| Close Window |
A typical Java developer knows that when you need to develop a GUI for a Java application, Swing is the tool. Eclipse SWT also has a number of followers, but the majority of people use Java Swing. For the past 10 years, it was a given that Swing development wouldn't be easy; you have to master working with the event-dispatch thread, GridBaglayout, and the like. Recently, the NetBeans team created a nice GUI designer called Matisse, which was also ported to MyEclipse. Prior to Matisse, JBuilder had the best Swing designer, but it was too expensive. Now a good designer comes with NetBeans for free.
Why even consider Flex for developing Rich Internet Applications (RIA)? First, we'll give the short answer. Just look at the code in Listing 1. This code compiles and runs in the Flash player and produces the output shown in Figure 1. Yes, it's a tree control with several nodes that know how to expand, collapse, and highlight the user's selection. Imagine the amount of Java code you'd need to write to achieve the same functionality.
The code is nice and clean; the GUI looks rich and appealing. The Flex compiler automatically converts the MXML code in Listing 1 into an object-oriented language called ActionScript, and then compiles it into an SWF file, the format that the Flash player understands.
To add the business processing to this example, we'd need to write event handlers in the ActionScript 3.0 language, which is very similar to Java, but we wouldn't have to worry about routing all events to the event-dispatch queue. Below are some of the other reasons that the Flex/Flash combination is a very promising technology for RIA development:
Flex client applications are compiled SWF files that can be delivered to the client and run by the Flash player, which is installed as a plug-in to your browser. On the client side, Flex consists of a Flash player, a framework of predefined components, a couple of command-line compilers, and an Eclipse-based Flex Builder IDE. On the server side, Flex is a Web application that includes Flex Data Services and a Flex Charting component and can be deployed in any JEE server.
To see Flex in action, let's write the functional specification, then develop and deploy a sample application.
Designing the Stock Portfolio Application
We'll create a Web application that will receive and display a feed containing security prices and the latest news as in Figure 2 and 3.
The top portion of the screen must be populated by the stocks included in the user's portfolio. For simplicity, store the user's portfolio in the XML file as in Listing 2.
When the user clicks on a row with a particular stock (i.e., ADBE as in Figure 2), populate the lower data grid with the headlines related to the selected stock. The news should be coming from http://finance.yahoo.com/rss/headline. The column link has to contain the URLs of the news, and when the user clicks on the link, a new browser's window should pop up displaying the selected news article.
The top of the screen should contain the toggle buttons Show Grid/Show Chart. When the Show Grid option is selected, the user will see the screen as in Figure 2, and when Show Chart is selected, the top data grid has to be replaced with the pie chart (see Figure 3), preserving the same functionality (clicking on the pie slice repopulates the news headlines).The market data has to be refreshed on the screen every second or so.
Since we are limited by the size of this article, for the server side we'll just write a simple POJO that will generate random numbers and push them back to the Flash client. However, in the next article we'll add the data feed from an external Java application via JMS.
Sounds like an ambitious task for a short magazine article, doesn't it? This article is not a tutorial on Flex MXML or ActionScript, but as a Java developer, you should be able to understand how the code is put together with minimal explanations. After deploying and running this application you can study Flex more formally. So let's roll up our sleeves...
Installing Flex Builder and Flex Data Services
At the time of this writing, Flex 2 is still in beta (http://labs.macromedia.com/flexproductline/ ) and its production release is expected this summer. The burning question is, "Is it free?" The answer is, "It depends." There is a free lunch (or rather a complimentary appetizer from the chef), but it'll just whet your appetite, and pretty soon your hand will start slowly reaching for the wallet. Flex Data Services (FDS) that provide seamless integration of Flex into enterprise data and messaging have deployment fees. For high-performance enterprise applications supporting client authentication, messaging, and the like, you may want to purchase the FDS license. For smaller applications, you might use integration via HTTPService, Web Services, and Remote Java invocation using standard and open source technologies.
The free part is Flex command-line compilers and the Flex framework that includes all these libraries of rich components. This means you can use any plain text editor for writing Flex code, compile it into the SWF file, and run it in a free Flash 9 player. As of today though, Adobe says that the Eclipse-based Flex Builder IDE, which makes development a lot more productive, will not be free. Flex Charting components that offer client-side interactive charts are not free either.
The Flex Builder IDE comes in two flavors: a standalone version and the Eclipse plug-in. You'll have to choose one of these versions during the installation process. Let's install Eclipse 3.1 first from www.eclipse.org/downloads/ (unless you have it already), and then the plug-in version of Flex Builder. During the installation, you'll have to specify the directory where your Eclipse resides (the one that has a subdirectory called plug-ins). After Flex Builder is installed, start the Eclipse IDE, select the menus Window | Open Perspective | Other and add the Flex Development perspective (the Flex Debugging perspective will be added automatically to the Eclipse IDE). The Flex Builder comes with great Help content, which in the plug-in version is hidden under the menu Help | Help Contents | Adobe Help.
To run client-only Flex applications, we need to install one more program: Flash Player 9. However, for our stock portfolio application, we'll need to install and deploy FDS, which is a Web application that can be deployed in any JEE compliant container. Since FDS is a Web application, you can deploy it in the application server of your choice. First, we'll download and install Apache Tomcat 5.5 from http://tomcat.apache.org/. By default, on Windows machines Tomcat is installed in "C:\Program Files\Apache Software Foundation\Tomcat 5.5" and runs on port 8080. At the end of the installation, the Apache Tomcat server will be started as a Windows service. Now download from and run the FDS executable installer (default directory ID C:\fds2), picking the optional Flex Data Services J2EE Application. After the installation is finished, unzip the content of the flex.war into Tomcat's webaps\ROOT directory and restart the Apache Tomcat service. Enter the URL http://127.0.0.1:8080/ in your browser, and you should see the FDS Welcome screen. Flex Data Services are deployed now under Tomcat.
Developing Stock Portfolio Application with Flex
What Runs Where
First, let's see which components will run on the client and what belongs to the Web server. The code shown in Listings 3-6 will be compiled into portfolio.swf, which will run on the client either independently in the Flash 8.5 player, or with the help of portfolio.html (auto-generated by Flex Builder) in the Web browser with the Flash plug-in. Tomcat is our server, which will host compiled Java classes shown in Listings 7 and 8. We'll also make small additions in the server configuration files (see Listings 9 and 10) to specify where to find the Java classes and to allow access to the external Yahoo! service. While Flex provides different ways of client/server communication (RemoteObjects, HTTPService, WebService), in this article our client will use the RPC by means of the <mx:RemoteObject> component that will find the matching Java classes located and configured on the server as described at the end of this article.
Developing the GUI Part
Open the Flex perspective in Eclipse and create a new project (e.g., Portfolio_RCP) containing files from Listings 3-7, and then compile it into one file called portfolio.swf. To simplify deployment, set the project's output directory to Tomcat's \ROOT\portfolio.
Our main file portfolio.mxml (see Listing 3) divides the screen with the vertical Flex box-type layout (<mx:VDividedBox>) with an adjustable divider between its children (think of Swing's split panes), and it includes two separate code fragments. The bottom child contains a data grid programmed in <FinancialNews> (see Listing 6). The top child contains <PortfolioView> (see Listing 4).
Java developers know that some Swing components store the data and GUI parts separately, for example, the JTable gets its feed from a data model class that can store its data in a Java collection. On a similar note, some Flex controls (e.g., <mx:DataGrid>) can also use ActionScript collection classes as data providers. Our data model is the file portfolio.xml depicted in Listing 2.
This code represents a typical master-detail relationship, where a change in the selected security in the master component <PortfolioView> repopulates the detail component <FinancialNews>.
Getting the Price Quotes
Let's go over some interesting programming techniques that we've applied while coding the top panel, which contains several Flex framework components wrapped into a Canvas container (see Listing 4). After learning that the user's portfolio is represented by portfolio.xml, you may expect at least some number of lines performing XML parsing.
What do you think of this line:
<mx:XML format="e4x" id="portfolioModel" source="portfolio.xml" />
The XML parsing is complete! After this line you can traverse the entire XML document using a simple dot notation, e.g., portfolioModel.security. This magic is possible because Flex supports the e4x format that was introduced by W3C as a simple alternative for reading and processing XML documents.
Another interesting feature to comment is data binding, which is a way to link the data in one object to another. In particular, it's a convenient way to move the data between the GUI and non-visual components. Data binding can be specified using the curly braces syntax. The following lines from Listing 4 provide an immediate refresh of the top data grid and chart as soon as the data in the underlying security changes.
<mx:DataGrid id="portfolioGrid" width="100%" height="100%"
dataProvider="{portfolioModel.security}"
<mx:PieChart id="portfolioPie" dataProvider="{portfolioModel.security}"
The variable portfolioModel.security represents the data source, and the destination will change as soon as any property of the source changes. Note that we have one data source mapped to two destinations (the data grid and pie chart), so we don't need to do any additional programming to refresh the data grid and the chart when the incoming market data changes the properties of the security.
The tag <mx:RemoteObject destination="Portfolio" id="freshQuotes" > declares the connection of our client to a remote object that resides on the server and goes by the name "Portfolio" (see Listing 10).
Listing 4 has the ActionScript code embedded into mxml in the tag <mx:Script>. It has a public variable marked with a [Bindable] metadata:
[Bindable] public var selectedSecurity:String;
This variable will automatically send notifications as soon as its value changes to all registered listeners. This may happen as a result of the change event in the data grid or if the user clicks on the slice of the pie. Selecting a different stock symbol (security) will trigger the repopulation of the news grid. No additional programming is required. Wait a minute, we've never programmed any event listeners for the variable selectedSecurity! That's correct, but when Flex compiles Portfolio.mxml (see Listing 3), it'll notice the bindable public variable pv.selectedSecurity, so it'll not only generate a registered required listener, but it will also call the function set security in the FinancialNews panel (see Listing 6).
The show begins as soon as the painting of the top portion of the screen (<mx:Canvas>) completes. Its creationComplete event handler calls the ActionScript function startQuotes(), which connects to the remote POJO and calls its method getQuotes() every second. The tag <mx:RemoteObject> also contains the element declaring how the method getQuotes() should be invoked. As soon as the result of this method call arrives, it's being passed to the function applyQuotes(). All Flex remote calls are asynchronous and require an ActionScript method that will process the result of the call. In our example, applyQuotes() is such a method. The concurrency attribute defines how the remote object should process multiple requests. For example, the user requested quotes for one security, but it takes a bit longer than usual, and she sent a quote request for another security. The attribute concurrency="last" means that older responses should be ignored.
Another interesting line illustrating ActionScript capabilities is e4x object access in Listing 4:
var row:* = portfolioModel.security.(Symbol==quote.symbol);
Flex is a great platform for any XML-related processing. It'll automatically handle Symbol==quote.symbol as XPath expressions behind the scenes, providing a code-free approach for data navigation.
Just to tease you, we've decided to keep commented lines <mx:Consumer> and consumer.subscribe()in the code to give you an idea what has to be added to replace the AJAX-style price quote polling performed by setInterval() with a server push implemented by publish/subscribe messaging. We'll cover this topic in our next article.
The top toggle buttons Show Grid/Show Chart (see Figure 2) use images, and it's a good idea to embed these resources right into the SWF file. This way the browser can download the entire client in one HTTP request (but the size of the SWF file becomes larger). For the same reason multi-file Java applets are packaged into a single JAR. There are two lines in Listing 4 that embed the images and assign them to reference variables. PorfolioView displays them as follows:
<mx:VBox label="Show Grid" icon="iconGrid" ...>
<mx:HBox label="Show Chart" icon="iconChart" ...>
Listing 5 contains the ActionScript code that's used on the client just to help Flex perform Java introspection more efficiently; we've declared a class with all properties that exist in its Java peer on the server (see Listing 8).
Getting the Financial News
The bottom data grid
(see Listing 6) displays the news headlines supplied by the Web site.
This time we're not using <mx:RemoteObject> but another flavor of
RPC called <mx:HTTPService> that points at http://finance.yahoo.com/rss/headline, which is known to clients as YahooFinancialNews (see Listing 9).
Again, the HTTPService call works asynchronously: as soon as security
setter is called (set security), it sends the request
newsFeed.send({s:value}) to the server. In this line, "s" is the name
of the parameter, and the value should contain the selected security.
For example, if the user selects MSFT, the server destination
YahooFinancialNews will receive the following URL: http://finance.yahoo.com/rss/headline?s=MSFT. Just try to enter this URL manually in your Web browser to see the XML our Flash client will receive as a result of this call.
Received XML is used as a source property of the <mx:XMLListCollection>. The data grid's columns are mapped to the nodes of this data provider. The collection and then GUI are refreshed upon arrival of the result (a property of newsFeed). You don't have to use a collection as a middleman between the data and the GUI component. However, collections may become handy if you'd like to perform some additional data massaging before presenting it to the user, e.g., filtering or search using e4x.
The following line puts the result of the HTTPService request (newsFeed.result) into the source property of an XMLListCollection:
<mx:XMLListCollection id="newsList" source="{newsFeed.result.channel.item}" />
If you look again at the structure of the XML returned by the URL http://finance.yahoo.com/rss/headline?s=MSFT, you'll notice that the channel is the root element there and each headline is represented by the element called item.
The last column in the A composite element is called <mx:LinkButton>. It's a button that looks like a hyperlink. We've carefully wrapped it into a container <mx:Component>, which in turn sits inside <mx:ItemRenderer> (imagine a Swing cell renderer on steroids). Click on this link to navigate to this URL, which will be opened in the popup window. The argument of the URLRequest constructor is data.link; data is a reference to the current row in the data, and link is the name of the column.
Now open the Java perspective in Eclipse and create a new project (e.g., Portfolio_RCP_RO), which will contain our Java classes. The POJOs that live in the Tomcat server are very simple. The StockQuoteDTO .java (see Listing 7) contains the last price of a particular stock. The class Portfolio.java (see Listing 8) is a simple, random number generator simulating market-like real-time price changes for several hard-coded securities. Compiled Java classes should go under Tomcat's directory WEB-INF\classes.
Configuring the Server-Side Destination and Proxy
For security reasons, Flash clients can only access the servers they
came from, unless there is an agreement between our server and external
service providers, which agreed to deal with our server and are
declared in a crossdomain.xml file. However, our portfolio SWF was not
loaded from finance.yahoo.com, and we are not allowed to install
crossdomain.xml on the Yahoo! Servers. We'll use another technique
called Flex proxy. When the user clicks on the News link in the data
grid, the portfolio client will connect to our FDS application deployed
under Tomcat, which will proxy our communication with Yahoo!. To
configure the Flex proxy service, see the following section to the
flex-proxy-service.xml located in the Tomcat's \WEB-INF\flex directory
(see Listing 9).
Now FDS will contact http://finance.yahoo.com, get the news for a specified symbol, and return it back to the Flash client.
As for POJO on the server side, Flex provides configuration files that allow you to hide the exact service providers details (e.g., actual Java class names) by specifying so-called destinations. The following section from Listing 10 has to be added to the flex-remoting-service.xml file.
Clients won't know that the actual name of our POJO is com.theriabook.ro.Portfolio, but they'll be able to refer to it by the nickname Portfolio.
Now start your Tomcat service and run the Flex application (portfolio.mxml) in the Flex Builder's project Portfolio_RCP. You should be able to see the screens as in Figures 2 and 3, the "market data feed" should modify the prices and you'll be able to read the latest Yahoo! news on your stocks.
Conclusion
In this short article we managed to
develop an application with not so trivial functionality, but the
amount of code we had to write was minimal. Unfortunately, the amount
of explanation we could provide was minimal as well. If you'd like to
start learning Flex 2, you can find lots of quality introductory
materials and product documentation at http://labs.macromedia.com/wiki/index.php/Main_Page . Our upcoming book Rich Internet Applications with Adobe Flex and Java (www.theriabook.com ) will show how to face-lift your enterprise Java applications using advanced and not-so-obvious Flex techniques.
© 2008 SYS-CON Media Inc.