Targetting for beta and assigning to Ted for now. This will likely be more of a group exercise and be part of a more general review process of the current product.

Task for 2.0 FCS is limited to testing / timing a specific System Event Listener to determine if there is significant overhead related to using multiple listeners vs. fewer.

Mojarra source code shows reflection for each invocation of a SystemEventListener, so there is definitely performance overhead that could be avoided. In any case, this is an optimization, so can be addressed after 2.0.

If profiling shows a clear need for performance improvements, this can be investigated sooner.

The use of a specific Mojarra class in the ICEPushResourceHandler causes problems under MyFaces:

if (projectStage != null && !projectStage.equals("Production"))

...
Caused by: java.lang.reflect.InvocationTargetException
at sun.reflect.NativeConstructorAccessorImpl.newInstance0(Native Method)
at sun.reflect.NativeConstructorAccessorImpl.newInstance(NativeConstructorAccessorImpl.java:39)
at sun.reflect.DelegatingConstructorAccessorImpl.newInstance(DelegatingConstructorAccessorImpl.java:27)
at java.lang.reflect.Constructor.newInstance(Constructor.java:513)
at org.apache.myfaces.shared.util.ClassUtils.buildApplicationObject(ClassUtils.java:539)
... 15 more
Caused by: java.lang.NoClassDefFoundError: com/sun/faces/facelets/tag/ui/UIDebug
at org.icefaces.impl.push.servlet.ICEpushResourceHandler.<init>(ICEpushResourceHandler.java:80)
... 20 more
Caused by: java.lang.ClassNotFoundException: com.sun.faces.facelets.tag.ui.UIDebug
at org.apache.catalina.loader.WebappClassLoader.loadClass(WebappClassLoader.java:1676)
at org.apache.catalina.loader.WebappClassLoader.loadClass(WebappClassLoader.java:1521)
... 21 more

Perhaps simply wrapping the relevant code in EnvUtil.isMojarra() check would suffice.

Reverted the HeadBodyListener and BridgeSetup optimizations as it broke our portlet support.

Partly reverted one of the optimizations. It now checks for the String com.sun.faces.facelets.tag.ui.UIDebug again instead of the class.

Originally the optimizations seemed to have improve performance of the lifecycle by a bit over 10%. After reverting some of the optimizations there is still a performance gain of about 5% left.

No additional work has been done for this one lately. Some of the original improvements are still in causing the 5% performance increase. Marking this one as FIXED now.