Cool, I was actually in doubt whether to use containerId there as I didn't properly understand the workings of the transaction manager. Thanks for looking into it David.
Regards Manu On Sat, Mar 15, 2008 at 12:36 PM, David Jencks <[EMAIL PROTECTED]> wrote: > I took a look at OPENEJB-702, containing a (partly wrong) fix for the > lack of participation of message delivery in transaction logging. As > noted there, there's still the problem that the activemq vm transport > is broken and hangs if you attempt to get the resource manager id. > > So, just applying the patch (with fixes) breaks a lot of openejb > tests as the mdb delivery hangs, times out, runs out of connections, > etc etc. > > However, using the tcp transport works. While not ideal, I think > using a working although slower transport is a reasonable compromise > to a faster, broken transport until we can get a fixed activemq out. > > I changed 2 places configuring the vm transport, in openejb-core > service-jar.xml files. There's another one in openejb-tomcat-webapp > that doesn't appear to cause any test failures and I didn't change > knowing less than nothing about. > > This also directly uses the geronimo tm classes NamedXAResource and > WrapperNamedXAResource. I don't understand enough about plugability > in openejb to see how to make this dependency optional. > > Committed in rev 637368. > > I'm going to also look into enabling transaction logging for DBCP > although that looks to be considerably more complicated. > > thanks > david jencks > >
