-
1. Re: UserTransaction and non-default JNDI context.
zacjacobson Jul 7, 2004 5:11 PM (in response to zacjacobson)I downloaded the source for JBoss 3.2.5, and the UserTransaction works the same way, so updating JBoss won't help me in this case.
If someone knows a workaround for this situation, I would definitely appreciate hearing it. -
2. Re: UserTransaction and non-default JNDI context.
starksm64 Jul 8, 2004 9:46 AM (in response to zacjacobson)Your going to have to hack the ClientUserTransaction to expose a thread local to which a Properties/Hashtable can be placed for use by the InitialContext. The j2ee appclient naming context factory should be providing a binding for the user transaction as well that would give more control over this, but currently its not.
-
3. Re: UserTransaction and non-default JNDI context.
zacjacobson Jul 8, 2004 1:41 PM (in response to zacjacobson)Thanks for the suggestion, I will look into that.
I also thought about making a UserTransaction available from Tomcat's JNDI (eg Tyrex) - but I don't really undersand if or how that would work - could commit/rollback on a UserTransaction that lives in Tomcat affect my beans in JBoss? -
4. Re: UserTransaction and non-default JNDI context.
perisb Mar 14, 2005 2:10 PM (in response to zacjacobson)"scott.stark@jboss.org" wrote:
Your going to have to hack the ClientUserTransaction to expose a thread local to which a Properties/Hashtable can be placed for use by the InitialContext. The j2ee appclient naming context factory should be providing a binding for the user transaction as well that would give more control over this, but currently its not.
I've got this problem, too, at version 3.2.5. Since this doesn't seem to be an overly exotic requirement, is there, or will there be a fix in the future? This may be a problem with the design of JNDI--I'm not sure.
Short of a fix, can anyone post a working version of the hack described? Thanks.
-Peris