Re: svn commit: r933348 - in /openwebbeans/trunk/webbeans-impl/src/main/java/org/apache/webbeans/intercept: ApplicationScopedBeanIntereptorHandler.java DependentScopedBeanInterceptorHandler.java Int

2010-04-13 Thread Gurkan Erdogdu
Hi; CreationalContext is a thread safe, you have to provide thread safety when running multiple threads and the most important object in CreationalContext is objectInstance and its thread safe. There is no problem I have seen. Also, your solution eats client provided CreatinalContext because

Re: svn commit: r933348 - in /openwebbeans/trunk/webbeans-impl/src/main/java/org/apache/webbeans/intercept: ApplicationScopedBeanIntereptorHandler.java DependentScopedBeanInterceptorHandler.java Int

2010-04-13 Thread Mark Struberg
Hoi! The problem is not that CreationalContext is not thread safe, but that NormalScopedBeanInterceptorHandler is not thread safe if you store the CreationalContext as member variable :) And yes, I'm eating the provided CreationalContext for NormalScoped bean proxies, because in my opinion it

Re: svn commit: r933348 - in /openwebbeans/trunk/webbeans-impl/src/main/java/org/apache/webbeans/intercept: ApplicationScopedBeanIntereptorHandler.java DependentScopedBeanInterceptorHandler.java Int

2010-04-13 Thread Mark Struberg
Yes Gurkan, I think I know _why_ you introduced it: because the TCK currently needs it. But I honestly think the TCK is wrong here. And this has also been confirmed by Marius and Pete already in the CDITCK-140. Of course, if we could find a way to resolve those cyclic references _without_

Re: svn commit: r933348 - in /openwebbeans/trunk/webbeans-impl/src/main/java/org/apache/webbeans/intercept: ApplicationScopedBeanIntereptorHandler.java DependentScopedBeanInterceptorHandler.java Int

2010-04-13 Thread Gurkan Erdogdu
Mark; Full logic is based on this stuff in OWB. Becuase TCK requires to use them. Therefore it is not a good idea to remove them completely. Therefore we have to find a solution to resolve this instead of changing logic. My point is that, there is a working code and maybe this code is not

Re: svn commit: r933348 - in /openwebbeans/trunk/webbeans-impl/src/main/java/org/apache/webbeans/intercept: ApplicationScopedBeanIntereptorHandler.java DependentScopedBeanInterceptorHandler.java Int

2010-04-13 Thread Gurkan Erdogdu
Your points are not aligned with specification. You have viewed a different perspective on using CContexts from how TCK and OWB are implemented . :) From: Mark Struberg strub...@yahoo.de To: dev@openwebbeans.apache.org Sent: Tue, April 13, 2010 11:38:18 AM

Re: svn commit: r933348 - in /openwebbeans/trunk/webbeans-impl/src/main/java/org/apache/webbeans/intercept: ApplicationScopedBeanIntereptorHandler.java DependentScopedBeanInterceptorHandler.java Int

2010-04-13 Thread Mark Struberg
Which point of my argumentation chain? Which sentence? Which section in the spec does get injured? LieGrue, strub --- Gurkan Erdogdu gurkanerdo...@yahoo.com schrieb am Di, 13.4.2010: Von: Gurkan Erdogdu gurkanerdo...@yahoo.com Betreff: Re: svn commit: r933348 - in

Re: svn commit: r933348 - in /openwebbeans/trunk/webbeans-impl/src/main/java/org/apache/webbeans/intercept: ApplicationScopedBeanIntereptorHandler.java DependentScopedBeanInterceptorHandler.java Int

2010-04-13 Thread Gurkan Erdogdu
Anyway I have reverted to beginning, and create an issue. Because tinkering in somewhere must not destroy some other places as a side affect. At evening, we could discuss the current design of the OWB regarding CC and setup a plan how to handle this weird situation without not effecting other

Re: svn commit: r933348 - in /openwebbeans/trunk/webbeans-impl/src/main/java/org/apache/webbeans/intercept: ApplicationScopedBeanIntereptorHandler.java DependentScopedBeanInterceptorHandler.java Int

2010-04-13 Thread Joseph Bergmark
I know I'm jumping into the conversation a bit late, but Im trying to catch up and get my head in the game a bit. I looked over the code, and read over Mark's example scenario earlier, which I'll paste a snippet of: And now consider a webserver with 10.000 different users clicking on the send

Re: svn commit: r933348 - in /openwebbeans/trunk/webbeans-impl/src/main/java/org/apache/webbeans/intercept: ApplicationScopedBeanIntereptorHandler.java DependentScopedBeanInterceptorHandler.java Int

2010-04-13 Thread Mark Struberg
That SessionScoped object isn't stored in MailService's creational context is it? Indirectly it is! a) Our CreationalContextImpl form a hierarchy tree which contains all CreationalContexts of all @Dependent scoped beans (CreationalContextImpl#dependentObjects). And currently we _also_ store

Re: svn commit: r933348 - in /openwebbeans/trunk/webbeans-impl/src/main/java/org/apache/webbeans/intercept: ApplicationScopedBeanIntereptorHandler.java DependentScopedBeanInterceptorHandler.java Int

2010-04-12 Thread Mark Struberg
This makes no difference, really. The problem is that we store a CreationalContext as a member in NormalScopedBeanInterceptorHandler. And subsequent function calls rely on it. But if in the meantime _another_ thread overwrites this NormalScopedBeanInterceptorHandler.creationalContext (by using

Re: svn commit: r933348 - in /openwebbeans/trunk/webbeans-impl/src/main/java/org/apache/webbeans/intercept: ApplicationScopedBeanIntereptorHandler.java DependentScopedBeanInterceptorHandler.java Int

2010-04-12 Thread Gurkan Erdogdu
Please attach a small web application code that shows your issue. Without it, it is very hard to visualize your siutation. As a side note, serialization issues always a problem in Java Web Applications. You have to manage this via synchoronization blocks or third party other libs, for example