for the end to end processing.

I like the idea of a fast-path, but I'd rather not tackle the schema evolution 
problems right away (i.e the planner thinks it wants a Decimal(25,4), but the 
data is a Decimal(18,2) etc).

We've never had that problem before where there are potentially two different 
vector column types for the same underlying column, which complicates things 
with no perceivable gains right now.

I've moved the patch over to HIVE-17266 so that we can tackle it in a later 
iteration.

> Add ORC Decimal64 Serialization/Deserialization (Part 1)
> --------------------------------------------------------
>
>                 Key: HIVE-17235
>                 URL: https://issues.apache.org/jira/browse/HIVE-17235
>             Project: Hive
>          Issue Type: Bug
>          Components: Hive
>            Reporter: Matt McCline
>            Assignee: Matt McCline
>            Priority: Critical
>         Attachments: HIVE-17235.03.patch, HIVE-17235.04.patch, 
> HIVE-17235.05.patch, HIVE-17235.06.patch, HIVE-17235.patch
>
>
> The storage-api changes for ORC-209.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

Reply via email to