[jira] [Commented] (HIVE-17321) HoS: analyze ORC table doesn't compute raw data size when noscan/partialscan is not specified

[liyunzhang_intel (JIRA)]

    [ 
https://issues.apache.org/jira/browse/HIVE-17321?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16129923#comment-16129923
 ] 

liyunzhang_intel commented on HIVE-17321:
-----------------------------------------

[~lirui]: patch looks good. But I  have 1 question， why the statistic of 
limitpushdown.q changes? 
before
{code}
    if ((OrcInputFormat.class.isAssignableFrom(inputFormat) ||
          MapredParquetInputFormat.class.isAssignableFrom(inputFormat)) && 
(noScan || partialScan)) {

{code}

Now
{code}
  if ((OrcInputFormat.class.isAssignableFrom(inputFormat) ||
          MapredParquetInputFormat.class.isAssignableFrom(inputFormat))
{code}
If the InputFormat is TextFile, i think your patch will not change the result.  
 If my understanding is not right, tell me.

> HoS: analyze ORC table doesn't compute raw data size when noscan/partialscan 
> is not specified
> ---------------------------------------------------------------------------------------------
>
>                 Key: HIVE-17321
>                 URL: https://issues.apache.org/jira/browse/HIVE-17321
>             Project: Hive
>          Issue Type: Bug
>          Components: Spark
>            Reporter: Rui Li
>            Assignee: Rui Li
>            Priority: Minor
>         Attachments: HIVE-17321.1.patch, HIVE-17321.2.patch
>
>
> Need to implement HIVE-9560 for Spark.



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