[jira] [Commented] (DRILL-304) Implement destroy() method for HyperBatches

["Jacques Nadeau (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/DRILL-304?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13829574#comment-13829574 ] 

Jacques Nadeau commented on DRILL-304:
--------------------------------------

As part of this, we should also modify the vector.transferpair.transferto() methods so that they clear the target before pointing to the new memory.

> Implement destroy() method for HyperBatches
> -------------------------------------------
>
>                 Key: DRILL-304
>                 URL: https://issues.apache.org/jira/browse/DRILL-304
>             Project: Apache Drill
>          Issue Type: Task
>            Reporter: Ben Becker
>
> Ideally, classes like AbstractSingleRecordBatch() should call container.clear() to decrement the reference count of the incoming ValueVectors before calling next() on the upstream operator.  There is an issue with HyperBatches in this case though; the underlying ValueVectors should not be freed implicitly.  That is, the operator should only clear the vectors when the entire HyperBatch has been consumed.
> It has been suggested that a destroy() method be used to handle the HyperBatch case.



--
This message was sent by Atlassian JIRA
(v6.1#6144)