Really, it's not a bug. Different JSF specifications have a different
behaviors for a this case. So, RichFaces classes should be JSF 1.2
compatible for a ViewExpiredExceptions.
SUN RI check VievHandler/StateManager instances at the restore view
phase. For a JSF 1.2, no view is created by framework, but, for a JSF
1.1 classes, they should call ViewHandler.createView method , for an
empty UIViewRoot creation.
I'll chech implementation code for a correct support of the all cases.
This is not a problem for a RichFaces 3.2. development branch, so it
already targeted for a JSF 1.2
Pete Muir wrote:
But we're not using a4j/rf components, we are just submitting a
h:form
with a h:commandButton - so the behaviour should *not* deviate from
the underlying JSF impl.
And actually, if you do some debugging, you'll see that the problem
isn't that RF needs to throw a ViewExpiredException, but that, in the
updateModel method of UIInput without RichFaces the viewRoot is null,
so the RI throws the exception, with RichFaces, it isn't null, so the
exception isn't thrown.
I don't know if this is a bug in RF or in the JSF RI - if you need me
to chase this down I guess that I can spend time on it but I would
prefer not to.
Thanks
On 19/02/2008, Alexandr Smirnov <asmirnov(a)exadel.com> wrote:
> ViewExpiredException is a JSF 1.2 feature. RichFaces 3.1.x keep JSF 1.1
> compatibility, so, it follow to jsf 1.1 specifications.
> To support ViewExpiredException, we need a switch dependencies to JSF
> 1.2, as far as no such class in the JSF 1.1 API.
> Pete Muir wrote:
>
>>
http://jira.jboss.com/jira/browse/RF-2294
>>
>> This is one is affecting
SeamFramework.org and is urgent. We need to
>> get a fix into the 3.1.x branch.
>>
>> Thanks guys :)
>>
>>
>>
>