Yes, removing the call to updateActiveTransactionCache makes the test fail. My hope is that by adding an update timestamp condition to it, there will be many instances where the update will happen asynchronously while the commit is being completed so that on a busy system we don't add an overwhelming number of synchronous executions.
Good point about L2. It is now a subclass of T2 with almost no code and corrected Javadoc.
Yes, removing the call to updateActiveTra nsactionCache makes the test fail. My hope is that by adding an update timestamp condition to it, there will be many instances where the update will happen asynchronously while the commit is being completed so that on a busy system we don't add an overwhelming number of synchronous executions.
Good point about L2. It is now a subclass of T2 with almost no code and corrected Javadoc.