[HiveMind] Is interceptor creation on threaded model unnecceseraly slow?

[Christian Essl <ch...@yahoo.de>]

For the threaded model as I see interceptors are created for each core- 
impl. Wouldn't it be enough that interceptors wrap the proxy. This would 
certainly be more performant. Or maybe have interceptors under order lets 
say 1000 wrap the core-impl (and be created each time) and others wrap the 
proxy?

It think it should also be commented that InterceptorFactories should cache 
their interceptor-classes by service-id (as AbstractInterceptorFactory 
does). Otherwise you may have a memory leak. As I see the referenced 
example on how an InterceptorFactory uses config elements 
(hivemind.test.services.impl.FilterLoggingInterceptor) does not do that.



---------------------------------------------------------------------
To unsubscribe, e-mail: commons-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: commons-dev-help@jakarta.apache.org