Nope, I just figured it out... not a JPA problem at all, it was ActivityRecoveryTest that was using the PMAPI behind my back..
-mbs On 9/10/07, Paul Brown <[EMAIL PROTECTED]> wrote: > > On 9/10/07, Maciej Szefler <[EMAIL PROTECTED]> wrote: > > On one occasion, while in the debugger, I had a run that did not have > the > > problem (but that run had another problem with a different field being > > null). My observations suggest that the transactions are fine, that is, > the > > problematic transaction is happening after the transaction that should > have > > set this value is good and committed (actuallly otherwise the row would > not > > exist at all). > > Is this the same problem that we had years ago (2003?) that drove us > to Hibernate from Kodo? Is it specific to a database, e.g., Derby but > not MySQL? > > -- Paul >
