[
https://issues.apache.org/jira/browse/QPID-6114?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14147591#comment-14147591
]
Alex Rudyy commented on QPID-6114:
----------------------------------
Keith,
Thanks for good comments.
I believe that we need to address all your comments
{quote}
I think this change makes the overridden #getAttribute methods in ACO
subclasses redundant. Perhaps ACO#getAttribute should be made final too?
{quote}
Yes, it seems we missed that. We should remove overridden getAttribute from our
CO tree and modify ACO to call getState() for STATE attribute. That will also
fix a bug on AbstractQueue where ACTIVE is always returned as a state which is
incorrect. We missed that.
{quote}
Why is _state no longer protected by a memory barrier?
{quote}
The state is changed from Configuration thread, so, it is to some extent not
critical but I agree that we should strengthen that change _state var to be
volatile or atomic.
{quote}
Not directly related to this commit, but I wonder if
BDBHAReplicaVirtualHost should implement validateChange to reject all attempted
mutations?
{quote}
Yes it can be done. Looks like a good suggestion. Also, we might need to do
something with ManagedAttributeFields. It does not look that we can manage the
fields from the replica host at all.
> Migrate the State member variable to AbstractConfiguredObject class
> -------------------------------------------------------------------
>
> Key: QPID-6114
> URL: https://issues.apache.org/jira/browse/QPID-6114
> Project: Qpid
> Issue Type: Improvement
> Components: Java Broker
> Reporter: Andrew MacBean
> Assignee: Alex Rudyy
> Fix For: 0.31
>
>
> As all ACOs should have State, this is the most sensible place for this and
> its getter and setter to live just now.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
