On 02/02/2007, at 4:11 AM, David Jencks wrote:
On Feb 1, 2007, at 5:00 AM, Gianny Damour wrote:
Hi,
I am ready to check-in some changes to the way online and offline
deployers are booted and configured such that ModuleConfigurer can
be easily registered with the JMXDeploymentManager.
As per David J. suggestion, deployer.jar now boots a Kernel and
starts a list of configurations to register the ModuleConfigurers.
The list of configurations to start is named jsr88-configurer-
config.xml and it is no more no less a persistent configuration
list file (a file a la config.xml). In this file, there is a
module declaration for each type (jar, rar, war and ear) of
ModuleConfigurer implementations. When the deployer is started in
offline mode, an additional set of configurations is started to
register the relevant builders. This additional list of
configuration is named offline-deployer-config.xml and it is also
a persistent configuration list file.
A couple of problems I am aware of:
1. ModuleConfigurers and ModuleBuilders classes are still in a
same jar.
we can fix these one by one, right?
Yes. I contemplated doing the relevant changes to split
ModuleConfigurers and ModuleBuilders; however, I simply put it on
hold as we may not need it at the moment.
2. offline-deployer-list will be dropped and no more supported (it
is replaced by offline-deployer-config.xml).
and this is a problem how exactly :-) ? I'd say this is an
improvement?
I agree :). I just wanted to be sure that people will not be surprise
when they discover that offline-deployer-list has been dropped.
3. the JMXDeploymentManager instances used by the maven plugins do
not have access to the ModuleConfigurers.
I can't see any bad effect from this, can you? IIUC they don't now
either.
Indeed. There is no bad effect and we keep the status quo.
around (e.g. the Maven2Repository).
I intend to commit these changes over the week-end. So, let me
know if you have any concerns or if you want me to first address
1, 2 or 3.
Does this slow down normal deployment due to starting up the
additional kernel?
It is rather unnoticeable from an user perspective.
Thanks,
Gianny
thanks
david jencks
Thanks,
Gianny
