Shane Hathaway wrote: > Furthermore, you should not expect B to be updated implicitly at > transaction boundaries unless some object in B actually got changed. This is icky and unintuitive. Why is ZODB like that? Chris