March 2008 - jboss-dev-forums - Jboss List Archives

[Design of EJB 3.0] - Re: webservices ejb3 api usage

by wolfc

Deployment handling should be done by an intermediate deployer. How does the webservice meta data merge with the ejb3 meta data? You should be able to pick up the merged ejb3 meta data and instantiate some meta data for the web deployer to initiate your endpoint. As for injection. That should have happened before method invocation. I was thinking about a ThreadLocalStack construction which is injected during construction and set during invocation. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4139480#4139480 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4139480

18 years

1
0
0 / 0

[Design of JBoss ESB] - Re: Features of JBoss Messaaging Message Bridge present in E

by beve

About the acknowledgement mode, this can be set on the jms-message-filter element: | <jms-message-filter dest-type="QUEUE" dest-name="queue/Test" acknowledge-mode="AUTO_ACKNOWLEDGE"/> | Can be one of: AUTO_ACKNOWLEDGE (default) CLIENT_ACKNOWLEDGE DUPS_OK_ACKNOWLEDGE. /Daniel View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4139470#4139470 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4139470

18 years

1
0
0 / 0

[Design of EJB 3.0] - Re: Determine whether an invocation is local or remote

by bdecoste

We pretty much do the same thing just in IsLocalInterceptor. If you wanted to remove the potential problem if a user removing the interceptor, the GUID/VMID would need to go into the proxy impl. There is already a containerGuid in the proxies which contains the VMID where the proxy was created as well as the ObjectName of the container. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4139469#4139469 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4139469

18 years

1
0
0 / 0

[Design of EJB 3.0] - Re: Determine whether an invocation is local or remote

by anil.saldhana＠jboss.com

The EJB2 proxy interceptors that deal with transport (InvokerInterceptor) use GUID (which is UUID actually and underneath uses the RMI vmid). I bank on this one to set the invocation to be intervm. Maybe we need this thing moved over to ejb3 world. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4139468#4139468 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4139468

18 years

1
0
0 / 0

[Design of EJB 3.0] - Re: Determine whether an invocation is local or remote

by ALRubinger

"anil.saldhana(a)jboss.com" wrote : Good point. In my opinion there should just be a flag on the invocation whether the call was intra-vm or intervm. Why not put a VMID in the Invocation at the container/core level, along with a method "boolean isLocalInvocation()" in EJBContainer, and then the security interceptor can call upon that? Then the interceptors would depend on the containers, not each other. S, ALR View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4139466#4139466 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4139466

18 years

1
0
0 / 0

[Design of JBoss Build System] - Re: aop pom for CR8

by pgier

Ok, thanks for reverting the pom. I was adding the aop jars that were not in maven. (pluggable-instrumentor, etc.). I must have overwrote that one by mistake. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4139465#4139465 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4139465

18 years

1
0
0 / 0

[Design of EJB 3.0] - Re: Determine whether an invocation is local or remote

by ALRubinger

"bdecoste" wrote : Generally. The containers already have a dependency on the IS_LOCAL settings that IsLocalInterceptor sets in order to know whether to marshall the response. | | If a user removed IsLocalInterceptor, we behave as if it's always a remote call and go through rmi and marshall everything. The question is, if the user removed it, what would security do? It would treat it as a remote call? So the containers don't really have a dependency then? Just take advantage of optimization if a flag's available in invocation metadata? We should be able to fall back gracefully. :) S, ALR View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4139464#4139464 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4139464

18 years

1
0
0 / 0

[Design of EJB 3.0] - Re: webservices ejb3 api usage

by heiko.braun＠jboss.com

anonymous wrote : | Define an interface for the SessionContainer.localInvoke. | Right, I already did that. Currently I am playing with two integration interfaces. One for deployment and another one for invocation. Both need to be implemented by the Ejb3Deployment and SessionContainer respectively. A generic way to access injectable properties (2.2) is covered by the later as well: Deployment handling: | | /** | * Access web service decl. within an EJB3 deployment. | */ | public interface WebServiceDeployment { | | java.util.List<spi.integration.WebServiceDeclaration> getServiceEndpoints(); | } | | public interface WebServiceDeclaration | { | /* aka bean name */ | String getComponentName(); | | /* aka bean impl */ | String getComponentClassName(); | | /* generic meta data retrieval */ | <T extends java.lang.annotation.Annotation> T getAnnotation(Class<T> t); | } | Invocation handling: | | /** | * Invoke a web service implementation hosted by a StatelessSession container. | */ | public interface ServiceEndpointContainer { | | java.lang.Class getServiceImplementationClass(); | | java.lang.Object invokeEndpoint(Method method, Object[] objects, | spi.InvocationContextCallback invocationContextCallback) | throws java.lang.Throwable; | } | | /** | * Callback for invocation context properties that may be injected | * in the service endpoint prior to actual method invocation. | */ | public interface InvocationContextCallback | { | <T> T get(Class<T> propertyType); | } | | View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4139462#4139462 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4139462

18 years

1
0
0 / 0

[Design of EJB 3.0] - Re: Determine whether an invocation is local or remote

by bdecoste

Setting a local vs remote flag in the invocation is exactly what IsLocalInterceptor does. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4139460#4139460 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4139460

18 years

1
0
0 / 0

[Design of POJO Server] - Re: migrating TransactionManager and Invokers to POJO

by bstansberry＠jboss.com

"jhalliday" wrote : Would one of the AS people care to try moving the invokers again? Not sure what you mean by this. The error you posted is because the WarDeployer has a now unnecessary <depends>jboss.jca:service=CachedConnectionManager</depends> That just needs to be moved to the deploy/jbossweb.sar's jboss.web:service=WebServer mbean; that's where the dependency really is. And the jca-hack-for-tomcat-service.xml move. I just did that locally and the server starts fine. I'd be happy to check that in, but don't know anything about any issues with the invokers. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4139458#4139458 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4139458

18 years

1
0
0 / 0

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

jboss-dev-forums March 2008