On Mon, 2004-02-23 at 02:51, Niclas Hedhman wrote: > I haven't spend _that_ much thinking about it, maybe Stephen has a > better > view.... > > My perception is that although LifeCycle extensions can be used for > this, we > have postponed the JMX implementation to a great degree because the > "Deployment Model" (how components hangs together) was static until a > "few > hours ago". > Now it is not anymore! > Therefor, JMX subsystem should be created as a "facility", practically > a > service that has access to the DeploymentModel, and is typically more > trusted > than your average component. > > It then registers itself to the DeploymentModel and listens for > changes, and > creates/destroys MBeans as the model changes. > Likewise the JMX subsystem can create new components by modifying the > DM > (which results in events and destruction/creation of MBeans...) Sounds pretty cool. I'll look into the DeploymentModel and look at the HTTP facility implementation as an example. > > If you want to "do it right", we would be grateful, and I suggest that > Stephen > in that case give you a set of pointers on where to look to understand > how to > migrate your current effort. > I'd definitely like to "do it right" so any pointers would be appreciated.
> Niclas > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [EMAIL PROTECTED] > For additional commands, e-mail: [EMAIL PROTECTED] > --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
