How is the performance/network traffic if using client-side saveState? I have a data model that has hundred of rows.

Don't know.  I have not done any peformance testing as I am really just evaluating jsf and myfaces.  I think I would really consider design of the UI carefully.  Unless I really was required to display tables or trees of hundreds of rows/nodes, I would make the UI such that only a subset of the information is presented to the user at any point.  If there was such a requirement I would suggest that it should be changed - users will be happier.

My test application uses myfaces/spring/hibernate stack.  The UI and hibernate are coupled such that the queries have limits and filters passed to them.  The application is completely stateless - the jsf tree is saved on the client and all the beans are request scope.

Does anyone know of any performance results/thoughts comparing stateless versus stateful jsf apps?

My guess is that network traffic/marshalling (unmarshalling) of jsf-view still beats having a bunch of users with a session object as long as you keep the saveState stuff to a reasonable amount of data.

=======================================
Ryan Wynn
Websphere Portal Developer
IBM Business Consulting Services - Public Sector
Reston Office: 703-668-2478
Cell:  703-622-1977
[EMAIL PROTECTED]



Dave <[EMAIL PROTECTED]>

09/27/2005 01:46 PM

Please respond to
"MyFaces Discussion"

To
MyFaces Discussion <[email protected]>
cc
Subject
Re: How to deal with ugly issue - Back/Refresh





How is the performance/network traffic if using client-side saveState? I have a data model that has hundred of rows.

Ryan Wynn <[EMAIL PROTECTED]>
wrote:


No, I use client that must be why.  Thanks.


=======================================
Ryan Wynn
Websphere Portal Developer
IBM Business Consulting Services - Public Sector
Reston Office: 703-668-2478
Cell:  703-622-1977
[EMAIL PROTECTED]


Martin Marinschek <[EMAIL PROTECTED]>

09/27/2005 10:19 AM

Please respond to
"MyFaces Discussion"


To
MyFaces Discussion <[email protected]>
cc
Subject
Re: How to deal with ugly issue - Back/Refresh







It' s only relevant for server side state saving - do you use that?

regards,

Martin

On 9/27/05, Ryan Wynn <
[EMAIL PROTECTED]> wrote:

How does this problem present itself?  My app is stateless and uses saveState at some points.  I have not had any problems with the back button.



=======================================
Ryan Wynn
Websphere Portal Developer
IBM Business Consulting Services - Public Sector
Reston Office: 703-668-2478
Cell:  703-622-1977

[EMAIL PROTECTED]
Werner Punz <[EMAIL PROTECTED]>
Sent by: news <
[EMAIL PROTECTED]>

09/27/2005 02:20 AM

Please respond to
"MyFaces Discussion"


To
[email protected]
cc
Subject
Re: How to deal with ugly issue - Back/Refresh









Ok I think there is a pattern to deal with http post problems,
however I have not dealt with it yet, others might be able
to help out
I think it is called somewhat like redirect after
post...
You also might have to check out the shale tokens
system it might help you out as well.

Sorry that I can not help you further.

Werner


Dave wrote:
> HI Werner,
> Thanks. <x:saveState> does not work in this case. Data saved using
> <saveState> are only available to the immediately following
> request.(post back in most cases).
>  
> When user click Back, the browser says something like "not available
> since previous Form post, ask to refresh ". When use click Refresh, the
> data saved using <x:saveState> is not there any more because it is not
> immediate PO ST back.
>  
> Dave
>
> */Werner Punz <
[EMAIL PROTECTED]>/* wrote:
>
>     Dave wrote:
>     > When a backing bean is dependent on some data that are initialized
>     > by previous page, it works fine normally. But if user click
>     Back/Refresh
>     > on browser, then the data is not available anymore to the backing
>     bean.
>     > This will result in NullPointerException because the data is null. How
>     > to catch this issue and display a nice informative message, definitely
>     > not Exception? Thanks for ideas. Dave
>     >
>     Sorry, my first answer was too short, to clarify things a little bit,
> &n bsp;   the problem you encounter is due to the request scope, a bean declared
>     in request scope lives only one request, now if you do a refresh you
>     might lose
>     important data along the way, the same goes for back.
>
>     There are various solutions to the problem, the shale token stuff, I
>     yet have to check out,
>     but one solution is to use constructs w hich enable scopes bigger
>     than request.
>     One solution definitely would be to dump the backend bean into the
>     session
>     that way non traversed values would survive a refresh.
>     Another one would be to use a lot of hidden fields, for the non
>     visible data which has to survive
>     which is a maintenance nightmare.
>
>     Generally the cur rently possibly best approach would be to use
>     x:saveState which stores
>     beans for every page where a x:saveState tag is hit with the same
>     declaration,
>     or to use some kinde of dialog-scoping system:
>
>     Following systems currently are available:
>     Shale Dialog
>     Seam
>
>     Seam at least is rather heavy on the configuration side, but it
>     looks like the most interesting
>     approach in this area.
>     Shale Dialog I have not looked into yet,
>
>     the possible most lightweight approach however probably is
>     x:saveState...
>
>
> ________ __________________________________________
> Do You Yahoo!?
> Tired of spam? Yahoo! Mai l has the best spam protection around
>
http://mail.yahoo.com
>






--


http://www.irian.at
Your JSF powerhouse -
JSF Trainings in English and German


Yahoo! for Good
Click here to donate to the Hurricane Katrina relief effort.

Reply via email to