Nice improvement. This will make VWP a lot cleaner and neater...

Posted by Yossarian on January 14, 2008 at 11:29 PM PST #

Hi Winston,

That is great news...

May I ask you a question:
I am trying to dynamically change the parent of JSF component..
Let me explain my context:
Components are initially put via Netbeans interface into JSF pages but without any style (that is creating an initial component tree structure)..
The user of my application might (via an heavy client whick takes the role of layout administrator) change the components look (style:position, background color...etc) and also move a component from Panel (or GroupBox) in Tab1 to a panel (or GroupBox) in Tab2 for instance.

What i have achieved till then: JSF component are dynamically positioned and styled when the user click on each tab rendering exactly what he is seeing on heavy client application.... lot of work indeed, but this works perfectly (I need however to have a look on performance issues) when user do not change any parent of the components....In other words, when initial component tree structure is not modified.

BUT, when I try the use case described above (from tab1 to tab2)...it works when the page is first rendered (component is actually in Tab2 and well positioned) but when I click on Tab1, the component shows up again in Tab1. When I click on Tab2, the component is not there anymore.....?????

The way I am proceeding is as follows:
All initialisation is done in the init function of a managed bean.
@Override
public void init() {
// Perform initializations inherited from our superclass
super.init();
// Perform application initialization that must complete
// *before* managed components are initialized
// TODO - add your own initialiation code here
System.out.println("_________ENTERING FICHE BEAN INIT________________");
resizerEJB.init();
System.out.println("...........Init components ............");
this.initComponents();
System.out.println("_________EXITING FICHE BEAN INIT________________");

// <editor-fold defaultstate="collapsed" desc="Managed Component Initialization">
// Initialize automatically managed components
// *Note* - this logic should NOT be modified
try {
_init();
} catch (Exception e) {
log("FicheBean Initialization Failure", e);
throw e instanceof FacesException ? (FacesException) e: new FacesException(e);
}

// </editor-fold>
// Perform application initialization that must complete
// *after* managed components are initialized
// TODO - add your own initialization code here
}

and in intComponents, I do:

UIComponent component= ComponentUtil.findComponentInRoot(componentId);
if (component != null){
//Retrieving the parent from heavy client:
String newParentId = componentMap.get(componentId).getParent().toLowerCase();
//Getting the parent in JSF tree structure
String oldParentId =component.getParent().getId();
UIComponent oldParentComponent = component.getParent();
UIComponent newParentComponent = ComponentUtil.findComponentInRoot(newParentId);
if (newParentComponent != null){
oldParentComponent.getChildren().remove(component);
System.out.println("newParentComponent getRendersChildren " + newParentComponent.getRendersChildren());
newParentComponent.getChildren().add(component);
//newParentComponent.encodeChildren(FacesContext.getCurrentInstance());
System.out.println("newParentComponent " + newParentComponent.getChildren());
}
........................

COuld you please help me on this ...Do you have any idea.?

Am I misunderstanding something here ..??

Many thanks in advance
Regards,
Nicolas

Posted by Nicolas Gaudin on January 18, 2008 at 02:40 AM PST #

Hi Ryan, thanks for the nice comment and thanks for all the constructive discussions we had (as you nicely articulated in your blog http://www.ryandelaplante.com/rdelaplante/entry/my_netbeans_visual_web_pack), that lead to these performance improvements.

About your question about facelet support for NB 6.1. The main focus for NB 6.1 seems to be

- Performance improvement.
- Out of the box PhP development support.
- Full fledged Javascript editing support including debugging.

Given these priorities, I'm not sure if facelet could make it in to the product release, but could be an autoupdate module.

Posted by Winston Prakash on January 27, 2008 at 11:20 AM PST #

I'm happy to hear that my comments had some influence on NetBeans. Regarding Facelets, I see that the issue is in the "TBD" section, which probably means that it might or might not make it in. The summary gives me the impression that the existing Facelets plugin needs to get Sun's seal of approval and bundle it with the IDE instead of having to add it from the update center. http://wiki.netbeans.org/NB61PlanTBDs

Posted by Ryan de Laplante on January 27, 2008 at 02:12 PM PST #

Ryan, Most important performance improvement is on Windows. One of the main culprit was BeanInfo search. To find the BeanInfo all jars in the classpath was searched. As you know, I/O on Windows sucks. When we restrict the BeanInfo search to limited jars, our tests shows as much as 50% improvement in opening a page with several components and first time Table Component drop on Windows Vista. We didn't see that much improvement on Unix machines (Linux/Solaris/Mac).

BTW, you won't see drastic improvement with binding removal if your project is small (tens of pages). Actual performance boost is seen with 100 of pages. We tested with 128 page project.

Anyway, for 6.1 we are not changing the overall architecture of Visual Web, which is planned for later major release (may be NB 7.0).

Posted by Winston Prakash on January 30, 2008 at 10:27 AM PST #

[url=http://www.neuroendoscopy.cn/]垂体瘤[/url]
[url=http://www.neuroendoscopy.cn/]脑积水[/url]
[url=http://www.neuroendoscopy.cn/]小儿脑积水[/url]
[url=http://www.neuroendoscopy.cn/]蛛网膜囊肿[/url]
[url=http://www.neuroendoscopy.cn/]脑室内囊肿[/url]
[url=http://www.neuroendoscopy.cn/]透明隔囊肿[/url]
[url=http://www.955680.cn/]传世私服[/url]
[url=http://www.955620.cn/]传世私服[/url]
[url=http://www.955630.cn/]魔女游戏[/url]
[url=http://www.955650.cn/]传奇私服[/url]

Posted by learn chinese on September 17, 2008 at 09:10 PM PDT #