From: oberon777 <nu...@jb...> - 2005-07-25 21:25:52
|
Can anyone explain why this exception is happening? Thanks. View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=3886490#3886490 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=3886490 |
From: rmcdonough <nu...@jb...> - 2005-07-27 02:59:07
|
What's your jboss-aop.xml look like? It's hard to tell what the issue is by just looking at bunch of stack trace. Other than trying to run an AO app, what and how are your trying to apply your aspects? Ryan- View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=3886714#3886714 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=3886714 |
From: oberon777 <nu...@jb...> - 2005-07-27 05:30:07
|
A wide pointcut like this yields this problem: | <bind pointcut="call(* org.jgap.*->*(..)) AND !within(CallerInterceptor) AND !within(StackTraceCatcher)"> | <interceptor class="CallerInterceptor"/> | </bind> | Similarly, if I were to have call(* *->*(..)) as the matching expression, the same issue would arise. View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=3886728#3886728 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=3886728 |
From: <kab...@jb...> - 2005-07-27 08:20:35
|
We do have some isues with wide pointcuts like these, and opinion is divided whether this is by design or not. Assuming your CallerInterceptor and StackTraceCatcher classes are not within the default package, try fully qualifying their names. Otherwise try narrowing down the pointcut View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=3886742#3886742 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=3886742 |
From: oberon777 <nu...@jb...> - 2005-07-27 08:35:01
|
Kabir, thanks for your reply. Yeah, those belong to the default package. Those pointcuts are intentially wide, narrowing them down is not really an option. Should I assume that matching on every method or even every constructor call is not likely to work? What exactly are the issues you're referring to?.. On a related note, I was talking to JBossProfiler folks and was given the impression that they succeed with using AOP for getting profile information. Getting profile info most definitely requires having a large number of pointcuts... I wouldn't be surpised if these same issues arose for even a small application. I should probably double-check with a smaller test case. View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=3886745#3886745 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=3886745 |
From: <kab...@jb...> - 2005-07-27 10:29:56
|
The issues are that infinite recursion can take place, since interceptors/aspects themselves can be weaved. I tried something similar to what you describe above based on one of the examples in our tutorials, but could not reproduce your problem. | <bind pointcut="call(* *->*(..))"> | <interceptor class="SimpleInterceptor"/> | </bind> | causes recursion as expected | <bind pointcut="call(* *->*(..)) AND !within(SimpleInterceptor)"> | <interceptor class="SimpleInterceptor"/> | </bind> | was fine. Do you have other bindings declared in your jboss-aop.xml file? If you do, and haven't already, make sure that all your interceptors/aspects are included in !within() clauses View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=3886765#3886765 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=3886765 |