It was a request bean, set to session scope to test out the new build. Actually only a few items in the bean was meant to be in the saveState, saving the whole bean was only a lazy way to get everything working before determining which objects need to be saved.
What is the performance implication of SERIALIZE_STATE_IN_SESSION=true? Should this be set to false unless I need to implement clustering that requires sharing session between containers? Regards, Yee -----Original Message----- From: Mathias Brökelmann [mailto:[EMAIL PROTECTED] Sent: Monday, 19 December 2005 4:55 PM To: MyFaces Discussion; [EMAIL PROTECTED] Subject: Re: Session Scope Bean does NOT require serialization. Myfaces bug? I wonder why you use t:saveState for a session bean? We changed state saving (only server side) to serialize state into the session after 1.1.1. This new behavior can be changed by a web-app init parameter org.apache.myfaces.SERIALIZE_STATE_IN_SESSION which is true by default. 2005/12/19, Yee CN <[EMAIL PROTECTED]>: > > > > Hi, > > > > I have a bean that is NOT serializable but yet I am able to set its scope to > session. So does that mean that putting a bean to session scope does not > requires serialization? Or is it a myfaces bug? > > > > I am raising this question because I have a piece of code that worked in > 1.1.1 release but failed in latest nightly build. The piece of code is > trying to do: <t:saveState value="myNotSerializableBean" />, so it should > rightly fail and it is a bug that got fixed in the nightly built. That this > bean is still able to participate in Session scope makes me wonder whether > there is a lingering bug in Myfaces state saving. I am using the 20051217 > build. > > > > Best regards, > > Yee > > > > -- Mathias