JBCACHE-1514 is not an issue for the AS. As Galder described, the
CacheManager is bound in JNDI, and only in the java: namespace. The
NonSerializableFactory is just a javax.naming.spi.ObjectFactory that
uses a static map to cache objects. It will only return objects to
callers inside the same VM.
On 08/11/2009 09:14 AM, Galder Zamarreno wrote:
On 08/11/2009 02:57 PM, Galder Zamarreno wrote:
> On 08/11/2009 01:54 PM, Manik Surtani wrote:
>> On 11 Aug 2009, at 12:43, Galder Zamarreno wrote:
>>> On 08/11/2009 01:20 PM, Manik Surtani wrote:
>>>> Brian, how important is
>>>> to you? Or do you not bind cache instances to JNDI/not see this
>>>> problem with JBoss AS' JNDI impl?
>>> Afaik, Only AS's cache manager
>>> (org.jboss.ha.cachemanager.CacheManager) is bound to JNDI. I don't
>>> think we recommend binding cache instances but instead binding some
>>> form of cache manager instead. Not sure if you can directly bind the
>>> JBC's default cache manager.
>> Surely then AS's cache manager has references to Cache instances? So
>> these are either transient or exposed to potentially being serialized as
The thing is, org.jboss.ha.cachemanager.CacheManager is not Serializable
and hence, Brian has used a different method that involves binding
CacheManager to the NonSerializableFactory and then binding a JNDI a
Serializable reference (javax.naming.Reference) that seems to point to
the non serializable cache manager.
I don't know much about the method other than I was able to read from
the code but just in case, here's the CacheManager class file:
So, to sum up, I can't say whether 1514 affects AS or not.
> Let me check AS trunk...
>>>> Manik Surtani
>>>> Lead, Infinispan
>>>> Lead, JBoss Cache
>>>> jbosscache-dev mailing list
>>> Galder Zamarreño
>>> Sr. Software Engineer
>>> Infinispan, JBoss Cache
>> Manik Surtani
>> Lead, Infinispan
>> Lead, JBoss Cache