The creation of stress tests are great news for JBoss AOP.
These results are already encouraging, even because they are just initial tests, being run
against code that we intend to improve.
Regarding the per_instance issue, I agree with the idea of differentiating domains that
haven't overriden the bindings from those that have.
We could store the generated joinpoint classes in a WeakHashMap, where we would map the
list of advices/interceptors to the generated class. That way, if:
| the joinpoint class is replaced by another one, due to dynamic operations -> the
class would eventually be removed from the map
| two or more per instance domains (with overriden bindings) happen to have the same
collection of advices/interceptors to apply to the same joinpoint -> the needed
joinpoint class would be generated only once
| two or more instances have a non-overriden domain -> the needed joinpoint class
would be generated only once
|
|
| Plus, after making generated advisors work 100% with hotswap, I suggest that, only in
hotswap mode, we do the advices call inside the wrapper instead of a joinpoint class. This
way, we would avoid the extra call to the joinpoint class. With hotswap, we will be able
of swapping the wrapper code when needed.
View the original post :
http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4031050#...
Reply to the post :
http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&a...