You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@drill.apache.org by "Timothy Chen (JIRA)" <ji...@apache.org> on 2013/04/08 09:25:15 UTC

[jira] [Commented] (DRILL-50) Build a single node reference interpreter that leverages Google's Supersonic

    [ https://issues.apache.org/jira/browse/DRILL-50?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13625208#comment-13625208 ] 

Timothy Chen commented on DRILL-50:
-----------------------------------

Seems like the typical supersonic usage requires it to first specify schema info to the supersonic view object, add rows to it and then bound operations for results.
Our logical operator usage typically attempts to access expressions passed to each operator that assumes it's column existence.
I could work around this to create a supersonic obj each operator and buffer all rows or batch a number of rows to pass into it, to avoid a per row JNI call, but it seems like to use supersonic effeciently might need to explore more options
                
> Build a single node reference interpreter that leverages Google's Supersonic
> ----------------------------------------------------------------------------
>
>                 Key: DRILL-50
>                 URL: https://issues.apache.org/jira/browse/DRILL-50
>             Project: Apache Drill
>          Issue Type: New Feature
>            Reporter: Jacques Nadeau
>            Assignee: Timothy Chen
>
> Supersonic is a very interesting project that could help speed certain operations for the full execution engine.  As a first step towards exploring places where it could help, a single node partial interpreter should be built to explore Supersonic's strengths and weaknesses.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira