[seam-issues] [JBoss JIRA] (JBSEAM-4395) Endless loop when an exception occurs during commit inside a long running conversation, and there is an <exception> clause to handle it in page.xml

[Johan Gärde (JIRA)]

    [ https://issues.jboss.org/browse/JBSEAM-4395?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12736703#comment-12736703 ] 

Johan Gärde commented on JBSEAM-4395:
-------------------------------------

I think the problem is that the exception handler parser (org.jboss.seam.exception.Exceptions) ignores the before-redirect attribute of the end-conversation element.
If the exception handler parser is changed so that the before-redirect attribute is read and the RedirectHandler would make use of this information and actually end the conversation before the redirect I think this will solve the situation.


                
> Endless loop when an exception occurs during commit inside a long running conversation, and there is an <exception> clause to handle it in page.xml
> ---------------------------------------------------------------------------------------------------------------------------------------------------
>
>                 Key: JBSEAM-4395
>                 URL: https://issues.jboss.org/browse/JBSEAM-4395
>             Project: Seam 2
>          Issue Type: Bug
>          Components: Exception Handling
>    Affects Versions: 2.2.0.GA, 2.2.1.CR1
>         Environment: WebSphere v7.0.0.5, Seam 2.2.1-SNAPSHOT
>            Reporter: Denis Forveille
>
> Scenario:
> - process is in a long running conversation
> - a persistence exception occurs (e.g. OptimisticLockingException), during either a call to flush() or during the automatic commit() of the transaction 
> - a clause <exception> is declared in pages.xml to handle the exception and redirect to a generic error page (e.g. <exception> that will handle all exceptions..) like this:
>   <exception>
>       <end-conversation/>
>       <redirect view-id="/erreur_exception.xhtml" />
>   </exception>
> During the commit, the exception occurs, Seam catch it and pass it to the exception handler.
> The exception handler sees that the user has to be redirected to the error page.
> The long running conversation is still active, handling the "dirty" objects that cause the persistence failure 
> The request to process the error page starts. During the "RESTORE_VIEW", the dirty object that are in the conversation are then flushed again to the database, causing again the persistence exception, triggering the seam exception handling mechanism, that will redirect to the error handling page, causing an endless loop
> A workaround to this is to add an observer on "org.jboss.seam.exceptionHandled" that will kill the current conversation and so remove the dirty objects from the session and the error page is display (see the forum reference)
> Some thoughts:
> - the documentation gives an example on how to handle OptimisticLockingException, but in our case, the way it is, this causes an infinite loop
> - is the <end-conversation/> clause working correctly? At what time during exception handling is the end-conversation supposed to occur?

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira