[
https://issues.apache.org/jira/browse/HBASE-15793?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15275319#comment-15275319
]
stack commented on HBASE-15793:
-------------------------------
Putting an HBase Interface in front of the netty future is the way to go.
Netty Future is rich in utility. Looking at it made go over our current Future
use. There is a bunch of use of java.util.concurrent.Future. There is
SyncFuture and AsyncRequestFuture... there is a CallFuture and
java.util.concurrent.ScheduledFuture. Our new async WAL actually does
io.netty.util.concurrent.Future.
That said, I think its ok having this new Future at top level in the client.
There does not seem to be overlap w/ our current Futures'.
I also think it ok NOT giving this 'Future' a hbasey name -- i.e. HFuture or
some such...
The genericization (word?) is pretty fancy.
Let me commit to help move this project along... [~ghelmling] you might be
interested in whats going on in here sir.
> Port over AsyncCall improvements
> ---------------------------------
>
> Key: HBASE-15793
> URL: https://issues.apache.org/jira/browse/HBASE-15793
> Project: HBase
> Issue Type: Improvement
> Components: rpc
> Reporter: Jurriaan Mous
> Assignee: Jurriaan Mous
> Attachments: HBASE-15793-v1.patch, HBASE-15793.patch
>
>
> Again a part of HBASE-13784 patch.
> This issue is about porting over the improvements in AsyncCall so it is a
> more flexible for usage in true async calls. It will not depend anymore on
> PayloadCarryingRpcController which is not needed in the async implementation.
> Furthermore it introduces the MessageConverter and IOExceptionConverter so
> the responses of future async calls can be converted in an easy async way.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
