Weld 1.0.1.CR1

[pmuir]

I've tagged and published Weld 1.0.1.CR1. It's also integrated in JBoss AS trunk (available in nightly builds newer than #1135 and the upcoming JBoss AS 6.0.0.M2 :-) and should soon be integrated into GlassFish. However, we aren't doing a full distribution this time - the examples, APIs, Javadocs etc. are all identical to 1.0.0. There will be a full distro for CR2 :-)

You can also download a Weld SE snapshot or a Weld Servlet snapshot containing this version of Weld.

[meetoblivion]

I assume the glassfish integration is dependent on the completion of changes between the glassfish-weld bridge and how weld has changed (e.g. the changes in BeanManagerImpl)

[pmuir]

John Ament wrote on Feb 01, 2010 01:07:


I assume the glassfish integration is dependent on the completion of changes between the glassfish-weld bridge and how weld has changed (e.g. the changes in BeanManagerImpl)

I believe Roger has fixed this bug in the integration already, but I'm not sure if this is in GlassFish nightly builds yet (once it is, you can drop the new OSGi bundle into GF and have it just work). Largely, the reason for the slower integration into GF is firstly due to it being done by a different team (Roger K @ Sun/Oracle), whilst I do the JBoss AS integration, but also in this case a bit delayed by the whole Sun/Oracle thing I guess ;-).

BTW, John, sorry we didn't address your EJB/Weld clustering issues this time around, but as you saw we need some changes to JBoss EJB3, so didn't manage to get them in. FWIW the clustering of managed beans should now work well :-)

[meetoblivion]

Aye, right now I'm trying to convince people that not everything is mission critical so that if a node fails, it's not necessary to demand that the session state is propagated.  Either way, an interesting issue w/ JBAS is that it's not an instant replication by design.  Hopefully they change that w/ infinispan.

[asookazian]

John Ament wrote on Feb 01, 2010 18:57:


an interesting issue w/ JBAS is that it's not an instant replication by design.  Hopefully they change that w/ infinispan.

I thought that JBoss replicates state to the other nodes in the cluster at the end of method calls (and that you can re-configure this behavior as well) when replication is turned on.  I distinctly remember the CSP EAP techs telling me that regarding EAP 4.3.

So what do you mean it's not instant replication by design?

[meetoblivion]

This, from what they tell me

http://community.jboss.org/thread/68783
http://community.jboss.org/wiki/WebClusteringBestPractices