> This makes alot of sence... almost so that I wonder
> why the JMX folks did
> not make it this way from the start... perhaps the
> JINI guys wanted all the
> glory.

Jason, I really like the "ra-ra-ra-ism" of your attitude, that being said, the JMX 
folks didn't do it this way for a good reason : detaching. 

the base of the microkernel is broken when you do that as you have direct references.  
The whole idea of JMX is to detach the invocation.  

Then JINI doesn't do that, it just provides with hard links (pointers, not 
objectNames) but being able to look up pointers by class is simple if you are doing 
typed programming and also means you don't need to pass "names" of instances to look 
up, just the class is sufficient.


> the ugly MBeanServer.invoke() calls can be done away
> with too?

NO! (yes)

Again those calls are important in the pure detyped, detached invocations, as it is 
the basis for an advanced microkernel.  without these we are stuck (as everyone else) 
in the linked invocations and if our reference dies we have a problem. 
MBeanServer.invoke() is powerful.

The bottom line is "no" we are not getting rid of MBS.Invoke() in the large "modular" 
parts as without it you really don't have detached programming. 

<however/> In some *sub*parts where I do want the MBeans management and singletons and 
lookups but I don't want the server.invoke() semantics all the time (and that catch 
diarreah that comes with it) then the 2 little hacks implemented in the Registry are 
really neat and clean and cool.
marcf


______________________________________________________________________
View this jboss-dev thread in the online forums:
http://jboss.org/forums/thread.jsp?forum=66&thread=6254

_______________________________________________
Jboss-development mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/jboss-development

Reply via email to