On Fri, 2008-10-03 at 16:42 +0200, Adrian Brock wrote:
> 2) AOP ClassLoading
>
> A similar criticism applies for the AOPClassLoaderDeployer
> This has no chain definition at all? Only a relative order based
> on the main classloader deployer.
>
> Again this is brittle to a refactoring of how the classloader
> deployer
> works.
https://jira.jboss.org/jira/browse/JBDEPLOY-96
Kabir, can you modify your deployer to specify
java.lang.ClassLoader as an input instead of using the relative
ordering.
Also this deployer says it needs a VFS deployment.
Other than the names used within AOP;
ScopedVFSClassLoaderDomain
VFSClassLoaderDomainRegistry
I don't see any dependency on the VFS api?
It might have done in the past until we refactored the
way in the memory url part of the classpath is done?
Finally, this still hasn't been updated to understand
how subdeployment classloaders work. It should be asking
whether the (sub-)deployment has an
org.jboss.classloading.spi.dependency.Module
attachment and hence its own classloader.
Even the old code which has some hardwiring for JBossWebMetaData
looks broken since it only unregisters top level deployment
classloaders in undeploy() ?
--
xxxxxxxxxxxxxxxxxxxxxxxxxxxx
Adrian Brock
Chief Scientist
JBoss, a division of Red Hat
xxxxxxxxxxxxxxxxxxxxxxxxxxxx
_______________________________________________
jboss-development mailing list
jboss-development(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/jboss-development