You are right. It is slower than the current strategy. Indeed, the number of method invocations required to traverse the pipeline is multiplied by two. Actually, I thought that the time required to call a method, which forwards the invocation to the next Interceptor, is not significant compared to the execution time of an Interceptor. This assumption may be wrong.Sounds interesting but kinda slow. Do I have to go through all those iterators to get my call done?
However if it is right, then the order of magnitude of performance is the same for these two strategies and the proposed one is more flexible.
IIRC, this strategy is used by Catalina's containers. They call it a Valve.
Thanks for that. Gianny
_________________________________________________________________
MSN Messenger http://g.msn.fr/FR1001/866 : dialoguez en direct et gratuitement avec vos amis !
