On Dec 31, 2010, at 1:31 PM, David Jencks wrote:

> Some of these ideas are implemented in the patch attached to 
> https://issues.apache.org/jira/browse/OWB-511
> 
> thanks
> david jencks
> 
> On Dec 31, 2010, at 12:02 PM, David Jencks wrote:
> 
>> I looked at the WebBeansContext recently and was a little surprised at what 
>> it does and doesn't do.  I wonder if what I want it to do is what David 
>> Blevins was thinking of as "next steps".

You pretty much got it.  The changes are a good next step.  We still have a 
ways to go as in OpenWebBeans proper (using the DefaultSingletonService) we 
still get most our hits through the static access to ServiceLoader.  

68639 org.apache.webbeans.corespi.ServiceLoader
34129 org.apache.webbeans.util.AnnotationUtil
6063 org.apache.webbeans.util.WebBeansUtil
2170 org.apache.webbeans.test.TestContext
1832 org.apache.webbeans.component.AbstractOwbBean
1464 org.apache.webbeans.intercept.WebBeansInterceptorConfig
1402 org.apache.webbeans.inject.impl.InjectionPointFactory
1170 org.apache.webbeans.decorator.WebBeansDecoratorConfig
 736 org.apache.webbeans.container.InjectionResolver
 594 org.apache.webbeans.xml.WebBeansXMLConfigurator

The number #2 "false" static usage was AnnotationUtil, I just cut that.

>> As a more minor point I'm a little confused by the choice of some of the 
>> hardcoded services now in WebBeansContext.  Some make sense like 
>> ConversationManager where there is no interface and no subclasses, but some 
>> are specific implementations of service interfaces such as JndiService and 
>> ScannerService where IMO the Default* implementations are unlikely to be 
>> used in most integration contexts.

The hardcoded services are just the ones that I found in the map after a 
complete build/test (having printed all entries to file).  We can certainly 
trim those out if we want.


-David


Reply via email to