Hi!
I am absolutely sure, that it works with myfaces-orchestra-core-1.0.jar - I
need only to replace the .jar to see the difference! Nothing else changes!
Strange, I have no clue why Orchestra 1.0 will avoid this problem.
Unfortuantly, we can not allow the user to stay online all the time. Each
opened session consumes memory (we have many statefull beans with session
life time).
I understand, but notice: Using the poller keeps the session open only
as long as a browser is pointing to your application. On the other hand,
once the browser has been closed the session can die much faster.
If the session scoped beans are a problem think about putting them into
a separate (probably non-persistence-linked) conversation.manual scope.
If accessed every now and then it lives as long as the session.
Together with the poller then, the conversation.manual scope is able to
timeout and release memory, but the session is still alive with just a
handful of data in it.
I do it that way with our application.
Can you explain pls, why It works, if I use navigation with
<navigation-case> without <redirect /> and does not work wit <redirect />?
No, I can't. Are you using client-side state saving? Probably without
<redirect /> JSF is able to restore itself to the point where it is
possible then to render again, though, it should not work either I think.
Do you have any chance to try the JSF RI Mojarra? Just wondering if it
makes any difference there too.
Is there any workaround to generally avoid this exception or to handle it
correctly?
Probably you can configure an error handler for this exception in your
web.xml (I think it is called error-type where you can configure an
error page per exception) and explain the user that the session has expired.
Ciao,
Mario
