[
https://issues.apache.org/jira/browse/HIVE-955?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
He Yongqiang updated HIVE-955:
------------------------------
Attachment: hive-955-2009-11-30.patch
The basic idea is to pass column needed into ColumnarStruct, and let the init
method of ColumnarStruct only init those columns. The columns needed by the
query is stored in TBS, so we need to get them out and set into the
configuration. We did that when we call getRecordReader, where we clone a conf,
and set column ids into the conf, and then pass the conf into native
RecordReader. We need to do it over again when we init the operator tree. At
this time, we need to pass a conf with these column information into the serde
layer.
thanks, Namit!
> Passing columns needed onto ColumnarStruct
> ------------------------------------------
>
> Key: HIVE-955
> URL: https://issues.apache.org/jira/browse/HIVE-955
> Project: Hadoop Hive
> Issue Type: Improvement
> Reporter: He Yongqiang
> Assignee: He Yongqiang
> Attachments: ColumnarStruct_columnids_2009-11-25.2.patch,
> hive-955-2009-11-30.patch
>
>
> This can avoid unneeded loop cycle in ColumnarStruct's init method. This can
> saves a lot when the table's column number is big.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
