if gili wants to go test our new PageState mechanism by making a nice subclass of PageState using xstream (XStreamPageState?), that might make a really nice contrib or extensions class...


Christopher Turner wrote:

Erm, how is that of any use to a generic web framework? People are going deploy Wicket applications on Tomcat, Jetty, JRun, Websphere, Weblogic, Pramati, Oracle/Orion and possibly many others. I doubt every one of these vendors is going to switch to XStream. The best we can ever do is support the SERVLET 2.4 specification because that is the only given in every environment.
Could I therefore politely request that we (you) forget about XStream or any other approach that is not compatible with the servlet specification. That way we can spend our time solving the problem in a realistic and portable way.
Regards,
Chris


    Chris,

    Ok, so I filed a RFE with Tomcat to add XStream support (ability
    to use it instead of the default mechanism). Although, I don't
    have much confidence in the Tomcat crowd anymore :(

    Thanks,
    Gili

    --Original Message Text---
    *From:* Christopher Turner
    *Date:* Tue, 8 Mar 2005 16:45:46 +0100

    Message The reason for not using something like XStream is that we
    don't have any control over the serialization of objects in the
    session. This is entirely the responsibility of the application
    server and clustering engine. It is not something that we have any
    option to override. For a typical request the following happens:

    - request arrives at application server
    - application server finds the JSESSIONID cookie and retrieves the
    session from somewhere (memory, disk, cache server etc.)
    - the target servlet is identified and the objects are sent to
    this for processing
    - Wicket code runs
    - the servlet returns
    - the application server looks to see if the session is changed
    and synchronizes the session across the cluster
    - the application server stores the session away somewhere
    (memory, disk, cache server etc.)

    Everything related to session storage, synchronization and so on
    happens IN the application server code base so we really have
    ascolutely no control over this. In fact there is no guarantee
    that the application server will even use the standard Java
    serialization mechanism in its own implementation. To quote the
    Servlet 2.4 specification:

    "If distributed containers persist or migrate sessions to provide
    quality of service features, they are not restricted to using the
    native JVM Serialization mechanism for serializing HttpSessions
    and their attributes. Developers are not guaranteed that
    containers will call readObject and writeObject methods on session
    attributes if they implement them, but are guaranteed that the
    Serializable closure of their attributes will be preserved."

    Therefore anything other than making sure objects implement
    Serializable is not possible!

    regards,
    Chris



    Again coming back to this point: why are we not using XStream that
    allows us to handle everything automagically?

    Gili

    --Original Message Text---
    *From:* Johan Compagner
    *Date:* Tue, 08 Mar 2005 15:49:18 +0100

    because the model itself will still be serialized.
    A detachable model doesn't mean the model itself is detached but
    the developer can detach the contents.

    For example if you have a heavy weight component in your model
    that you can relookup from somewhere
    by an ID. Then in the detach you store that id in a non transient
    instance reference. And youre heavy weight object
    is cleared or hold only in a transient reference.

    Then when the component and its model are serialized only the id
    is serialized with the model not youre complete object.

    johan


Cameron Braid wrote: Speaking about serializable stuff...

    With detachable models, why do they also have to be serializable ?

    Cameron.

    -----Original Message-----
    From: [EMAIL PROTECTED]
    [mailto:wicket-develop-
    [EMAIL PROTECTED] On Behalf Of Johan Compagner
    Sent: Wednesday, 9 March 2005 12:28 AM
    To: Wicket Developer List
    Subject: [Wicket-develop] should we make AbstractResource
    serializeable or
    not?

    Our clone by serialize did encounter a few serialization
    exceptions for
    example
    all the Resources (like image). Should we make AbstractResource
    serializeable?
    those can be in the session..

    johan




------------------------------------------------------- SF email is sponsored by - The IT Product Guide Read honest & candid reviews on hundreds of IT Products from real users. Discover which products truly live up to the hype. Start reading now. http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click _______________________________________________ Wicket-develop mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/wicket-develop




------------------------------------------------------- SF email is sponsored by - The IT Product Guide Read honest & candid reviews on hundreds of IT Products from real users. Discover which products truly live up to the hype. Start reading now. http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click _______________________________________________ Wicket-develop mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/wicket-develop









-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
_______________________________________________
Wicket-develop mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/wicket-develop

Reply via email to