RE: [Springframework-developer] Portlet support (again)

[Nick L. <nic...@es...>]

Comments inline.

> 
> Aloha!
> 
> I sent a patch, but meanwhile I've done more.
> I've attached a zip file containing my changed and added files to 
> confluence:
> http://opensource.atlassian.com/confluence/spring/display/JSR1
> 68/Home?showComments=true#comments
> Below is a copy of the comment I posted there.
> 
> Cheers,
> Rainer
> 

From a quick look this looks pretty promising.

Does org.springframework.web.portlet.mvc.SimpleFormController replace
org.springframework.web.portlet.SimplePortletFormController?

If not then when should SimplePortletFormController be prefered?


> 
> This zip contains some (many) changes to the CVS code.
> Mainly the form handling code is affected, trying to build a form 
> controller hierarchy similiar to the servlet tree.
> A new controller type is ActionFormController for forms requiring 
> processing in the action phase, e.g. to call 
> ActionResponse.sendRedirect(String), 
> ActionResponse.setPortletMode(PortletMode), or link 
> ActionResponse.setWindowState(WindowState).
> A controller extending this is RedirectFormController.
> 
> Another new class is ParameterControllerMapping to map 
> controllers by a 
> given request parameter, and ParameterMappingInterceptor to intercept 
> those parameters in ActionRequests and forward it to the according 
> RenderRequest.
> 
> The code works for me but needs some more testing and documentation.
> 
> Generally it might be necessary to change some method 
> signatures in base 
> classes from using ActionRequest/Response or 
> RenderRequest/Response to 
> use the more general PortletRequest/Response. Then implementing 
> controllers have more freedom to choose wether the processing 
> will take 
> place mainly in the render phase (as proposed in the original 
> code) or 
> in the action phase (as implementd by ActionFormController). After 
> implementing a few portlets I can imagine the need for portlets doing 
> more work in the action phase and forward the result to the 
> render phase 
> as parameter or even in the portlet session.

Yes, I agree this would be a good change and I think it should be made.


> In addition the next version of the portlet specification might relax 
> the communication constraints between action and render 
> phases (e.g. by 
> allowing to set RenderReuqest attributes).  The proposed changes will 
> make it easier for implementing controllers to implement 
> those extensions.
> 
> 

I'd like to think that might be the case but some people on the JSR
committee seemed pretty opposed to this when it came up on the pluto-dev
list.

Nick