[jira] [Commented] (DRILL-3345) TestWindowFrame fails to properly check cases involving multiple batches

["Aman Sinha (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/DRILL-3345?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14599980#comment-14599980 ] 

Aman Sinha commented on DRILL-3345:
-----------------------------------

+1 with one minor comment: 
Perhaps you should modify the following message to not say 'frame' 
{code}
    logger.trace("frame changed, we are ready to process first saved batch");
{code}
Now  it refers to partition instead of frame.

> TestWindowFrame fails to properly check cases involving multiple batches
> ------------------------------------------------------------------------
>
>                 Key: DRILL-3345
>                 URL: https://issues.apache.org/jira/browse/DRILL-3345
>             Project: Apache Drill
>          Issue Type: Bug
>          Components: Execution - Relational Operators
>    Affects Versions: 1.1.0
>            Reporter: Deneche A. Hakim
>            Assignee: Deneche A. Hakim
>            Priority: Critical
>              Labels: window_function
>             Fix For: 1.1.0
>
>         Attachments: DRILL-3345.1.patch.txt
>
>
> as part of adding unit tests to TestWindowFrame, I added a debug option to MSorter to force the batches passed downstream to be of a specific size. This was supposed to help test edge cases involving multiple data batches while controlling precisely when a partition and/or window frame ends in a batch.
> Turns out the change to MSorter was incomplete and all cases end up processing one single big batch of data.
> The purpose of this JIRA issue is to fix this, add some check to detect when the batches are no longer split correctly and make sure all unit tests are passing.



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