This content has been marked as final.
Show 6 replies
-
1. Re: JBPAPP-864/EJBTHREE-1109 - clustered home proxies
galder.zamarreno Jun 10, 2008 5:45 AM (in response to galder.zamarreno)Seems like there're issues with patches generated with "svn diff", see http://subversion.tigris.org/issues/show_bug.cgi?id=2164, so I've decided to commit the fix:
http://viewvc.jboss.org/cgi-bin/viewvc.cgi/jbossas?view=rev&revision=74347 -
2. Re: JBPAPP-864/EJBTHREE-1109 - clustered home proxies
brian.stansberry Jun 10, 2008 7:41 AM (in response to galder.zamarreno)OK. I was trying to apply the patch to my JBPAPP_4_2_0_GA_CP checkout so I could look at it that way and the 4.2/4.3 webservices differences were adding a lot of confusion. Just updating will save the hassle.
-
3. Re: JBPAPP-864/EJBTHREE-1109 - clustered home proxies
brian.stansberry Jun 10, 2008 8:28 AM (in response to galder.zamarreno)I'm guessing the reason you use the Dispatcher to find the StatefulClusteredProxyFactory in StatefulContainer.createClusteredProxy(Object id) is you lose the id info if you do it via JNDI?
-
4. Re: JBPAPP-864/EJBTHREE-1109 - clustered home proxies
galder.zamarreno Jun 10, 2008 8:50 AM (in response to galder.zamarreno)"bstansberry@jboss.com" wrote:
I'm guessing the reason you use the Dispatcher to find the StatefulClusteredProxyFactory in StatefulContainer.createClusteredProxy(Object id) is you lose the id info if you do it via JNDI?
Indeed. I think it might have been possible to retrieve the proxy factory by looking it up in JNDI and the calling any of the createProxies accordingly. I have the feeling that this solution would have indirectly gone through the Dispatcher as well.
What was clear though is that I needed to go through factory.createProxy() call to generate this so that the partition, family wrapper...etc where initialised correctly (I did test other alternatives and would fail without these being initialised). In the case of SLSBs, this is much simpler cos getting an instance of the SLSB proxy is just a matter of retrieving it from JNDI as proxies are the same. But in the case of SFSB, it's different cos you have to retrieve the proxy for a specific instance. -
5. Re: JBPAPP-864/EJBTHREE-1109 - clustered home proxies
brian.stansberry Jun 10, 2008 10:47 AM (in response to galder.zamarreno)"galder.zamarreno@jboss.com" wrote:
I think it might have been possible to retrieve the proxy factory by looking it up in JNDI and the calling any of the createProxies accordingly. I have the feeling that this solution would have indirectly gone through the Dispatcher as well.
Yes, it would go through the dispatcher as well. But that becomes an internal detail of how the proxy factory works, and doesn't leak into StatefulContainer.
I'm not advocating changing this for 4.x, as the usage of dispatcher is not going to change. :) But for trunk, think about it a bit. -
6. Re: JBPAPP-864/EJBTHREE-1109 - clustered home proxies
galder.zamarreno Jun 10, 2008 2:40 PM (in response to galder.zamarreno)"bstansberry@jboss.com" wrote:
"galder.zamarreno@jboss.com" wrote:
I think it might have been possible to retrieve the proxy factory by looking it up in JNDI and the calling any of the createProxies accordingly. I have the feeling that this solution would have indirectly gone through the Dispatcher as well.
Yes, it would go through the dispatcher as well. But that becomes an internal detail of how the proxy factory works, and doesn't leak into StatefulContainer.
I'm not advocating changing this for 4.x, as the usage of dispatcher is not going to change. :) But for trunk, think about it a bit.
Ok, I think you've got a point there. I'll leave it as it is for 4.x and rethink it for trunk. Thanks for the feedback :))