springframework-developer

[Springframework-developer] PROPAGATION_REQUIRES_NEW and JTA

[Jim E. <jd...@ca...>]

Hi Folks,

I'm wondering if it's possible to use PROPAGATION_REQUIRES_NEW *without*
using JTA.  I'm trying to use PROPAGATION_REQUIRES_NEW with the
following simple setup (using spring-framework-1.1):

HibernateTransactionManager whose sessionFactory is
a LocalSessionFactoryBean, and the sessionFactory
dataSource is a JDBC DriverManagerDataSource.

This isn't yielding the results I want yet.  
I don't get any exceptions or obvious error
output, but I don't get the transactional behavior I expect either.
I'll need to do further work to diagnose, but first I  
just want to make sure that such a configuration is intended to work.

Based on this post from Colin Sampaleanu, I suspect that JTA
may be required:

----
REQUIRES_NEW actually just suspends the current transaction, and then 
 starts a new, unrelated transaction transaction to run the contained 
 code in. If you commit the inner, new transaction, then rollback the 
 outer, the inner transaciton will still be committed, as they are 
 unrelated. This should work properly inside any JTA environment (as long 
 as spring can get at the jta transaction manager) and the model is also 
 essentially the same as the requires new support in the EJB spec.
----

Thanks for listening,
Jim Engquist

Re: [Springframework-developer] PROPAGATION_REQUIRES_NEW and JTA

[Seth L. <set...@gm...>]

On Thu, 06 Jan 2005 10:27:35 -0700, Jim Engquist <jd...@ca...> wrote:
> 
> Hi Folks,
> 
> I'm wondering if it's possible to use PROPAGATION_REQUIRES_NEW *without*
> using JTA.  I'm trying to use PROPAGATION_REQUIRES_NEW with the
> following simple setup (using spring-framework-1.1):
> 
> HibernateTransactionManager whose sessionFactory is
> a LocalSessionFactoryBean, and the sessionFactory
> dataSource is a JDBC DriverManagerDataSource.
> 
> This isn't yielding the results I want yet.
> I don't get any exceptions or obvious error
> output, but I don't get the transactional behavior I expect either.
> I'll need to do further work to diagnose, but first I
> just want to make sure that such a configuration is intended to work.


If your setup is something like this, it won't work:

public class Bean {
   // marked as REQUIRES
  public void doA() {
    doB(); // run as REQUIRES_NEW
    doB(); // run as REQUIRES_NEW
  }

  public void doB() {
    ...
  }
}

Because when doB is called from inside doA, it's not going through the
proxy.  Therefore, it bypasses the transaction code.  Maybe this is
why you're not seeing the behavior you expect?

Seth

-- 
<a href="http://www.picklematrix.net/foaf.rdf">Seth Ladd's FOAF</a>
<a href="http://www.foaf-project.org/">What is FOAF?</a>

RE: [Springframework-developer] PROPAGATION_REQUIRES_NEW and JTA

[Juergen H. <ju...@in...>]

Transaction suspension should work with all of Spring's local transaction
managers too, including HibernateTransactionManager. It's even simpler than
in the JTA case: no need for special server-specific setup.

Juergen


-----Original Message-----
From: spr...@li...
[mailto:spr...@li...]On Behalf
Of Jim Engquist
Sent: Thursday, January 06, 2005 6:28 PM
To: spr...@li...
Subject: [Springframework-developer] PROPAGATION_REQUIRES_NEW and JTA



Hi Folks,

I'm wondering if it's possible to use PROPAGATION_REQUIRES_NEW *without*
using JTA.  I'm trying to use PROPAGATION_REQUIRES_NEW with the
following simple setup (using spring-framework-1.1):

HibernateTransactionManager whose sessionFactory is
a LocalSessionFactoryBean, and the sessionFactory
dataSource is a JDBC DriverManagerDataSource.

This isn't yielding the results I want yet.
I don't get any exceptions or obvious error
output, but I don't get the transactional behavior I expect either.
I'll need to do further work to diagnose, but first I
just want to make sure that such a configuration is intended to work.

Based on this post from Colin Sampaleanu, I suspect that JTA
may be required:

----
REQUIRES_NEW actually just suspends the current transaction, and then
 starts a new, unrelated transaction transaction to run the contained
 code in. If you commit the inner, new transaction, then rollback the
 outer, the inner transaciton will still be committed, as they are
 unrelated. This should work properly inside any JTA environment (as long
 as spring can get at the jta transaction manager) and the model is also
 essentially the same as the requires new support in the EJB spec.
----

Thanks for listening,
Jim Engquist





-------------------------------------------------------
The SF.Net email is sponsored by: Beat the post-holiday blues
Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
_______________________________________________
Springframework-developer mailing list
Spr...@li...
https://lists.sourceforge.net/lists/listinfo/springframework-developer