I recently read this article from Neil Griffin Making Distinctions Between Different Kinds of JSF Managed-Beans and it got me thinking about the distinction between different beans in my own application. To quickly summarise the gist:
Model Managed-Bean: This type of managed-bean participates in the
"Model" concern of the MVC design pattern. When you see the word
"model" -- think DATA. A JSF model-bean should be a POJO that follows
the JavaBean design pattern with getters/setters encapsulating
properties.
Backing Managed-Bean: This type of managed-bean participates in the
"View" concern of the MVC design pattern. The purpose of a
backing-bean is to support UI logic, and has a 1::1 relationship with
a JSF view, or a JSF form in a Facelet composition. Although it
typically has JavaBean-style properties with associated
getters/setters, these are properties of the View -- not of the
underlying application data model. JSF backing-beans may also have JSF
actionListener and valueChangeListener methods.
Controller Managed-Bean: This type of managed-bean participates in
the "Controller" concern of the MVC design pattern. The purpose of a
controller bean is to execute some kind of business logic and return a
navigation outcome to the JSF navigation-handler. JSF controller-beans
typically have JSF action methods (and not actionListener methods).
Support Managed-Bean: This type of bean "supports" one or more views
in the "View" concern of the MVC design pattern. The typical use case
is supplying an ArrayList to JSF h:selectOneMenu drop-down
lists that appear in more than one JSF view. If the data in the
dropdown lists is particular to the user, then the bean would be kept
in session scope.
Utility Managed-Bean: This type of bean provides some type of
"utility" function to one or more JSF views. A good example of this
might be a FileUpload bean that can be reused in multiple web
applications.
This made sense to me and for the past few hours I have been refactoring my code and came up with the following with respect to the user login:
The AuthenticationController is an example of a Controller Managed-Bean. It is request-scoped and features two getters and setters for setting a username and password, and two navigation methods, authenticate and logout, navigating the user to either their private area upon successful login, or back to the main page when logging out.
The UserBean is an example of a Support Managed-Bean. It is session-scoped and features an instance of User class (which would be null when you are not authenticated) with a getter and setter, nothing more.
The AuthenticationController has this user as a managed property (#ManagedProperty(value = "#{userController.user} private User user;). Upon successful authentication, the AuthenticationController would set the managed property to the actual user instance with the corresponding username that was used for the login.
Any new beans would be able to grab the user as a managed property as well and pull the data they need, such as group membership for instance, if the User class would feature a list with group names.
Would this way be the proper way to go about with regard to the seperation of concerns?
This is a very subjective question. I personally disagree that article and find that it's giving really bad advice to starters.
Model Managed-Bean: This type of managed-bean participates in the "Model" concern of the MVC design pattern. When you see the word "model" -- think DATA. A JSF model-bean should be a POJO that follows the JavaBean design pattern with getters/setters encapsulating properties.
I would absolutely not make or call it a managed bean. Just make it a property of a #ManagedBean. For example a DTO or JPA #Entity.
Backing Managed-Bean: This type of managed-bean participates in the "View" concern of the MVC design pattern. The purpose of a backing-bean is to support UI logic, and has a 1::1 relationship with a JSF view, or a JSF form in a Facelet composition. Although it typically has JavaBean-style properties with associated getters/setters, these are properties of the View -- not of the underlying application data model. JSF backing-beans may also have JSF actionListener and valueChangeListener methods.
This way you keep duplicating and mapping the properties of the entity in the managed bean. This makes no sense to me. As said, just make the entity a property of the managed bean and let the input fields refer it directly like #{authenticator.user.name} instead of #{authenticator.username}.
Controller Managed-Bean: This type of managed-bean participates in the "Controller" concern of the MVC design pattern. The purpose of a controller bean is to execute some kind of business logic and return a navigation outcome to the JSF navigation-handler. JSF controller-beans typically have JSF action methods (and not actionListener methods).
This describes the #RequestScoped/#ViewScoped #ManagedBean class pretty much. Whether event listener methods are allowed or not depends on whether they are specific to the view which is tied to the bean and/or are for their job dependent on the bean's state. If they are, then they belongs in the bean. If not, then they should be a standalone implementation of any FacesListener interface, but definitely not a managed bean.
Support Managed-Bean: This type of bean "supports" one or more views in the "View" concern of the MVC design pattern. The typical use case is supplying an ArrayList to JSF h:selectOneMenu drop-down lists that appear in more than one JSF view. If the data in the dropdown lists is particular to the user, then the bean would be kept in session scope.
Fine. For application wide data like dropdown lists just use an #ApplicationScoped bean and for session wide data like logged-in user and its preferences just use a #SessionScoped one.
Utility Managed-Bean: This type of bean provides some type of "utility" function to one or more JSF views. A good example of this might be a FileUpload bean that can be reused in multiple web applications.
This makes not really sense to me. Backing beans are usually tied to single views. This sounds too much like an ActionListener implementation which is to be used by <f:actionListener> in command components to your choice. Definitely not a managed bean.
For kickoff examples of the right approach, see also:
Hello World example in Our JSF wiki page
"Bookstore CRUD" example in this answer
"Master-detail" example in this answer
JSF Service Layer
Communication in JSF 2
Related
We're making usage of Primefaces Wizard component and the forms been included within the Wizard's TABs were declared inside some UI Fragments with their respective Managed Beans. Each one of the Managed Beans contains properties representing some Entities.
What we need to achieve is that, for every completed step, we take the above mentioned Entity property that was previously fulfilled and designate the same Entity property inside the Wizard Managed Bean, and later we may present the Entities data like a summary, so the user would decide whether to proceed or not.
I'm wondering whether there are some other options to accomplish that rather than set a Session variable across the Wizard navigation.
Is the JSF #ManagedProperty usage feasible?
Tks.
As described in this question I try to perform some field validation in a form on the backing bean side. For this I would like to access the violating fields to mark them.
From searching the web there seem to be two ways to do this:
store the components in the backing bean for access and use them in the JSF pages via the binding attribute.
Use standard value binding in the JSF pages and when needing access to a component from the bean, look it up via UIViewRoot.findComponent(String id)
As far as I can see both ways have drawbacks:
Component bindings blows up the backing bean with variables and getters/setters, some sites strongly discourage the use of component binding at all. In any case, a request scope is advised. On the other hand, findComponent() always traverses the tree, which may or may not be costly, right? (Plus, at the moment I can't find my component at all, but that is another problem)
Which would be the way to go? Are these interchangeable alternatives and if not, based on what criteria do you chose? Currently I just don't have enough insight to make a decent decision...
First of all, regardless of the choice, both are a poor practice. See also How does the 'binding' attribute work in JSF? When and how should it be used?
If you had to make the choice, component bindings are definitely faster and cheaper. It makes logically completely sense that a tree scan as done by UIComponent#findComponent() has its performance implications.
Indeed, the backing bean holding the component bindings must be request scoped, but you could easily inject a different scoped backing bean holding the business logic in it by #ManagedProperty.
A cleaner approach would be to use a Map as holder of all component bindings. You only need to add the following entry to faces-config.xml:
<managed-bean>
<managed-bean-name>components</managed-bean-name>
<managed-bean-class>java.util.HashMap</managed-bean-class>
<managed-bean-scope>request</managed-bean-scope>
</managed-bean>
This can just be used as
<h:inputSome binding="#{components.input1}" />
<h:inputSome binding="#{components.input2}" />
<h:inputSome binding="#{components.input3}" />
And this can be obtained in other beans as
Map<String, UIComponent> components = (Map<String, UIComponent>) externalContext.getRequestMap().get("components");
This way you don't need to worry about specifying individual properties/getters/setters. In the above example, the Map will contain three entries with keys input1, input2 and input3, each with the respective UIComponent instance as value.
Unrelated to the concrete question, there may be a much simpler solution to the concrete problem as you described in the other question than performing the validation in the action method (which is actually Bad Design). I've posted an answer over there.
I can't find any guidance on this question. I am writing a composite component that needs its own backing bean because it interacts with a data base.
The new component also needs to be able to set a value in some other backing bean as the result of some user action.
To do this, the question is do I have to write a #FacesComponent java class or a regular #Model/#Named (I use CDI annotations) type of bean? If you can use either, what is the advantage of one or the other?
Secondary question: will I be able to use CDI #Inject into a #FacesComponent to get my DAOs and such?
Update: I discovered that I can access cc.attr objects with the following code in a regular backing bean:
FacesContext fc = FacesContext.getCurrentInstance();
Object obj = fc.getApplication().evaluateExpressionGet(fc,
"#{cc.attrs.model.location}", Location.class);
So this allows me to obtain attributes. I haven't found out how I can write them yet.
So it seems that the only real reason to do a #FacesComponent is if you want to write rendering code that will output something the normal Facelets tags won't render. Is this correct?
I think BalusC responded to this basic question in this thread.
The main advantage is the ability of a #FacesComponent to access attributes that a UIComponent normally has access to, rather than trying to tie in with EL expressions executed in the bean.
Im currently using JSF2, and i notice the JSF bean could have a lot of responsibility, and if combined will look like lots of codes. These include :
holding the state / data
could be a backing bean for the UI component
action methods definition
action listener methods definition
navigation
calling the services
all the setter n getters
Does it make anysense to break these into several classes or do you usually combine all of them together ?
Currenly for every JSF Bean, i define another class to hold the view data / state along with the setter getters.
How do you usually do it ? Please share your experience !
Thank you =)
Every property which is been used in action(listener) methods needs to stay in the backing bean. The remnant most likely belongs in its own class which can in turn be a different (managed/entity)bean, eventually as a (managed)property of the bean where it originated.
Hey guys. I'm designing a site that uses the Icefaces framework. I've been reading a book called Real World Java EE patterns. I'm kind of confused how to layout the pages. Normally I would have just a POJO class implement serializable for a bean. This bean would then back each page. With a single page design I'm going to have a bunch of elements on the page. Datatables, trees, inputs, calendars etc. Is it normal or best practice to have separate beans for each datatable, calendar, etc or put that all in one bean? I'm not sure how to approach this. Right now each element is a bean and I'm using the #Inject annotation to have the data table talk to the tree and vise versa. This creates really bad code and if I put this as a member of the class I will get a circular reference because the data table bean has to inject the calendar and the calendar has to inject the data table.
Thanks for any help.
I tend to use a single bean per <h:form> or at least per view (XHTML/JSP file). Any related beans will just be injected in this particular "main" bean (and thus not among each other).