users@glassfish.java.net

Re: EntityManager's forced version update (lock(WRITE) + flush()) not atomi

From: <glassfish_at_javadesktop.org>
Date: Fri, 14 Nov 2008 13:21:09 PST

You are wrong - the where clause can't check the new value (otherwise it'll always fail, right?), but only the previous one to confirm that there were no parallel updates (which *is* the point of the optimistic concurrency).

I'll leave that thread to those who will be doing updates to the JPA APIs - you can't just change them without an approval from the spec lead, or even without a spec update (like in the javadoc case) - it's against complicated JCP rules.

Regards,
-marina
[Message sent by forum member 'mvatkina' (mvatkina)]

http://forums.java.net/jive/thread.jspa?messageID=316860