I have been working on an implementation of the JCache as proposed in the community process for JBoss.
I seem to have run into a brick wall regarding documentation. (I have purchased the full set of docs available here).
My questions are there:
1) I was planning to use the HAJNDI environment to contain the cached data or to use some other mechanism in JBoss (DistributedState?). The advantage to the HAJNDI structure is its clarity (easy to inspect) and it's tree structure (easy to store hierarchical data). The problem is that I can find no way of inserting data or NamingContext(s) into HAJNDI. How do I do it?
2) Is there a better way? A truly independent mechanism would allow for the use of JMS or multicasting to manage the data naming tree and allow for inter(ra)server requests for data. It not too difficult to write, but why would I re-write what is already there?
3) Is there an existing example of the use of the DistributedState outside of entity beans?
Have you looked at the JBoss Cache project ? We're working on a replicated transactional cache, and it will be JCache compliant.
My goal is to be able to provide it as a standalone JCache implementation, that will even be able to run outside the JBoss container.