Ok, so getting the datasources straightened out was not like differentiating a quadratic equation, but it was just a bunch of painful drudgery. A lot of it was slop lying around that Tomcat was not concerning itself with, but JB7 was finding, and some of it was the crackpot eclipse cycle of having to clean and refresh, etc. every time you do a config change.
But that wasn‘t the end either. Our project uses eh-cache and we wasted almost another hour trying to make that work.
I'm actually really interested in using Infinispan for the cache..
But in the meantime, literally hours went in a hole fiddling around with the caching configuration. First, trying to set it to ehcache produced a slew of errors (mostly class crap). So we decided to try and go in the Infinispan direction. It didn‘t like the READ_WRITE annotation on entities. Started replacing those with TRANSACTIONAL. Boy, that was such fun. Finding every annotation in every project. Annotations can be kind of stupid, for things like this for instance. How is separation of concerns being honored here? What on earth does a cache policy have to do with the entity itself? But, finally, that all was working and hibernate was sufficiently satisfied that it rolled through all its crap. So we started making arrangements to pull in the fatted calf again. But it was not to be. We struggled for a while longer with a crazy casting error (of the Hibernate Session) that is done inside Lucene, it turns out. Then, alack, Pugachev‘s little ox cart ran aground after wandering the countryside with a simple proclamation: JB7 does not support Hibernate Search 3.
So the total amount of time now has officially reached kookie. Like, I could have probably ported a large amount of the code to 6 in this amount of time.
That said, I am far from soured on JB7. I am still really looking forward to it. I would warn any who are thinking of wandering into the woods with a 5-era project that has a significant number of dependencies, you may be down in the sewer a long time.