Thanks all for your responses. The behaviour of Derby client driver does not seem right to me too. Also, it is different from embedded driver. I will open a JIRA issue shortly.
I agree this is a corner case. It is unlikely for an application to have the code I mentioned in my mail. If autocommit is set to false, the application will be doing a commit/rollback at the end and then we would not see this exception. Thanks, Deepa On 10/21/05, Lance J. Andersen <[EMAIL PROTECTED]> wrote: > The JDBC javadocs indicate the following > > > Note: If this method is called during a transaction, the result is > implementation-defined. > > > Unfortunately many of the corner-cases such as this will vary from vendor > to vendor so the best thing to do is avoid this type of behavior in your > apps if you are concerned about portability. > > We have tried where we can in JDBC 4 to clarify issues such as this to at > least let you know if your milage may vary and if it hurts, dont do it ;-) > > > Andreas Korneliussen wrote: > Roy's email was meant to be sent to me personally to correct my answer. > > He says that according to the SQL spec, it is not allowed to set > transaction isolation inside a transaction, however setting transaction > isolation should not open a transaction either. > > My answer was based on the "JDBC API Tutorial and Reference, Third > Edition". > > -- Andreas > > > Nope. Det er ikke tillatt å sette transaction isolation inne i en > transaksjon, men det å sette transaction isolation skal heller ikke åpne en > transaksjon... > > Roy > > >
