[
https://issues.apache.org/jira/browse/ORC-49?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16457099#comment-16457099
]
ASF GitHub Bot commented on ORC-49:
-----------------------------------
Github user omalley commented on the issue:
https://github.com/apache/orc/pull/257
Part of what we absolutely need in RLEv3 is zero suppression for the whole
row batch. So that if you have integers like:
- 1000
- 2000
- 3000
- 10000
- 4000
- 1000
it will record 3 for the number of zeros and encode:
- 1
- 2
- 3
- 10
- 4
- 1
> Improve decimal encoding to use RLE
> -----------------------------------
>
> Key: ORC-49
> URL: https://issues.apache.org/jira/browse/ORC-49
> Project: ORC
> Issue Type: Sub-task
> Reporter: Owen O'Malley
> Assignee: Owen O'Malley
> Priority: Major
>
> Because decimal was added in Hive 0.11 with no maximum precision, the ORC
> encoding for decimal uses arbitrary var ints. Since Hive 0.12, the maximum
> precision for decimals fits in 128 bits and thus we can use a much better
> encoding using the long rle.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
