Pet Store with JavaFX 1.0. Part 1.

I work for a company called Farata Systems. Laura is our office manager, and she doesn’t know Java. Everyone else does.  We work in the enterprise RIA space using Adobe Flex as a main development tool for creating all these Web pages connecting to the server side applications written in Java.

In December of 2008, Sun Microsystems has released the first version of a language for developing UI for RIA. I’ve been watching the evolution of JavaFX from its announcement at JavaOne 2007 till its release, but wasn’t able to find any demos that would prove that this extension of Java platform can be considered as a candidate for boring needs of enterprise applications. Watching demos like a bunch of balls moving inside a glass didn’t look too convincing to me.

We decided to contribute to Java community and invest some time and resources to figure out what JavaFX 1.0 is capable of. Most of the Java developers know what Pet Store application is. For many years this reference implementation has been used by Sun Microsystems as a part of Java BluePrints program for Java EE platform.

The current version of this application is described in the article called “Introducing the Java Pet Store 2.0 Application” that’s available at http://java.sun.com/developer/technicalArticles/J2EE/petstore/.

We re-used the existing back end of the Pet store, but performed a little facelift by replacing the UI developed with AJAX/JSP/JavaServer Faces with JavaFX 1.0.

I was working on this project with our senior Java developer Alex Litskevich who’s done lots of digging and research trying to find the best solutions whenever we would run into limitations of JavaFX.
The goal of this introductory article is to demo the JavaFX Pet Store. You can watch a quick prerecorded demo at http://myflex.org/demos/petstore/petstore.html .

In the future, I’ll deploy this application in one of our public servers and will review the code.

Last year, while attending a conference in India, I met an excellent Sun’s software engineer Greg Murray, one of the authors of the idea of Pet Store application.   Greg, I hope you’ll like this version of your pet store.

Flex, JavaFX, Silverlight…

I’ll be honest with you, during development of this application in the back of my mind I’ve been constantly comparing the way things are done in JavaFX with how I’d do the same with Adobe Flex.
Simply put, today, Flex is a lot more productive than JavaFX.  But let’s remember that we are comparing products of versions 3 vs. version 1. If you know how to develop Swing applications, you might want to continue doing this rather than manually creating JavaFX 1.0 wrappers for Java code as we did in our Pet Store.

The biggest issue is that JavaFX 1.0 doesn’t include any UI component library that a Web developer needs. I’m not talking about data grids or tree views. Even such simple things as a hyperlink component aren’t there. I can’t think of any other reason but a trivial lack of time and limited human resources allocated by Sun for development of this new UI set.

Hopefully, Sun will release a JavaFX 2.0 in foreseeable future with a full set of UI components, containers, navigators, and layouts. Just to draw an analogy, Microsoft has released absolutely naked Silverlight 1.0 in 2007, but full featured Silverlight 2.0 became available a year after.
The good news is that Java developers have already started building their own open source components for JavaFX. The name of this project is jfxtras .

On the other hand, I see some features available for JavaFX developers that are missing in Flex.

For example, ActionScript doesn’t support multi-threading. This is not an issue when it comes to communication with remote sites – all remote calls are asynchronous and use callbacks for server-to-client data flows. But if you’d need to write or use an existing Java library that performs some CPU-intensive calculations on the client, I’m sure Java will beat ActionScript.  The syntax of both languages is pretty similar though (see http://flexblog.faratasystems.com/?p=115).
JavaFX supports eval() function that allows to compile and execute a piece of code on the fly. Flex doesn’t  support this useful feature.

Two-way binding already exists in JavaFX but in Flex will get it only in version 4.

I’ve yet to see some benchmarks comparing Flex and JavaFX network protocol.  The latter just offers XML and REST for data exchange between client and server.  Of course, you can program Java sockets or wrap up an RMI client into JavaFX – would it be faster than ActionScript Real-Time Messaging Protocol? We don’t know. Or take HTTP tunneling between   Java Applet and a server-side Java stream. Would it be faster than using AMF, an HTTP-based binary protocol? We don’t know yet.

JavaFX removes a lot of complications in developing mobile applications. It promises to remove the need of using Java SE for desktop and Java ME for mobile – JavaFX should be the only solution you need. JavaFX Mobile offers hardware acceleration.  Keep in mind, there are 2.6 billion of Java enabled mobile devices! JavaFX has serious advantage in this domain over Flash Lite.

JavaFX 1.1 is about to be released. It offers performance improvements, bug fixes, new features in the mobile space, but no major additions to the desktop platform.

I understand that this is not a good time to ask for some serious cash injections, but if executives of Sun Microsystems won’t find a way to double the size of the team working on the future version of JavaFX, entire Adobe Flex team can safely go on sabbatical for a year.  If the money is tight, the sooner JavaFX will go open source, the better.

Even though I wouldn’t start developing with JavaFX a project that needs to go in production this year, overall, JavaFX looks promising. Try it out and see for yourself.

Yakov Fain

My upcoming talks on Flex, JavaFX and $300 coupon

During the next three months I’ll be making a number of technical presentations for Flex and Java developers. These are the confirmed events:
1.    New York Flex User’s Group, February 19, Flex design patterns, details should be available soon at http://nyflex.org
2.    Java developers conference in Atlanta, March 11 – a keynote on comparing technologies for rich Internet applications and a technical session on automatic code generation for CRUD RIA. Details at http://www.devnexus.com/
3.    Chicago Flex User’s Group, March 18, Flex design patterns, http://chicagoflex.org/
4.    Princeton Java User’s group, Developing Pet Store in JavaFX. I’ll schedule this meeting some time in March. Keep an eye on http://myflex.org/princetonjug
5.    Flex conference 360Flex, Indianapolis, IN, May 18-20, Design patterns: a Deep Dive: http://www.360flex.com
My colleague Victor and I will run a two-day public workshop for enterprise Flex architects in New York city on March 2-3, details are available at http://www.eventbrite.com/event/247789144 .   Readers of this blog will get $300 discount by entering a special code adjusting2crisis during registration.

That’s all for now.

Yakov Fain

http://twitter.com/yfain

Getting my hands dirty with JavaFX

I’ve started looking into JavaFX and plan to publish a couple of articles going through development of an application that real-world software developers can relate to. Enough of the demos of bubbles moving up and down in a glass with water. Can’t recall any application in my rather long career that would require to program something like this.

While looking at the JavaFX way to do stuff, I subconsciously compare it with Flex.  So far as expected, Flex is a hundred miles ahead. But to be fair, one shouldn’t forget that we are comparing products of 1.0 and 3.0 versions. At this point, JavaFX would require to do a lot more of manual coding in Java than I’d like to see.

Lack of such important  components as data grid would require you to wrap Java’s JTable and program to it. You need to program such basic things as a hyperlink component…

Eclipse plugin for JavaFX is not real. As of today (January of 2009), to use JavaFX you need to install NetBeans 6.5 and have a background of  a Java Swing programmer, which will be an obstacle for many Java developers.

But, let’s give this product a little bit of time to mature.

I’ll write my first JavaFX article for Java Developer’s Journal in February. BTW, have you seen a new look of JDJ and  Flex  Developer’s Journal ? Expect some more very interesting announcements from sys-con that may turn you into a published author.  Just keep an eye on the word Ulitzer.

Yakov Fain

JavaFX : The birth is a little premature

On December 4, JavaFX  arrived to this world with lots of “It’s a Boy” balloons. The family of RIA development tools gets bigger and as a Java programmer, I wish best of luck to this new kid on the block.

But the party is a little spoiled because the baby is premature and will have to spend some time in barocamera to survive. To make myself clear, I’ll mixin here some definitions from Wikipedia in italic:

Human fertilization is the union of a human egg and sperm, usually occurring in the ampulla of the fallopian tube.

Illustration by Yuri Fain.

In my opinion, this definition is OK for bulls and cows, but in the human world it would be nice if the bearers of sperm and egg loved each other and wanted this baby.   In May of 2007, when Sun Microsystems announced JavaFX, I did not see any traces of love there.

This was forced fertilization. Sun, which technically created the first rich Internet application in the form of applets in 1995 did not give enough love to that baby too concentrating its efforts on Java technologies that would help selling Sun’s servers. As a result, Java EE shines as a platform for development and deployment of the server-side enterprise applications.

Java Swing , the client side library, was too difficult to program and most importantly, required large Java runtime that was not too easy to install, had a slow 0-60 acceleration. Java on the client did not have a runtime environment that was feasible for consumer-facing RIA. A trucker from Alabama would not be able to install the proper version of JRE.

In the world of RIA, availability and high penetration of the runtime environment is the most important pre-requisite for any RIA technology or technique.  High penetration of Web browsers supporting DHTML plus XMLHttpRequest object is the explanation of the three years of AJAX craze that’s, thanks God, is coming to an end.
About ten years ago, Sun “won” a lawsuit with Microsoft and made $10B. These bad guys from Redmond wanted to make our Java dirty by quietly introducing their infidel libraries.  Sun went into a fight, got the money… and lost the mechanism for spreading Java runtime. The Internet Explorer, that’s being used by about 75% of the users, even today ships with the ten-year old Java 1.1 runtime.

On the other hand, Adobe’s Flash Player is a runtime for Flash and Flex application that’s small, has seamless installation and has a penetration rate of 98%. Flex 2 was announced three years ago after Adobe has acquired Macromedia, and not only offered nice framework for developing UI, but also came with advanced network protocols and easy integration with Java EE on the server.

Microsoft was first to recognize Flex as a new RIA platform leader, and as usually, responded, “Me too.” And when Microsoft says, “Me too”, they really mean it. Some very serious dependency injection (I’m talking about long and green) did the trick, and Silverlight 2.0 may start competing with Flex in a year, if the penetration of the Silverlight’s runtime will reach at least 80%.

During JavaOne 2007, Sun realized that they missed the RIA boat and announced JavaFX loud and clear. They were forced to bed, but the sperm met the egg and the process has begun. Marketing people and executives started make some serious noise about JavaFX.

Pregnancy (latin graviditas) is the carrying of one or more offspring, known as a fetus or embryo, inside the uterus of a female.

But just running around screaming, “I’m pregnant, I’m pregnant!” is not enough for delivering a healthy baby. Pregnant women should have better nutrition, and should be surrounded with love and support. I was following the evolution of JavaFX very closely and I didn’t see this love and support from Sun’s top management. Sun’s executives were trumpeting about new and great product, but I’d rather see them providing some real support to a small group of talented software engineers that were trying to do the bet pushing the product out the door.
Please read the story of an Chris Oliver, creator of JavaFX.  Sun published this strory under the title “Mind-Bendingly Cool Innovation ”. On the surface, it looks like a success story of a boy who didn’t even go to college but became a Senior Stuff Engineer at Sun.  But one of the Oliver’s statements tells me that he had lots of obstacles within Sun while working on JavaFX:
“For us, a bunch of tedious work – what I wanted to do but was obstructed for the last several years. Now I won’t be obstructed, thanks to the new management. Now the path is open for us to walk down it, but we have to actually walk down it,” he says.”

I’m sure, you know that it take a women nine months to deliver a baby. In software, it usually takes a lot longer, and the released version of JavaFX gives me a feeling that it’s premature.

The Web site javafx.com didn’t impress me.  The Web site itself was falling down several times during the opening day – it was not tested well enough. This is yet another evidence of insufficient resource allocated to this project. There is no demos of the UI for enterprise-grade applications that I was expecting to see.  As a matter of fact, long time ago I’ve been sending emails to people who were involved with development of JavaFX asking for at least a Pet Store demo. It didn’t happen yet.
The start of the available demos is still slow. The ugly messages like “Java Starting” and “You may not have the right certificate. Do you trust this application?” will scare that Alabama trucker to death and he’ll run away from this Web site.

On the positive side, JavaFX has pushed Sun to developer Java Kernel, a relatively small JRE that loads faster. The other smart thing is their attention to designer-developer workflow. Sun decided not to re-invent the wheel and create a plugin that should allow designers continue using Adobe Photoshop and Illustrator with automatic JavaFX code generator.

Sun made yet another smart decision – decoupling of the JVM from the browser. You should be able to start JavaFX from the browser, but then kill the browser and your applet should stay alive. You should be able to drag the applet from the browser and drop it right onto yourdesktop. Are there any demos of this features? I’m not sure – have not seen yet.

Yet another feature to watch is the automatic detection of the  JVM version. Inclusion of a small JavaScript piece into the HTML wrapper of your applet should check it out and ensure that the user machine has JVM 1.6U10 or above.  If not, the JVM should be SEAMLESSY upgraded. I haven’t seen this feature in action yet- running available demos on ny MacBook Pro, which has only JVM 1.5, and so far I haven’t been forced or offered to do an upgrade.

Now Sun has to offer tools for developing JavaFX applications. Sure enough, you can use NetBeans 6.5, but for some reason, majority of Java developers use Eclipse. Quick introduction of Eclipse plugin for JavaFX should be a highest priority item for Sun. Just bite the bullet and do it.

Even though I’m using Adobe Flex for development of the UI portion of enterprise RIA, I don’t fall in love with programming languages, and would be gladly using JavaFX too when it’s ready for production.   It may happen in 2010. But Sun has either to put the money where its mouth is or open source JavaFX to allow Java enthusiasts make this language competitive in the RIA world.

Yakov Fain