Hi Dan,
just saw your revision [1] on ManagedRoute today where you cleaned up the
compiler warnings regarding the deprecated usage of the Route.shutdown()
API. IMHO we should better keep on that compiler warnings until the *right*
moment (3.x.y release) because:
At the time the commiter who intends to *remove* the deprecated API of
Route.shutdown() and all it's
usages will most probably *not* realize that we would expose the shutdown()
semantics through the JMX as well, as you changed the logic there to be a
composite call of the Route stop() & remove() API making the compiler happy
with 0 warnings. However if we would keep on that warnings (until the right
moment) then the compiler will help the commiter to see that ManagedRoute
doesn't compile anymore!
I would say we should better keep on those warnings for the moment and maybe
even better, mark the ManagedRoute.shutdown() *itself* also as @Deprecated
so that for example it's usage through JMX-Proxy-API
(MBeanServerInvocationHandler.newProxyInstance() by unit-tests becomes also
visible, for example for the commiters who are implementing *new* unit-tests
*before* the 3.x.y major release
and would see and avoid the deprecated ManagedRoute.shutdown() JMX-API
usage. However the dynamic invocations through MBeanServer.invoke() on that
ManagedRoute.shutdown() API will be of course not become visible at
compile-time but first at runtime as the commiter runs the unit-tests before
commiting the Route.shutdown() removal.
IMHO sometimes it's even better to keep on the compiler warnings (they could
indeed help us!) for the sake of easier & safer refactoring at the *right*
moment.
My english is pretty poor, nevertheless I hope you get the point I'm trying
to address.
[1] http://svn.apache.org/viewvc?view=revision&revision=1222462
Babak
--
View this message in context:
http://camel.465427.n5.nabble.com/The-last-revision-on-the-ManagedRoute-shutdown-JMX-API-tp5098934p5098934.html
Sent from the Camel Development mailing list archive at Nabble.com.
