Not sure (one would think it is done once) but as a workaround you could have an application-level boolean
initialized (and check that the session is open) before doing any loading.

I got the same problem that the org.jboss.seam.postInitialization event is triggered several times, but for me it happens during/after hot deployment. An application scoped boolean doesn't work out since the event gets called all the time and it never stops doing so. This means I should probably switch off hot deployment again? That would probably work out for me, but not for Dmitry I guess. Any solution someone?

I guess hot deployment triggers org.jboss.seam.init.Initialize.redeploy
which runs org.jboss.seam.contexts.ServletLifecycle.endInitialization
which runs Events.instance().raiseEvent(org.jboss.seam.postInitialization);

Bug, please report in JIRA

Umm, me or someone who has actually seen the bug? ;-)

Reminds me of this closed bug.

Damian Harvey wrote on Oct 02, 2008 12:29:


Reminds me of this closed bug.

Slightly different ;-)

https://jira.jboss.org/jira/browse/JBSEAM-3489

I think you guys were faster than I :) Thanks for your quick response guys! As Dan said here My Link

It get's cold around here without that hot deployment ;)

I agree :)

Old code handling org.jboss.seam.postInitialization for scheduling stopped working after upgrade from 2.0.3CR1 to 2.1.0CR1.

In our case we have 3 beans.
First class A is application-scoped, stateful, and autocreated.
It has a method that observes org.jboss.seam.postInitialization, standard stuff.
It picks up an instance of another ordinary bean, B, which injects a bean C with

create=true

public class B {
  @In(create = true)
  private C c;

  @Asynchronous
  public QuartzTriggerHandle scheduleDelete(
    @Expiration Date when,
    @IntervalDuration Long interval,
    Date stoptime) {
      this.c.delete();
      return null;
  }
}

Now what stopped working is that class C that has alot of injected objects; logger, entity-manager and some other application-scoped
beans, that now are all nulled out, thus injection seem to fail.
This worked fine and they were all injected correctly before the version upgrade.

Is it a bug or is 2.1.0 handling startup and initializing in some different way?

/K

Please try trunk, I fixed a bug here very recently

Tried to reactivate the old code again after a synchronize/build of seam 2.1 Trunk. Still no luck, sorry.

Do you have enough info to persue this issue?

/K

Kristoffer Teuber wrote on Oct 14, 2008 20:04:


Tried to reactivate the old code again after a synchronize/build of seam 2.1 Trunk. Still no luck, sorry.

Do you have enough info to persue this issue?

/K

Can you file an issue in JIRA with a complete set of instructions to reproduce, as I get lost when we have a long thread like this ;-)