[
https://issues.jboss.org/browse/AS7-1656?page=com.atlassian.jira.plugin.s...
]
Paul Ferraro commented on AS7-1656:
-----------------------------------
I'm not sure this is really a problem. There are several mechanisms to demand the
jndi binding, @Resource injection, <resource-ref>, etc.
What I'd hate to lose is the coupling of the jndi resource to the application
lifecycle. One solution would be to introduce an additional start-mode for infinispan
caches that distinguishes between LAZY and ON_DEMAND. LAZY could install the binder
service using ACTIVE mode, where the cache container gets started lazily, while ON_DEMAND
would preserve the current behavior. This way, if some application needs to perform a
jndi lookup without a mechanism for demanding the binding to start, there is a
configuration option that supports this without having to resort to eagerly starting a
cache.
Binder services should not be ON_DEMAND
---------------------------------------
Key: AS7-1656
URL:
https://issues.jboss.org/browse/AS7-1656
Project: Application Server 7
Issue Type: Bug
Components: Clustering
Affects Versions: 7.0.1.Final
Reporter: Stuart Douglas
Assignee: Paul Ferraro
Fix For: 7.0.2.Final
IF a binder service is registered as ON_DEMAND then the item will not be available for
lookup in JNDI until something as expressed a dependency on it. This means that it will
work for resource injection, however will fail for manual JNDI lookups.
If lazy startup is required the binder service should use a ManagedReferenceFactory that
starts the lazy service and blocks until the lazy service is available.
--
This message is automatically generated by JIRA.
For more information on JIRA, see:
http://www.atlassian.com/software/jira