It is true that MethodProxy is super fast (I also gave it a try), however - I may be wrong here - the number of hits on the Manageable Objects will not be significant enought to see a true performance improvement.
It really depends on how we implement things. There may be some services that get hit often (especially if you have a monitoring application). Using a MethodProxy is just as easy a using reflections so I will most likely always use it.
Thanks for this answer.
I agree: it depends on how we implement things. For instance, a monitoring application could either poll the Manageable Objects or register an AttributeChangeNotification listener to keep its display updated.
My believe is that a monitoring application, which uses a polling mechanism, is not implemented correctly expecially if the Manageable Objects emits notifications. The fact that such applications may have an intrusive impact is "acceptable".
Acutally, my real concern was to have a simple implementation in a first stage and then roll-out an attribute cache, if required. Indeed, AFAIK this cache must be keep in sync. In order to keep this cache in sync, the manageable objects MUST emit attribute change notifications, and the cache needs to listen to them. Even if such an implementation could be trivial, it adds a complexity to the solution.
Whatever, Geronimo will be blasting fast.
Gianny
_________________________________________________________________
Hotmail : un compte GRATUIT qui vous suit partout et tout le temps ! http://g.msn.fr/FR1000/9493
