[
https://issues.apache.org/jira/browse/BOOKKEEPER-64?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13098066#comment-13098066
]
Ivan Kelly commented on BOOKKEEPER-64:
--------------------------------------
Actually,this patch is wrong. I misremembered what
computeDigestAndPackageForSending does. I need to think about this more.
> LedgerHandle#asyncAddEntry expects byte[] to remain available until completion
> ------------------------------------------------------------------------------
>
> Key: BOOKKEEPER-64
> URL: https://issues.apache.org/jira/browse/BOOKKEEPER-64
> Project: Bookkeeper
> Issue Type: Bug
> Reporter: Ivan Kelly
> Attachments: BOOKKEEPER-64.diff, BOOKKEEPER-64.diff
>
>
> When a call is made to asyncAddEntry, the byte[] passed must not change until
> the async confirmation comes back. Otherwise the result is undefined. While
> for some API this is acceptable, I BK is aimed at applications where
> buffering may be used. This may force an extra copy before sending the data
> to BK to ensure the data sticks around. This is pointless as one of the first
> things addEntry does is make a copy while staging the data to send.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
