[
https://issues.apache.org/jira/browse/HIVE-7832?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Prasanth J updated HIVE-7832:
-----------------------------
Attachment: HIVE-7832.3.patch
Following changes should fix the failing tests
1) flushDictionary() is updated to add row index entries for the flushed rows
2) Doing dictionary check before writing stripe for cases where number of rows
in stripe is less than the dictionary check after rows config.
> Do ORC dictionary check at a finer level and preserve encoding across stripes
> -----------------------------------------------------------------------------
>
> Key: HIVE-7832
> URL: https://issues.apache.org/jira/browse/HIVE-7832
> Project: Hive
> Issue Type: Improvement
> Affects Versions: 0.14.0
> Reporter: Prasanth J
> Assignee: Prasanth J
> Attachments: HIVE-7832.1.patch, HIVE-7832.2.patch, HIVE-7832.3.patch
>
>
> Currently ORC dictionary check happens while writing the stripe. Just before
> writing stripe if ratio of dictionary entries to total non-null rows is
> greater than threshold then the dictionary is discarded. Also, the decision
> of using dictionary or not is preserved across stripes. This sometimes leads
> to costly insertion cost of O(logn) for each stripes when there are too many
> distinct keys.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
