> Autocommit does work sometimes, right? So there must be commit
> happening sometimes with autocommit -- besides I coded it in ;-)
> It's definitely not in a call to connection.commit(), though.

No, it seems to me that we have forgotten to add commit() when we
change the auto-commit state (same applies when we change the
transaction isolation). I'm going to check the specs right now...

Here it is... JDBC 3.0, page 62: "If the value of auto-commit is
changed in the middle of a transaction, the current transaction is
committed."

With transaction isolation they are more tolerant: "It is recommended
that drivers implement the setTransactionIsolation method to change
the isolation level starting with the next transaction. Committing
the current transaction to make the effect immediate is also a valid
implementation."

So, lets mark this issue as a bug and try to fix it. What about
managed scenario, David?

Best regards,
Roman Rokytskyy