[
https://issues.apache.org/jira/browse/HAMA-568?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Edward J. Yoon updated HAMA-568:
--------------------------------
Attachment: HAMA-568.patch
Attach my patch. Memory-based queues are now use concurrent queue. So,
synchronize() method will return the queue itself if it is mem-based queue.
{code}
/*
* static constructor methods to be type safe
*/
public static <T> SynchronizedQueue<T> synchronize(MessageQueue<T> queue) {
if(queue.isMemoryBasedQueue()) {
return (SynchronizedQueue<T>) queue;
}
return new SingleLockQueue<T>(queue);
}
{code}
.... if no-one objects within three days, I'll assume lazy consensus and commit
it.
> Add faster synchronized collections for message queues
> ------------------------------------------------------
>
> Key: HAMA-568
> URL: https://issues.apache.org/jira/browse/HAMA-568
> Project: Hama
> Issue Type: Sub-task
> Reporter: Thomas Jungblut
> Assignee: Edward J. Yoon
> Fix For: 0.7.0
>
> Attachments: HAMA-568.patch, HAMA-568_1.patch, HAMA-568_2.patch
>
>
> Currently the synchronized messaging (accessed concurrently via RPC) is done
> with a single mutex object. See org.apache.hama.bsp.message.SynchronizedQueue.
> This can be improved for memory queue by adding the ConcurrentLinkedQueue
> from Java since it has higher throughput.
> Needless to say, we should add some custom synchronized version of a queue,
> maybe through a factory class in the normal version.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
