[jgroups-dev] New JChannel.getState() signature

I started work on [1] (changes to JChannel.getState()) and wanted to 
solicit some feedback.

The main reason we wanted to change getState() is to be able to learn 
what exactly went wrong when a state transfer fails, and to possibly 
retry programmatically.

The changes I want to implement are:

- No timeout. A timeout can always be either too low, or to high. No 
timeout means getState() blocks until the state has been fully received, 
period.
- Return value:
   - True: the state was transferred successfully
   - False: the member is the only member in the cluster, or the target 
member == self. NOTE: false does *not* mean the state could not be 
marshalled, or any other failure !
- As mentioned above, a failure in state transfer is thrown as a 
StateTransferException (or subclass), STE is a subclass of ChannelException
   - Failures could be (1) marshalling failures at the state provider or 
requester
   - Communication failures (e.g. state provider crashed or left 
half-way through the state transfer)

I'm not sure whether the boolean return value of getState() makes sense, 
should I make getState() return void ? Then if the state provider == 
state requester, or if the state requester is the only cludter member, 
we'd return immediately.

Note that getState() returns only after the state has been transferred 
completely, ie. the setState(byte[]) method returned, or the 
setState(InputStream) method closed the input stream.

Opinions, feedback ?

[1] https://issues.jboss.org/browse/JGRP-619

-- 
Bela Ban
Lead JGroups / Clustering Team
JBoss

------------------------------------------------------------------------------
EditLive Enterprise is the world's most technically advanced content
authoring tool. Experience the power of Track Changes, Inline Image
Editing and ensure content is compliant with Accessibility Checking.
http://p.sf.net/sfu/ephox-dev2dev
_______________________________________________
Javagroups-development mailing list