Going Mobile

with your RichFaces Applications

  • Brian Leathem
  • Senior Software Engineer, Red Hat
  • JAX - San Francisco
  • 2012-07-09
  • Welcome to the session "Going Mobile with your RichFaces Applications"

The plan

  • Why the Mobile Web?
  • Why JSF?
  • Why RichFaces?
http://www.flickr.com/photos/jdhancock/5845280258/
  • We're here to talk about writing mobile applications with JSF and RichFaces
  • We'll first establish that the mobile web is an important and unavoidable target in today's market
  • Then we'll demonstrate how JSF & RichFaces can be effective tools in this space.
  • Before we dive in, I'd like to get a feel from you as to how you are positioned today:
    1. Who amongst you are currently using RichFaces for Desktop Applications?
    2. Who is unsure whether they need a mobile solution at all?
    3. Who already has a mobile strategy in place?
    4. Those of you who do already have a mobile strategy in place, how many are doing it with JSF?
  • writing mobile web applications with...
  • establish the mobile web
  • demonstrate JSF/RichFaces can be effective
  • [survey]

Who am I?

  • Senior Software Engineer at Red Hat
  • RichFaces Project Lead
  • Represent Red Hat on the JSF 2.2 EG (JSR-344)
  • Seam Faces module lead
http://leslycorazon.wikispaces.com/file/detail/head-silhouette-with-question-mark.png/319199232
  • I am the RichFaces project lead and I represent Red Hat on the JSF 2.2 EG
  • If you have any question regarding Red Hat and JSF, feel free to corner me at any point throughout the conference
  • me
  • questions

Why Mobile?

I write desktop apps!

http://www.linuk.web.id/2010/09/biografi-singkat-linus-torvalds.html
  • It's not immediately obvious in many market areas that an organisation can benefit from a mobile application
  • There a number of typical, seemingly sound arguments providing a mobile interface:
  • "My Users need a desktop application!"
    • complex interface
    • data entry requirements
    • sit at their desk all day
    • need a mouse
  • However, despite these arguments, we have a game changer
  • not immediately obvious
  • typical sound arguments

Smartphones!

http://en.wikipedia.org/wiki/File:2007Computex_e21Forum-MartinCooper.jpg
  • [Hold up my phone]
  • These little devices in everyones' pocket, with enough computing power packed in them to compete with desktop PC's
  • This has provided your users and customers with the means to access your data anywhere.
    • Not the same kind of interaction as on a desktop app
    • Use cases may be different, but no less valid.
    • For instance, a use case may be an inventory status, rather then data entry
  • This is a photo of Martin Cooper (the inventor of the cell phone) holding the famous motorolla brick phone
  • For me it's a shocking reminder of how far mobile phones have come in such a short time.
  • little devices
  • different interaction

Mobile phones are everywhere!

http://gigaom.com/2010/04/12/mary-meeker-mobile-internet-will-soon-overtake-fixed-internet/
  • Mobile phones are everywhere!
  • with greater than 1 billion mobile users today
  • By 2014 there will be more mobile users than desktop users
  • Granted these aren't all smartphone users, but
    • the trend is such that mobile phones are increasing in complexity, not decreasing
    • further driving the need for mobile apps.

Choose: Native app or Mobile

  • Native:
    • Responsive
    • Touch-specific controls
    • Native L&F
  • Web:
    • Common across platforms
    • HTML 5 / CSS 3 / javascript
  • Those of you who have looked at developing mobile apps know you have a choice to make between Natvie and Web applications
  • Native:
    • has traditionally had the advantage of... [see advantages above]
    • but there are too many platforms to be able to develop for them all
  • Web:
    • Conversely, the mobile web is ubiquitous across platforms
    • but traditionally doesn't offer the native advantages
  • Then there is the hybrid approach which tries to offer the best of both worlds (See the Aerogear JBoss project for more details about hybrid apps)

Choose: Native app or Mobile

So how do I choose?

https://devcentral.f5.com/weblogs/macvittie/archive/2008/09/10/3604.aspx
  • So how do we choose?
  • On the one hand, we have native with apparent runtime advantages, at the cost of an increased development burden
  • On the other hand we have web apps with their ubiquitous run-time, but potential platform limitations
  • It seems we are "damned" either way
  • I love this graphic
    • it was on my father's coffee cup growing up
    • when I was young it was just a funny picture
    • but every year I understood a little more the predicament it was describing
    • now I think of it whenever I'm faced with any unpleasant dilemma
  • but not to worry - I have an answer for this mobile dilemma!

You need a mobile web app regardless!

  • You need a mobile web app regardless!
  • Not everyone can nor will install your native app, even if it is better
  • For consumer-facing applications this is obvious - it's not practical to install an app for *every* store at which one shops
  • I was in a hardware store recently - I wanted to by a garden hose and some related accessories
    • I wanted to price compare with a neighboring store
    • On visiting with website on my mobile phone, I was prompted to install a native application - No way!
    • Fall back to a mobile web-app that more than met my needs
  • Not always the case - some mobile web apps are horrible!
  • Let's just mare sure the ones we write aren't!

Mobile Web is Ready

  • HTML 5
  • CSS 3
  • JavaScript

Html 5 Web App → Native App

  • The good news is today's mobile web is ready and well suited to deliver full featured mobile applications.
  • With HTML 5, CSS 3, and the latest javascript API's
  • we can access many of the phones features and compete toe-to-toe with native apps

target = Webkit

  • We can further simplify our story by focusing on webkit based browsers.
  • This covers the browsers that ship with the Android, iOS, blackberry and Symbian OS'es
  • These also happen to be today's leading mobile platforms, justifying our focus on webkit browsers
  • This focus then allows us to target the subset of HTML 5 and CSS 3 features supported on webkit

JavaScript APIs

  • What javascript API's do we have available?
  • There are many, but some of the most interesting ones for JSF components are listed here
  • You can imagine the kind of JSF Components we can make with these API's
    • a location aware mapping component
    • Data Table pagers that pre-fetch and store the data locally
    • Push applications using web sockets to respond to server-side events - (in fact we have this one already!)
    • Canvas driven components drawing arbitrary graphics (Dials, sliders, etc)
  • The possibliities are seemingly litmitless
  • Let's look at some specific things we can achieve leveraging the HTML 5, CSS 3, and javascript APIs

Hide the Location bar

 
    hideURLBar: function() {
        //hide the url bar on mobile devices
        setTimeout(scrollTo, 100, 0, 1)
    }
     
  • Starting with a simple but effective example: hiding the location bar
  • Here's a screen-shot of a mobile web-app
  • Notice there is no location bar
  • It doesn't "look" like a web browser
  • Achieving this is simply a matter calling the DOM method "scrollTo", here invoked after a timeout to allow the page to load

Orientation Detection

  • Another "must have" for mobile apps is orientation detection.
  • When we have more screen real estate available, we should use it!
  • In this trivial example, we conditionally display table columns depending on the orientation of the device.
  • In the portrait mode, we display the id and name columns, and add the email and phone number in the landscape view

Orientation Detection

 
    /* portrait screen width */
    @media screen and (max-width: 480px) {
       .portrait {
       }
       .landscape {
          display:none;
       }
    }
    
    /* landscape/desktop device widths */
    @media screen and (min-width: 481px) {
       .portrait {
          display:none;
       }
       .landscape {
       }
    }
     
  • We achieve this with some simple CSS leveraging the max-width CSS 3 property
  • When the with of the display is less than the value specified (480px) we hide any element wit the CSS class "landscape"
  • We've also created the converse "portrait" css class, Although it tends to be less useful in practice.

Hardware acceleration

Activate the GPU for compositing each page

 
    .page {
      ...
      -webkit-transform: translate3d(0, 0, 0);
    }
     
  • Another important feature available with CSS 3 is the transform translate3d css property
  • We can achieve a performance boost for our mobile web apps wit this simple CSS rule
  • When this rule is present, the browser will activate the GPU for rendering visual effects
  • In fact the presentation you are looking at now is written in HTML 5, and uses the same CSS property to power the page transitions
  • [Press ESC to zoom out]
  • Again, we see seemingly limitless possibilities of what we can achieve with HTML 5.

HTML 5 presentation

 
    <section>
      <h2>The plan</h2>
      <ul>
        <li>Why the Mobile Web?</li>
        <li>Why JSF?</li>
        <li>Why RichFaces?</li>
      </ul>
    </section>
     
  • To drive the point home, here is the markup used to create one of the introductory slides of this presentation
  • As you can see, it consists of nothing more than the HTML 5 "section" tag, and a conventional HTML unordered list
  • The rest of the style and behaviour comes from CSS 3 and javascript

Plethora of Frameworks

+ JSF !!

  • There are a number of frameworks out there dedicated to simplifying this space and enabling developers to more easily and effectively write web applications.
  • Perhaps you've already heard of some of these - all great frameworks in their own right
  • I'm here today to convince you that JSF is a worthy candidate to be added to this list

Why JSF?

What do we like about JSF?

  • What does JSF have going for it that we should to consider it for mobile use cases?

JSF Because...

Standardized as part of Java EE

  • The biggest plus for me with JSF is that it is a standard
  • With multiple OSS implementations
  • Healthy component vendor marketplace - again OSS
  • No vendor lock-in!

JSF Because...

Ajax is baked-in
to the framework

  • AJAX plays a huge roll in developing mobile web apps.
  • With JSF 2.0, AJAX is front & center in the platform, ready for us to tap into to deliver fast and responsive applications

JSF Because...

Component-oriented
programming model

  • Similarly we will rely heavily on javascript as we develop mobile web applications.
  • But JSF components abstract away the javascript
  • We write to a standardized component model, rather than to arbitrary javascript libraies
  • As I'm fond of saying on behalf of the RichFaces project: "We write the javascript so you don't have to"

JSF Because...

Facelets:
A powerful templating mechanism

  • Facelets are one of the crown jewels of the JSF eco-system
  • We have a highly expressive and compositable templating language
    • includes
    • composite component
    • fragments and compositions
  • Keep our view code DRY

JSF Because...

Build on our investment in JSF/RichFaces technology!

http://openclipart.org/detail/17302/groupe-by-czara1 http://openclipart.org/detail/36571/tango-network-workgroup-by-warszawianka http://openclipart.org/detail/48427/clock-by-palomaironique http://openclipart.org/detail/22860/money-bag-icon-by-chovynz
  • Finally, let's use JSF to leverage our existing investment of people, infrastructure, time and money in JSF and RichFaces technologies
  • Why should we start with something altogether new if we don't have to?

Adapting JSF to be “Mobile Ready”

  • Responsive
    • Single Page programming model (via JSF AJAX)

  • Touch-specific controls
    • CSS 3 styling

  • Mobile L&F
    • Component libraries
  • JSF is great, but it needs some help to be made "mobile ready"
  • To create responsive applications, we want to make heavy use of AJAX and avoid full page loads.
  • To achieve thos, we adopt a single page programming model, which unfortunately is not well supported by the JSF navigation system - at least not yet ;)
  • JSF Components need to be made Touch "compatible".
  • For instance as a rule-of-thumb the touchable area should be 1cm x 1cm.
  • Fortunately achieving this is simply a matter of providing the right CSS skin
  • Finally we want to achieve a mobile L&F.
  • The responsibility here falls to us, the component providers
  • We'll take a close look at the RichFaces mobile components shortly, but first let's see what we can do with "vanilla" JSF

Device Detection

http://www.flickr.com/photos/ivyfield/4486938457/
  • One key requirement is to have the ability to tailor the content to the viewing device
  • We've seen earlier that we can use CSS techniques to tailor the display based on the screen size
  • But this is a client-side technique and works after the content is delivered to the browser
  • If we want to seamlessly provide both mobile and desktop applications, we need to know sooner:
  • on the server before the content is rendered and delivered

Device Detection

User Agent String

 
    Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/536.5
                (KHTML, like Gecko) Chrome/19.0.1084.56 Safari/536.5
     
  • Fortunately this is possible by inspecting the "user-agent" string which is sent from the browser with the initial request.
  • Here we can see a typical user agent, and can infer some significant details from it. For instance:
    • The browser claims to be Mozilla compliant (a useless claim nowadays)
    • The OS is 64-bit linux
    • The browser is a webkit browser
  • What's important here is that it's obvious from this user-agent string that this is a desktop browser, and not a mobile one.
  • The server knows this from the time of the initial request and we can react accordingly

Device Detection

Server Side: UAgentInfo

 
    public class UAgentInfo implements Serializable {
       ...
       public void initDeviceScan() {
          this.isIphone = detectIphoneOrIpod();
          this.isAndroidPhone = detectAndroidPhone();
          this.isTierTablet = detectTierTablet();
          this.isTierIphone = detectTierIphone();
          this.isTierRichCss = detectTierRichCss();
          this.isTierGenericMobile = detectTierOtherPhones();
       }
      ...
     
  • Fortunately we don't have to parse the user-agent ourselves.
  • We can leverage libraries to do this for us, like the UAgentInfo library shown here.
  • We can use this library to determine whether the request originated from:
  • [ go thorugh the API above ]

Device Detection

JSF Bean: UserAgent

 
    @Model
    public class UserAgent {
    
       private UAgentInfo uAgentInfo;
    
       @PostConstruct
       public void init() {
          FacesContext context = FacesContext.getCurrentInstance();
          HttpServletRequest request = (HttpServletRequest)
                      context.getExternalContext().getRequest();
          String userAgentStr = request.getHeader("user-agent");
          String httpAccept = request.getHeader("Accept");
          uAgentInfo = new UAgentInfo(userAgentStr, httpAccept);
       }
     
  • Using this library from JSF is then remarkably easy.
  • Here we have a CDI bean that instantiates an instance of UAgentInfo and feeds it the user-agent string from the FacesContext
  • We can then access the UAGentInfo's methods from our JSF beans, or from expression language to tailor our response accordingly

Device Detection

URL Rewriting: ReWrite

 
    public Configuration getConfiguration(final ServletContext context) {
       return ConfigurationBuilder.begin()
          .defineRule()
          .when(Direction.isInbound()
          .and(Path.matches("/").or(Path.matches("/index.jsf")))
          .and(UserAgent.isMobile())
          ).perform(Forward.to("/mobile/"));
       }
     
  • In fact, in our sample applications we take this one step further and make use of URL rewriting library called "ReWrite"
  • Rewrite has UAgentInfo built in, and we can define a simple rule to redirect mobile requests to a mobile specific page
  • The rule here can be interpreted as:
  • "If the requested page is the index page,and the request comes from a mbile browser, then redirect the request to a mobile specific page"
  • Rewrite works at the servlet level, before the request even gets to JSF.
  • This makes it transparent from a JSF perspective.

The RichFaces value add

For mobile JSF apps

  • By now we have a good idea of why we want to use JSF in our mobile web applications
  • But why use RichFaces?
  • What's the RichFaces value-add?

The RichFaces Showcase

  • The most obvious answer is the UI components
  • With RichFaces 4.1 and 4.2 we are shipping "mobile compatible" RichFaces components
  • These are our existing components adapted to mobile environments through the inclusion of some mobile specific CSS
  • Let's take a quick look at our showcase on a mobile browser to see some of these components in action
  • [ demo the RichFaces mobile showcase ]

a4j:push

Facelet Source

 
    <a4j:push address="pushCdi"
      ondataavailable="jQuery('<li/>').text(event.rf.data)
                                      .prependTo('#messages')" />
    <ul id="messages" />
     

Java Source

 
    @Inject
    @Push(topic = "pushCdi")
    Event<String> pushEvent;
    
    public void sendMessage() {
      pushEvent.fire("My Message");
    }
     
  • In addition to the UI components, RichFaces also offers a number of non-visual components
  • In particular the a4j:push component is particularly valuable in mobile applications
  • where we can have the mobile clients respond to server-side events
  • Here we see some sample code demonstrating how simple the push component is to use
  • On our facelet page, we place an a4j:push component addressing a particular topic
  • then in our backing bean we use CDI to inject a pushEvent for the same topic
  • finally we fire a message from the pushEven in our backing bean, and respond to that event on the client-side
  • here we respond to the event with a javascript event listener, but we could equally-well trigger an ajax update
  • I should also point out that CDI is not required, we also provide a non-CDI APi for working with the push component

a4j:push in tweetstream

JBW 2011 Keynote

  • Perhaps you remember the tweetstream demo from last year's JBoss world keynote?
  • the tweetstream demo is a terrific example of using the RichFaces push component in a mobile JSF application
  • we'll see the demo component in action shortly, when we get to a demo of our case-study sample application

Client-side validation

  • Perform validation locally
  • Another good example of the RichFaces value-add is our support for client-side bean validation
  • We provide a javascript implementation of the Java EE bean validation spec, and validate input values before they are sent to the server
  • This avoids a round-trip to the server, and improves the responsiveness of your mobile applications
  • With mobile web apps, the more we can do locally, the better off we'll be

a4j:jsFuntion

A facelet component interface to the JSF AJAX API

 
    <a4j:jsFunction name="updateName" render="showname">
       <a4j:param name="name" assignTo="#{functionBean.text}" />
     </a4j:jsFunction>
     
  • Lastly, I'd like to call out the a4j:jsFunction component.
  • Working with mobile web-apps can involve writing a lot of javascript
  • The a4j:jsFunction provides a powerful mechanism to interfacing javascript in the client with the JSF backend through the JSF Ajax API
  • We'll see some sample code making use of this component also in our sample application

Case Study: “kitchensink” Quickstart

1

2

3

4

  • Let's take a look at a more concrete "real-world" example: here we see the JBoss-as kitchensink sample project.
  • It's a CRUD application that uses many Java EE APIS's and there corresponding JBoss implementations.
  • We'll take a look at how we "mobilized" this application
    • First using vanilla JSF with custom CSS & Javascript
    • then we'll introduce RichFaces to simplify the app
    • lastly we'll look at some experimental components that further simplify things
  • Let's start with breaking the application down into logically to identify the mobile views we'll require
  • [ click through the reveal.js fragments ]

kitchensink: “Mobilized”

  • Demo time!
  • Here's a screenshot of the "mobilized" application.
  • You can see here the view is single purpose: this is the home page of the mobile application, with nice big clickable menu buttons (even with my fat fingers! )
  • The custom CSS and javascript resources required to make this work are packaged together into our "slidfast" library
  • Let's spend a few minutes running the application to get a feel for what it's all about
  • [ demo the app ]

Manipulate the location hash

Directly, using html links

 
    <a href="#about">i</a>
     

Using javascript

 
    <rich:panelMenuItem label="Add Member"
                        onclick="location.hash='new'"
                        mode="client" />
     
  • We can set the location hash directly using an html tag
  • Or we can use a javascript listener on a component to set it
  • either way, once the location hash is changed, slidfast will initiate the page transition
  • A big advantage of using th location is hash is it supports backwards navigation - through our custom back button, or via the browser controls
  • There is (at least!) one significant problem with the previous approach
  • Can anyone guess what it is?
  • The approach won't scale - we've front-loaded all our mobile pages
  • We need an AJAX driven approach that loads pages as required, if we want to scale

Mobile compatible with slidfast

A collection of lightweight CSS and Javascript assets

  • the key thing to keep in mind with slidfast is that it's designed to be light-weight
  • Slidfast is just what's required to write your mobile app the way you want, without requiring one to adopt a whole framework
  • This is in contrast with many of the other "heavy-weight" mobile frameworks we mentioned earlier
  • Slidfast is however AJAX centric, and works best when we write our applications with a single page programming model

“Vanilla” JSF mobile facelet

 
    <?xml version="1.0" encoding="UTF-8"?>
    <ui:composition template="/WEB-INF/templates/mobile.xhtml"
                    xmlns:ui="http://java.sun.com/jsf/facelets"
                    xmlns:f="http://java.sun.com/jsf/core"
                    xmlns:h="http://java.sun.com/jsf/html">
      <ui:define name="content">
        <h:outputStylesheet name="mobile.css" library="css"/>
        <h:outputScript>
          //initialize mobile RichFaces with slidfast
          slidfast({
            defaultPageID:'home-page',
            callback: 'handleHashChange',
            backButtonID: 'back-button'
          });
        </h:outputScript>
        <div id="browser">
          <header>
            <div>
              <div id="back-button" class="hide-button">Home</div>
              <div id="info-button" class="right-header-button info-link">
                <a href="#about">i</a>
              </div>
              <p>Mobile JSF</p>
            </div>
          </header>
          <h:form id="mobileForm">
            <div id="page-container">
              <div id="home-page" class="page stage-left">
                <ui:include src="/mobile/home.xhtml"/>
              </div>
              <div id="new" class="page stage-right">
                <ui:include src="/mobile/new.xhtml"/>
              </div>
              <div id="list" class="page stage-right">
                <ui:include src="/mobile/list.xhtml"/>
              </div>
              <div id="about" class="page stage-right">
                <ui:include src="/mobile/about.xhtml"/>
              </div>
            </div>
          </h:form>
        </div>
      </ui:define>
    </ui:composition>
     
  • Let's look closer at what the mobile facelet page looks like (Don't worry about the font-size, we'll zoom in)
  • This is for a vanilla JSF application + our slidfast CSS and javascript resources (we'll look at RichFaces enhancements to this shortly)
    1. We are looking at a facelet composite page that uses a specific mobile template (HTML 5 doctype, slidfast resources)
    2. Here we initialize the slidfast library, defining:
      • the initial/default page id,
      • the name of a callback method we'll see shortly
      • the id of the button we'll use for backward navigation
    3. the html 5 header tag is used to set the title bar across the top of the page
      • here's where we put the back-button with the id we configured above
    4. notice we navigate using location hashes - we'll see below how slidfast uses the location hash to slide content in and out of the view
    5. here we see a series of div's with css class "page" - this is our single page programming model
  • this is the juicy part, let's bring this up on a page on it's own

Single-page programming model

 
    <h:form id="mobileForm">
       <div id="page-container">
            <div id="home-page" class="page stage-left">
                <ui:include src="/mobile/home.xhtml"/>
            </div>
           <div id="new" class="page stage-right">
               <ui:include src="/mobile/new.xhtml"/>
           </div>
           <div id="list" class="page stage-right">
               <ui:include src="/mobile/list.xhtml"/>
           </div>
           <div id="about" class="page stage-right">
               <ui:include src="/mobile/about.xhtml"/>
           </div>
        </div>
    </h:form>
     
  • We can see a number of div elements with the page CSS class, each with an additional stage-* class
  • These are the page elements that slidfast slides in and out
  • The page ID activated is the one who's id is set in the location hash
  • so if I want to activate the "about" page, I have to set the location hash to "about"
  • Let's take a look at how we can manipulate the location hash to navigate through our pages

Scale w/ dynamic ui:include

 
    <div id="page-container">
        <div id="home-page" class="page stage-left">
            <ui:include src="/mobile/home.xhtml"/>
        </div>
        <div id="app-page" class="page stage-right">
            <ui:include src="#{pageBean.page}"/>
        </div>
    </div>
     
  • We can do this by making use of a ui:include tag with a dynamic src.
  • The trick here is to trigger an ajax update when the location hash changes setting the pageBean property
  • We'll first look at doing this with Vanilla JSF, leveraging the jsf.ajax javascript API

Update the PageBean - JSF

 
    <h:inputHidden id="page" value="#{pageBean.location}" />
    <h:outputScript >
      var handleHashChange = function (page) {
         var hiddenInput = 'mobileForm:page';
         $(document.getElementById(hiddenInput)).val(page);
    
         var options = {
            execute: hiddenInput,
            render : '@form',
            onevent : function(event) {
               if(event.status == 'success') {
                  slidfast.core.slideTo('app-page');
               }
            }
         }
         jsf.ajax.request(hiddenInput, null, options);
      }
    </h:outputScript>
     
  • Here we see the javascript necessary to invoke an ajax update when we change the location hash
    1. First we need somewhere to put the location hash so it can be submitted by the ajax request - we do this with a inputHidden element
    2. Here we define the "handleHashChange" listener we registered with slidfast during initialisation - this method will be called whenever slidfast detects a page change
    3. In that listener, we first grab a reference to our hidden input, and update it with the location hash value
    4. Next we create an options object we will pass along with the jsf ajax request
      • "execute" and "render are analogous to the same attributes in an f:ajax call
      • with the onevent callback we trigger the slidfast page transition when the ajax call is successful
    5. Finally we invoke the jsf.ajax method to trigger the ajax call and associated page updates
  • The value of the location hash gets stored in the "location" property of the pageBean, and is used to specify what page to include in the dynamic ui:include

Update the PageBean - RichFaces 4.2

 
    <h:outputScript target="head">
    <a4j:jsFunction name="handleHashChange"
                    render="@form"
                    oncomplete="slidfast.core.slideTo('app-page')">
        <a4j:param name="page" assignTo="#{pageBean.location}"/>
    </a4j:jsFunction>
     
  • We can simplify this by using the RichFaces a4j:jsFunction component
  • Notice how we no longer need the hidden input, as the a4j:param component let's us update backing beans directly
  • We still have the oncomplete callback that invokes the slidfast page transition
  • This is what we "ship today", if you are writing mobile apps with the current RichFaces release, this is what we provide
  • Let's take a sneak peak at some work-in-progress in our Sandbox

Update the PageBean - RichFaces Sandbox

 
    <slidfast:pageChanger activePage="#{pageBean.location}" render="app-page" />
     
 
    <slidfast:page id="app-page" stage="right">
        <ui:include src="#{pageBean.page}"/>
    </slidfast:page>
     
  • (The Sandbox is the RichFaces SkunkWorks, where we try out our new ideas and components before landing them into our component set)
  • here we have a pageChanger that encapsulates the above javascript, and also provides some additional added value
  • by allowing other components to take over the ajax rendering, in which case the pageChanger component only invokes the slidfast transition

“SlidFast” JSF mobile facelet

 
    <?xml version="1.0" encoding="UTF-8"?>
    <ui:composition template="/WEB-INF/templates/mobile.xhtml"
                    xmlns:ui="http://java.sun.com/jsf/facelets"
                    xmlns:h="http://java.sun.com/jsf/html"
                    xmlns:slidfast="http://richfaces.org/slidfast">
        <ui:define name="content">
            <h:outputStylesheet name="app.css" library="css"/>
            <h:outputStylesheet name="mobile.css" library="css"/>
            <div id="browser">
                <header>
                    <div>
                        <div id="back-button" class="hide-button">Home</div>
                        <div id="info-button" class="right-header-button info-link">
                            <a href="#about">i</a>
                        </div>
                        <p>Richfaces Mobile</p>
                    </div>
                </header>
                <h:form id="mobileForm">
                    <slidfast:pageChanger id="pageChanger" activePage="#{pageBean.location}" render="app-page" />
                    <slidfast:pageContainer id="pageContainer">
                        <slidfast:page id="home-page" stage="left">
                            <ui:include src="/mobile/home.xhtml"/>
                        </slidfast:page>
                        <slidfast:page id="app-page" stage="right">
                            <ui:include src="#{pageBean.page}"/>
                        </slidfast:page>
                    </slidfast:pageContainer>
                </h:form>
            </div>
           <h:outputStylesheet library="org.richfaces" name="rf-mobile-skin.css" />
        </ui:define>
    </ui:composition>
     
  • Looking at the facelet again as a whole, we can see how the introduction of the slidfast components greatly simplifies the page
  • The pageChanger component has replaced many lines of javascript
  • The pageContainer and page components have replaced the generic div elements. This offers a number of advantages
    • The page component is a valid target for jsf ajax updates (so we don't need to wrap everything in an h:outputPanel)
    • The use of CSS classes for imparting behaviour has been replaced with more meaningful attributes, like the "stage" attribute
  • The result is a page that looks and codes more like JSF, and less like a HTML 5/javascript app

Responsive layout

  • I'd like to wrap up this presentation with a look at another sandbox effort, where we are creating JSF components out of the twitter bootstrap CSS resources
  • If you haven't yet heard of the RichFaces bootstrap project, be sure to check it out.
  • Bootstrap provides a set of cross browser CSS and javascript resources you can use to build modern web applications
  • I could do a whole presentation on bootstrap components and how they fit with JSF
  • But the key point I want to raise here is the "Responsive Layout". Components adjust and resize themselves as the display size changes
  • let's take a look at a demo of this
  • [demo the bootstrap responsive layout]
  • Hopefully you can see hoe building RichFaces applications with these components will lend itself nicely to mobile development

The end?

The beginning - of your mobile adventure with RichFaces!

  • We've seen how mobile web-apps are a necessity in today's market
  • and how JSF and RichFaces can be effective means of providing these web apps, particularly if we are already invested in the technology
  • While RichFaces isn't required for building mobile JSF applications,we can see how the process is simplified by the introduction of javascript encapsulating mobile-specific components
  • Finally we had a peek into the future, to see how RichFaces is going to further empwer us to build mobile web apps.

Credits

Additional Resources

Source code:

Articles:

Stay in the loop