hadoop-hive-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Namit Jain (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HIVE-253) rand() gets precomputated in compilation phase
Date Mon, 16 Mar 2009 22:03:50 GMT

    [ https://issues.apache.org/jira/browse/HIVE-253?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12682483#action_12682483
] 

Namit Jain commented on HIVE-253:
---------------------------------

Looks good: I have some minor comments:

1. No need to include UDFRand in PartitionPruner
2. test rand_partitionpruner.q:
     can you add that on a partitioned table, srcpart, instead ?
     can you add a explain plan which will show that all partitions are included ? 
3. Can you add another tests for AND/OR of some expression and rand(), like
    Again can you perform a explain plan ?

> rand() gets precomputated in compilation phase
> ----------------------------------------------
>
>                 Key: HIVE-253
>                 URL: https://issues.apache.org/jira/browse/HIVE-253
>             Project: Hadoop Hive
>          Issue Type: Bug
>          Components: Query Processor
>    Affects Versions: 0.2.0
>            Reporter: Zheng Shao
>            Assignee: Raghotham Murthy
>            Priority: Blocker
>             Fix For: 0.2.0
>
>         Attachments: hive-253.1.patch
>
>
> SELECT * FROM t WHERE rand() < 0.01;
> Hive will say: "No need to submit job", because the condition evaluates to false.
> The rand() function is special in the sense that every time it evaluates to a different
value. We should disallow computing the value in the compiling phase.
> One way to do that is to add an annotation in the UDFRand and check that in the compiling
phase.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message