|
YOUR FEEDBACK
Did you read today's front page stories & breaking news?
SYS-CON.TV |
TODAY'S TOP SOA & WEBSERVICES LINKS Feature Wireless Web Serviceswith J2ME Part IISOAP or XML-RPC? The answer depends on your needs
Wireless Web Serviceswith J2ME Part IISOAP or XML-RPC? The answer depends on your needs
By: Kyle Gabhart; Jason Gordon
Feb. 1, 2002 12:00 AM
Last month in Part I (WSJ Vol. 2 Issue 1) we discussed J2ME and accessing Web services from wireless devices using the XML-RPC protocol. In this article, we will consider SOAP as a vehicle for accessing Web services from wireless devices, comparing and contrast-ing it with XML-RPC. Our sample application will again be a J2ME midlet, however, we will use EnhydraME's kSOAP rather than kXML-RPC to provide the protocol's implementation.
Overview of SOAP SOAP was originally conceived and developed at Microsoft between 1998 and 1999. SOAP did not, however, gain widespread attention until DevelopMentor, IBM, Lotus, and Microsoft submitted the SOAP 1.1 version to the W3C on April 26, 2000. With both IBM and Microsoft behind it, the industry began to give SOAP some serious consideration. As of this writing, the SOAP 1.2 specification is being drafted by the W3C's XML Protocol Working Group. The latest draft of the 1.2 spec can be found at www.w3.org/TR/soap12/. When examining SOAP, it is important to identify the three main components of any SOAP message: the SOAP envelope, rules for encoding data, and a request/response interaction mechanism. The SOAP messaging architecture can be compared to a postal system (see Figure 1). A document is enclosed in an envelope and that envelope is transmitted via the transport mechanism, the mail system in our analogy, or HTTP across a network in the case of SOAP. With the mail, a zip code identifies the locale an envelope should be delivered to; with SOAP, HTTP header data provides such routing information. At this point, the analogy breaks down. Whereas the "last mile service" provided by a postal carrier is directed by the street address on the envelope, it is information encoded in the body of the XML document itself that ultimately directs the SOAP request to the specific service on the server.
XML-RPC vs SOAP XML-RPC is an extremely lightweight mechanism for invoking XML-based services. It is a clean, simple protocol that provides the minimum overhead necessary to invoke remote services and exchange data in a platform-, language-, and vendor-neutral manner. It defines six simple data types and two complex types. The result is that XML-RPC is a highly-efficient lightweight protocol. Messages are simple to construct, simple to parse, simple to debug, and are easily human-readable. XML-RPC requires a minimal amount of active memory for processing, building, and parsing messages, and produces a thin message body to be exchanged between client and server. In contrast, SOAP is a fatter protocol (although generally considered lightweight). In exchange for some additional overhead, SOAP provides namespace awareness, a sophisticated data-typing mechanism, and a flexible messaging paradigm. The W3C's XML Namespaces specification is leveraged to provide namespace awareness, and the XML Schema specification provides the data-typing mechanism. As such, SOAP supports over 40 standard data types and provides the capabilities to define custom simple and complex data types. This provides a tremendous degree of flexibility in terms of describing robust data structures with intricate relationships with other data contained in the message body. The SOAP architecture also introduces a flexible messaging architecture, supporting a variety of messaging paradigms, including unidirectional, bidirectional, multicast (publish-subscribe), and sequentialmessaging (multiple parties chained together in a particular order). One aspect that facilitates these paradigms is the ability of a SOAP message to include a header section. This allows security, transaction, and routing information to be exchanged between multiple parties. For example, a client can send a SOAP request with an intended destination, but also declare in the SOAP envelope's headers that a particular party should receive the message, process the pertinent header information, and then pass the message on to the next party in the chain (or the intended recipient if no other parties have been declared). One final distinction is that SOAP supports asynchronous messaging, while XML-RPC requires a synchronous communication between the two parties in an exchange. With all of these differences between the two protocols, executives, project managers, and wireless architects alike are wondering - "How do I make a choice between SOAP and XML-RPC for a particular wireless project or system?" Well, even if you weren't wondering that, we're going to tell you anyway. In a nutshell, XML-RPC provides a fast, compact protocol for exchanging data and invoking services in a neutral, standardized way. If application size, memory, and bandwidth are your top priorities, then XML-RPC is the way to go. On the other hand, if your application requires a robust data-typing mechanism, extensive security or transaction support, or a flexible messaging architecture, then SOAP is your answer. Also, SOAP is more mainstream than XML-RPC, so if your application needs to access a variety of services that you have no control over, SOAP may be a better match from an interoperability perspective. To break it down into more detail, Table 1 that outlines the business justifications for using each protocol.
Wireless SOAP Example For our application, we will be using three classes from the kSOAP library to handle the marshalling and unmarshalling of data via SOAP:
In creating our sample SOAP midlet (AirportWeather.java), the process is identical to the one we used in Part 1. This time, the packages and names have been changed to protect the innocent. We will import, extend, and use classes from the kSOAP library rather than the kXML-RPC library. The first step, obviously, is to import the necessary packages and declare the MIDP components that will be used in the application. After this, we define the midlet's constructor, initializing all the UI components, and add them to the display as necessary. With that complete, we need to fill in the three other lifecycle methods. In the startApp() method, we simply bring the MIDP display into action. Since we don't use any shared resources, the pauseApp() method is blank. Finally, the destroyApp() method releases the local resources that we have allocated for our midlet. As with Part 1, all the action takes place in the commandAction() method. This method is called anytime the user performs a command event (pressing a key, selecting an item from a list, etc.). The Command and Displayable objects are then queried to determine which component has actually been activated/ deactivated, and the appropriate actions are performed. When our midlet is launched, it displays a list of popular international airports for which weather information can be retrieved (see Figure 2). It also displays an exit button to allow the user to exit the midlet. Upon selecting an airport, a list of weather information services for that airport is displayed (see Figure 3). A back button is also displayed to allow the user to return to the airport menu. The user then selects the weather information service he or she is interested in and the midlet sends a SOAP request to the server to collect the desired information. A SOAP response is returned, parsed by the midlet, and the result is displayed on the screen. So how does all of that work? See Listing 1 for the command Action() method. The outer level of the commandAct ion() method checks to see what component has been activated. We'll look at each of these four events individually, beginning with the airport menu: if ( dis == airportMenu && com == List.SELECT_COMMAND ) { When an airport is selected, we set the currentAirport variable, set the title for the services menu to reflect the selected airport, and display the services menu. Next we'll look at the services menu. There are seven possible weather services that can be retrieved: wind conditions, temperature, pressure, humidity, sky conditions, visibility, and a complete summary containing the six individual pieces of information: else if ( dis == servicesMenu && com == List.SELECT_COMMAND ) { The first thing to do is to declare a String variable to store the result of the service call that will be made. Next, determine which service has been selected and store the index value in an integer variable and perform a switch on that variable. The switch statement can be seen in Listing 2. The first case retrieves a summary and the other cases retrieve individual weather items. After the switch statement, the results are displayed on the screen via an Alert object. Within each case statement in the switch block, the callSer vice() method is used to handle the exchange of SOAP messages. See Listing 3). Four essential things take place in this method. An HttpTran sport object is created, a PrefixMap namespace is created for the SOAP envelope, a SoapObject is created to represent the request, and the request object is sent via the HttpTransport class's call() method. The next else/if block checks to see if the back command has been selected, in which case the list of airports is displayed: else if ( com == backCommand ) { Finally, the exit command is checked to allow the application to be exited: else if ( com == exitCommand ) { That's all there is to it. Download the source code to get the details for the interface and then you are ready to build the midlet interface, compile the code and access this weather service from a J2ME phone using SOAP!
Conclusion SOA WORLD LATEST STORIES
SUBSCRIBE TO THE WORLD'S MOST POWERFUL NEWSLETTERS SUBSCRIBE TO OUR RSS FEEDS & GET YOUR SYS-CON NEWS LIVE!
|
SYS-CON FEATURED WHITEPAPERS MOST READ THIS WEEK |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||