what is the reason of having encodeEnd and encodeBegin final?
I have never understood why Trinidad is trying to hide possibilities from
the developer. Isn't it safe enough to say that renderers are not part of
the API, what you are doing there, might break with an update. Instead,
everyone is forced by this into "compliance mode"... how strange.
regards,
Martin
On Jan 22, 2008 10:24 PM, Andrew Robinson <[EMAIL PROTECTED]>
wrote:
> In the development of a library based on Trinidad, I need to have some
> hooks in the renderer to improve performance. Specifically, I need to setup
> a context and strip down a context for a component. Right now,
> CoreRenderer.encodeEnd (FacesContext, UIComponent) is final, so I cannot
> extend that method and the methods it calls I cannot reliably use (see
> below).
>
> What I propose is:
>
> new methods on CoreRenderer:
> protected void startComponent(FacesContext context, RenderingContext
> context, UIComponent component, FacesBean bean) {}
> protected void endComponent(FacesContext context, RenderingContext
> context, UIComponent component, FacesBean bean) {}
>
> These methods would be invoked from within
> "CoreRenderer.encodeBegin(FacesContext,
> UIComponent)" and " CoreRenderer.encodeEnd(FacesContext, UIComponent)"
> respectively.
>
> The benefit is that someone creating a component framework on top or
> Trinidad can extend the ability to setup and tear down settings when a
> component is being rendered. In my case, I need to extend the skinning code
> in a framework to enable certain functionality on a per-component basis from
> within the skin and have this functionality global to all our renderers.
>
> You may ask "why not just extend encodeAll?". Well the problem is that I
> have no idea when the child class will call this. For example:
>
> protected void encodeAll(FacesContext context, RenderingContext arc,
> UIComponent component, FacesBean bean) throws IOException
> {
> ... code
> super.encodeAll(context, arc, component, bean);
> ... code
> }
>
> As you can see, encodeAll doesn't give me the ability to have a hook from
> the renderer class level for when the component starts and ends. It doesn't
> make good OO code to have to manually code it in every one of our renderers.
>
>
> I also see the reason in having the methods final, so would prefer the
> hooks over removing the final keywords from encodeBegin and encodeEnd.
>
> Any objections to adding these 2 methods/hooks?
>
> If ppl. don't mind to respond quickly, I'll make the change if I get some
> quick positive feedback and no negative feedback. This is a fairly high
> priority feature that I need this for and cannot afford the 3 day vote
> period.
>
> Thank you,
> Andrew
>
--
http://www.irian.at
Your JSF powerhouse -
JSF Consulting, Development and
Courses in English and German
Professional Support for Apache MyFaces
