Re: [jboss-dev] Deployment chains
On Mon, 2008-10-06 at 22:23 +0200, Ales Justin wrote:
What about that discussion you had with Carlo,
about porting the component visitor?
It's not exactly a high priority.
Fixing the PersistenceDeployer to specifiy that it outputs
PersistenceUnitMetaData is. :-)
Although I must say I didn't fully understand that discussion
:-)
as to what is currently missing in AbstractComponentDeployer.
I don't understand it either.
Why setComponentVisitor() didn't work for him
is a mystery that remains unresolved, it clearly works
for the MC beans and JMX services?
So there's something different or it is too easy
to it wrong. :-)
The class in question is this one:
http://viewvc.jboss.org/cgi-bin/viewvc.cgi/jbossas/projects/jpa/trunk/dep...
It obviously has nothing to do with JPA.
It's a useful abstraction so it belongs as a
deployers helper class.
It is similar to BeanMetaDataFactoryVisitor
except it only deals with the boiler plate by
adding a getName() callback.
There's a FIXME
// TODO: determine proper component meta data class
component.addAttachment(componentMD.getClass().getName(), componentMD);
Which needs to be resolved by having the subclass pass the actual component
class in the constructor, for correct use of generics.
public AbstractComponentVisitor(Class<C> componentType)
{
this.componentType = componentType;
}
component.addAttachment(componentType, componentMD);
Although if you are going to allow subclasses of C, then
the collections need fixing to be for example
List<? extends C>
An abstract component visitor?
To relieve you of this code?
protected static void addXComponent(DeploymentUnit unit, X x)
{
DeploymentUnit component = unit.addComponent(getComponentName(x));
component.addAttachment(getAttachmentName(), x);
}
protected static void removeAliasComponent(DeploymentUnit unit, X x)
{
unit.removeComponent(getComponentName(x));
}
Adrian Brock wrote:
> You can release it as is.
> More management stuff can be added required.
>
> The stuff I've done so far, was just so I could
> easily see what the deployers were actually doing
> for this task:
> https://jira.jboss.org/jira/browse/JBAS-5044
>
> I dare say Scott may have other requirements
> when he gets chance to review it.
--
xxxxxxxxxxxxxxxxxxxxxxxxxxxx
Adrian Brock
Chief Scientist
JBoss, a division of Red Hat
xxxxxxxxxxxxxxxxxxxxxxxxxxxx