Please keep in mind that its a fair amount of time as well for us to vet these issues (immersion and all that). I spent a lot time trying to help you and figure out what it is you are actually (1) reporting and (2) wanting here. I appreciate the fact that you spent time, but I do not appreciate arguing when I am simply trying to figure out what you are reporting/wanting. I was done arguing with you so I moved on to other things. I don't do arguing.

So again:

It seems you are reporting that the underlying exception gets "eaten" (lost) in WebLogic. But I have yet to see anything that shows this is more than just WebLogic-specific. Where would I look in your attachments/comments to see differently? And no, I do not consider the original exception being nested/wrapped inside some form of a "JTA rollback exception" to be "lost". That's just simply the best outcome you will get in a JTA environment where you are allowing the flush to be initiated from a commit flowing from the JTA TM to Hibernate.