On 08/21/2012 06:04 AM, Jason Greene wrote:
On Aug 20, 2012, at 7:35 PM, Stuart Douglas
<stuart.w.douglas(a)gmail.com> wrote:
> On 18/08/2012, at 2:22 PM, Thomas Diesler <thomas.diesler(a)jboss.com> wrote:
>
>> Just to clarify, for non-osgi deployments on server restart we process them in
parallel. For a large set, deploymentA could race many phases ahead of deploymentB with no
particular order in which these deployments are processed, right?
> Yes, unless there are inter-module class loading dependencies, in which case a
deployment will wait for the deployment it depends on to finish MODULIZE.
>
Note that the intention is that we have fine grained dependencies between components,
which allows the greatest level of parallelism. This will always be our preferred model.
We do plan, as Stuart notes, to add coarse grained deployment level deps as a workaround
for legacy deployments.
Yes, that is very well and fine. What we don't have is
the ability to
define an "initial" dependency on a component that may go away without
affecting the service that initially depended on it. i.e. A component
can start functioning when the initial set of deployments on server
restart have reached a certain state it continues to function when one
of those deployments gets undeployed.
--
Jason T. Greene
JBoss AS Lead / EAP Platform Architect
JBoss, a division of Red Hat
--
xxxxxxxxxxxxxxxxxxxxxxxxxxxx
Thomas Diesler
JBoss OSGi Lead
JBoss, a division of Red Hat
xxxxxxxxxxxxxxxxxxxxxxxxxxxx