[
https://issues.apache.org/jira/browse/ZOOKEEPER-847?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13101436#comment-13101436
]
Benjamin Reed commented on ZOOKEEPER-847:
-----------------------------------------
i'm also worried about changing the signature. i think this is an optimization
to save the server from detecting the problem. its not just about the change,
but none of the async methods throw exceptions. in reality, if we want to
detect this at the client, we need to invoke the callback, which is a bit
tricky due to threading issues. i think we should just allow the server to
detect it as it currently happens.
> Missing acl check in zookeeper create
> -------------------------------------
>
> Key: ZOOKEEPER-847
> URL: https://issues.apache.org/jira/browse/ZOOKEEPER-847
> Project: ZooKeeper
> Issue Type: Bug
> Components: java client
> Affects Versions: 3.3.1, 3.3.2, 3.3.3
> Reporter: Patrick Datko
> Assignee: Laxman
> Fix For: 3.4.0
>
> Attachments: ZOOKEEPER-847.patch
>
>
> I watched the source of the zookeeper class and I missed an acl check in the
> asynchronous version of the create operation. Is there any reason, that in
> the asynch version is no
> check whether the acl is valid, or did someone forget to implement it. It's
> interesting because we worked on a refactoring of the zookeeper client and
> don't want to implement a bug.
> The following code is missing:
> if (acl != null && acl.size() == 0) {
> throw new KeeperException.InvalidACLException();
> }
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
