ok, Howard, I added the bug to JIRA. Regards.
Howard Lewis Ship wrote: > > Before making changes, I'm checking on the Hibernate user forum. > > http://forum.hibernate.org/viewtopic.php?p=2387715#2387715 > > Please still add the bug to JIRA. > > On Wed, Jun 11, 2008 at 9:53 AM, Howard Lewis Ship <[EMAIL PROTECTED]> > wrote: >> Thanks for the heads up. >> >> Looking at the Hibernate source, I can see that this is true, when you >> commit() a transaction, the Session's connection to the transaction >> (via the JDBCContext) is lost. It seems like, at least for the JDBC >> case, the Transaction object should only have begin() invoked once. >> >> Please add a bug to JIRA; I'll change the source to invoke >> session.beginTransaction() rather than Transaction.begin() after >> commit or rollback. >> >> On Tue, Jun 10, 2008 at 5:30 PM, raulmt <[EMAIL PROTECTED]> wrote: >>> >>> Hi, >>> >>> I'm using T5.0.12, Hibernate 3.2.6GA and MySQL 5.1. I ran into a problem >>> when I wanted to make two commits on the same request. The first commit >>> worked fine, but the second works "partially" (for example, an entity >>> deletion on the second transaction doesn't ocurr....). >>> >>> Debugging, I realized that before the first call to >>> HibernateSessionManager.commit(), if I call >>> hsm.getSession().getTransaction() (which gets the underlying transaction >>> from the jdbcContext), the Transaction returned is the same instance >>> that is >>> stored in a property on the hsm, but after calling hsm.commit(), >>> hsm.getSession().getTransaction() returns another instance of >>> Transaction... >>> >>> I guess the problem is that hsm.commit() after committing the >>> transaction it >>> obtained when created the Session, instead of obtaining again the >>> (apparently new) underlying transaction calling >>> session.getTransaction(), it >>> calls begin() on the same transaction it has stored on its property, >>> which >>> isn't the real underlying transaction and because of that, the >>> transaction >>> doesn't really begin again. If, after calling hsm.commit() I do a >>> hsm.getSession().getTransaction().begin() (which I assume really starts >>> another transaction on the database) all works perfectly. >>> >>> I'm not neither a Tapestry nor Hibernate expert, but I think maybe >>> hsm.commit() should get again the transaction with >>> session.getTransaction >>> after it commits the current transaction. >>> >>> Regards. >>> -- >>> View this message in context: >>> http://www.nabble.com/-T5--tapestry-hibernate-and-transactions-tp17767677p17767677.html >>> Sent from the Tapestry - User mailing list archive at Nabble.com. >>> >>> >>> --------------------------------------------------------------------- >>> To unsubscribe, e-mail: [EMAIL PROTECTED] >>> For additional commands, e-mail: [EMAIL PROTECTED] >>> >>> >> >> >> >> -- >> Howard M. Lewis Ship >> >> Creator Apache Tapestry and Apache HiveMind >> > > > > -- > Howard M. Lewis Ship > > Creator Apache Tapestry and Apache HiveMind > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [EMAIL PROTECTED] > For additional commands, e-mail: [EMAIL PROTECTED] > > > -- View this message in context: http://www.nabble.com/-T5--tapestry-hibernate-and-transactions-tp17767677p17784231.html Sent from the Tapestry - User mailing list archive at Nabble.com. --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
