December 2007 - jboss-dev-forums - Jboss List Archives

[Design the new POJO MicroContainer] - Re: setting managed property of type List

by adrian＠jboss.org

"alesj" wrote : OK, this is deeper problem, than I anticipated earlier. | | There is a problem of interfaces/abstract classes. You should just define default implementations for the collection interfaces like JAXB or JMX does. http://java.sun.com/javase/6/docs/api/javax/management/MXBean.html anonymous wrote : | A List is reconstructed as a java.util.ArrayList; a Set as a java.util.HashSet; a SortedSet as a java.util.TreeSet. | Generically, we could allow some kind of plugin mechanism to allow somebody to configure: interface -> implementation class or factory View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4112235#4112235 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4112235

18 years, 4 months

1
0
0 / 0

[Design of POJO Server] - Re: Seam examples tests

by wolfc

And thus we'll have a circular dependency when Seam goes into EAP. Because currently EAP == AS instead of icing on top. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4112234#4112234 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4112234

18 years, 4 months

1
0
0 / 0

[Design the new POJO MicroContainer] - Re: Nested property ref

by adrian＠jboss.org

"alesj" wrote : | "adrian(a)jboss.org" wrote : | | I need to look at the thread context classloader usage anyway WRT the deployers | | there's a thread about it somewhere trying to define what the rules should be. | Are those weird CL issues of Seam apps releated with this? | - http://www.jboss.org/index.html?module=bb&op=viewtopic&t=104907 I doubt it. The two CNFEs on that thread are inside an MBean and an EJB3 container which should be setting the correct TCL. The CL issues aren't weird, it's just a case of making sure that most code runs with a TCL that would be intuitive. e.g. run all REAL deployers with the deployment's classloader as the TCL. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4112227#4112227 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4112227

18 years, 4 months

1
0
0 / 0

[Design the new POJO MicroContainer] - Re: Nested property ref

by alesj

"adrian(a)jboss.org" wrote : On a related issue with the classloader, we should be using the TCL | of the person registered the controller context if they don't define a ClassLoaderMetaData. | Currently we just the use current one (which is effectively random). | | I need to look at the thread context classloader usage anyway WRT the deployers | there's a thread about it somewhere trying to define what the rules should be. OK, let me know if you need me to do anything on the kernel side. btw: I think we should get this AOP interceptor usage asap But wouldn't that mean that we would always require to run AOP configuration ? "adrian(a)jboss.org" wrote : | I need to look at the thread context classloader usage anyway WRT the deployers | there's a thread about it somewhere trying to define what the rules should be. Are those weird CL issues of Seam apps releated with this? - http://www.jboss.org/index.html?module=bb&op=viewtopic&t=104907 View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4112224#4112224 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4112224

18 years, 4 months

1
0
0 / 0

[Design the new POJO MicroContainer] - Re: Nested property ref

by adrian＠jboss.org

On a related issue with the classloader, we should be using the TCL of the person registered the controller context if they don't define a ClassLoaderMetaData. Currently we just the use current one (which is effectively random). I need to look at the thread context classloader usage anyway WRT the deployers there's a thread about it somewhere trying to define what the rules should be. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4112223#4112223 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4112223

18 years, 4 months

1
0
0 / 0

[Design the new POJO MicroContainer] - Re: Nested property ref

by adrian＠jboss.org

" Now that I've changed ConfigureAction to use direct BeanInfo.set, I guess there is no need for this? This == preparing MetaDataStack & ContextCL before BeanInfo.set invocation. " What? I don't understand the logic? You modified the code and are now claiming that you don't need to implement a feature of the old code? Or are you saying it is already done elsewhere? The classloader/metadata stack stuff needs doing more generically anyway. i.e. It should be doing it for JMX as well (although in the case of, JBossMX it already does the classloader - the RI JMX doesn't though). And it should also be doing it for all callouts to the bean (I don't know that it is done currently?). My preference is that this code should be in an AOP interceptor not in the MC anyway. You want the classloader/metadata stack for all invocations (if you want it all) not just the MC lifecycle. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4112219#4112219 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4112219

18 years, 4 months

1
0
0 / 0

[Design of JCA on JBoss] - Re: ManagedConnection AND ConnectionEventListener relationsh

by vickyk

"adrian(a)jboss.org" wrote : | Where? That is not correct. The fact that we only register one ConnectionEventListener | (the connection manager) is an implementation detail of ours, it is not in the spec, | you can have many listeners if you want. Ok , that was due to my wrong interpretation of specs :( . View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4112215#4112215 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4112215

18 years, 4 months

1
0
0 / 0

[Design the new POJO MicroContainer] - Re: Nested property ref

by alesj

"adrian(a)jboss.org" wrote : | I've got no idea what you talking about? :-) About this code in KernelControllerContextAction: | static Object dispatchJoinPoint(final KernelControllerContext context, final Joinpoint joinpoint) throws Throwable | { | BeanMetaData metaData = context.getBeanMetaData(); | ClassLoader cl = Configurator.getClassLoader(metaData); | AccessControlContext access = null; | if (context instanceof AbstractKernelControllerContext) | { | AbstractKernelControllerContext theContext = (AbstractKernelControllerContext) context; | access = theContext.getAccessControlContext(); | } | | KernelController controller = (KernelController) context.getController(); | KernelMetaDataRepository repository = controller.getKernel().getMetaDataRepository(); | MetaData md = repository.getMetaData(context); | if (md != null) | MetaDataStack.push(md); | else | staticLog.warn("NO METADATA! for " + context.getName() + " with scope " + context.getScopeInfo().getScope()); | try | { | | // Dispatch with the bean class loader if it exists | ClassLoader tcl = Thread.currentThread().getContextClassLoader(); | try | { | if (cl != null && access == null) | Thread.currentThread().setContextClassLoader(cl); | | if (access == null) | { | return joinpoint.dispatch(); | } | else | { | DispatchJoinPoint action = new DispatchJoinPoint(joinpoint); | try | { | return AccessController.doPrivileged(action, access); | } | catch (PrivilegedActionException e) | { | throw e.getCause(); | } | } | } | finally | { | if (cl != null && access == null) | Thread.currentThread().setContextClassLoader(tcl); | } | } | finally | { | if (md != null) | MetaDataStack.pop(); | } | } | Now that I've changed ConfigureAction to use direct BeanInfo.set, I guess there is no need for this? This == preparing MetaDataStack & ContextCL before BeanInfo.set invocation. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4112214#4112214 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4112214

18 years, 4 months

1
0
0 / 0

[Design the new POJO MicroContainer] - Re: Nested property ref

by adrian＠jboss.org

"alesj" wrote : | Do I still need to wrap it up with MetaDataStack push and ContextClassloader change? | I've got no idea what you talking about? View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4112211#4112211 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4112211

18 years, 4 months

1
0
0 / 0

[Design of JCA on JBoss] - Re: ManagedConnection AND ConnectionEventListener relationsh

by adrian＠jboss.org

"vickyk" wrote : After going though the JCA specs I could make that there is one-one relationship between ConnectionEventListener AND ManagedConnection. | In the BaseWrapperManagedConnection | Where? That is not correct. The fact that we only register one ConnectionEventListener (the connection manager) is an implementation detail of ours, it is not in the spec, you can have many listeners if you want. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4112210#4112210 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4112210

18 years, 4 months

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 December 2007