[jira] [Commented] (HIVE-15300) Reuse table information in SemanticAnalyzer::getMetaData to reduce compilation time

["Pengcheng Xiong (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HIVE-15300?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15740976#comment-15740976 ] 

Pengcheng Xiong commented on HIVE-15300:
----------------------------------------

[~rajesh.balamohan], i just took a look. It seems that we need to catch the exception following the similar way used in other places in SemanticAnalyzer. Just like this
{code}
Table tab = null;
      try{
        tab = getTableObjectByName(tabName);
      } catch (HiveException e) {
        LOG.info("Table " + tabName + " is not found.");
      }
{code}
Could u resubmit the patch? Thanks.

> Reuse table information in SemanticAnalyzer::getMetaData to reduce compilation time
> -----------------------------------------------------------------------------------
>
>                 Key: HIVE-15300
>                 URL: https://issues.apache.org/jira/browse/HIVE-15300
>             Project: Hive
>          Issue Type: Improvement
>          Components: Parser
>            Reporter: Rajesh Balamohan
>            Assignee: Rajesh Balamohan
>            Priority: Minor
>         Attachments: HIVE-15300.1.patch, HIVE-15300.2.patch
>
>
> E.g Q88 in tpc-ds takes lots of time to compile and it ends up getting the table details for the same table repeatedly. It took 20+seconds to compile the query.
> It would be good to reuse the table information in SemanticAnalyzer::getMetadata.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)