Posts tagged: Vaadin

Map component for JVx applications

Post to Twitter

We played around with some interesting stuff in the last weeks. Some customers and users asked us if we have a Map component.

We don't have a ready-to-use component but there are many free and commercial solutions available. The integration in a JVx application with custom components is not a problem and doesn't need much effort. But sure, an out-of-the-box solution would be useful.

Our Research team did create a PoC for a Map. The results is very nice and we want to show you some screenshots from our tests.

Swing integration (Tab mode)

Swing integration (Tab mode)

Swing integration (frame mode)

Swing integration (frame mode)

Vaadin integration

Vaadin integration

Vaadin integratin (corporation mode)

Vaadin integratin (corporation mode)

We didn't add the component to JVx because it's just a PoC and not ready-to-use.
But the Map integration looks great :)

The future of JavaFX?

Post to Twitter

After one year of JavaFX development, we thought it's time to write about our experience.

We started in Nov. 2014 with our first bigger JavaFX tests. The idea was an UI implementation for JVx. We thought it should be an easy task because JavaFX is a desktop technology. We had another research project in 2012, but the result wasn't as expected. Sometimes it's a good idea to re-start from scratch!

JavaFX wasn't very popular and is still not a Superstar, but we love researching and we thought it might be a good idea to bet on JavaFX. Not only as pure desktop solution and Swing replacement, but also for mobile app development based on JavaFXPorts and for embedded devices, like Raspberry Pi. So we had this big picture in mind.
We thought it would be awesome to create a single JavaFX application, for all platforms and devices. Sure, some other companies tried to solve this problem as well, but we didn't try to do the same. Our idea was simple and based on JVx. We hade this awesome JVx application framework which enabled GUI and technology independent application development. So we tried to solve the GUI problem because all other things like persistence, remote communcation, i18n, ... were already solved by JVx.

We did different GUI implementations for JVx, before we started with JavaFX. Our first one was based on Swing, followed by QTJambi, GWT, GXT, a headless variant and - last but not least - vaadin. We also made some tests with Flex and Pivot, but we didn't create more than some simple PoC. To be clear: GUI technology know-how was not the problem!

We knew that JavaFX didn't have a solution for MDI and especially a DesktopPane was missing, but it wasn't a problem to create custom components/controls. We had some hard months of development because we found out that JavaFX wasn't a mature toolkit like Swing. We had to find many workarounds for very simple problems and had to create more custom components than expected. We solved most problems in our own code, but it wasn't possible to change some basic problems. We tried to get in contact with the developers, via JIRA, but it was hard to explain real-world use-cases to toolkit developers. Our developers at SIB Visions create custom products and projects together with customers for their needs. So we had many UI toolkit requirements from real users. Some issues were accepted but not all. Many bugs were fixed but we had big problems with Feature Requests. So far so good.

We finished our JavaFX UI implementation in April 2015 and it was a big success, because we had a fresh UI technology with lots new features and animations, transitions, web and media views, .... awesome.

We thought that JavaFX will be improved in 2015/2016 and it will be mature enough to use it for our customers.

Wait... what's the problem with JavaFX?

Performance, Memory, limited chances to modify standard controls

The performance is not good, if you have many controls (= nodes). It's different on Windows, Linux, MacOS. It's much better on MacOS than on any other supported OS. If you develop business/backoffice applications, the performance is very important. Sure, if you create simple form based applications, it doesn't really matter.

We use JVx to develop huge backoffice applications and currently, JavaFX should be used with care.

What do we mean with "performance is not good"?

  • Large TableViews have scroll delays
  • Layouting sometimes need some "extra repaints"
  • Scrolling with many controls is not immediate
  • Startup time of an application is too long, compared to e.g. Swing applications

The version number of JavaFX 8 is too high for it's maturity.

JavaFX has big potential but we're not sure if it'll survive it's predecessor Swing.

And we're not alone with our opinion: Should Oracle Spring clean JavaFX

But there are also companies with JavaFX applications: JavaFX Real-world apps

There's more

The toolkit issues are one thing. Another big surprise was that Oracle stopped official ARM support in Jan 2015 which didn't mean that JavaFX doesn't run on ARM devices (RaspberryPi, ...), but it's not officially supported and Oracle doesn't fix bugs or implement improvements.
Thanks to Gluon it's still very easy to use JavaFX on ARM devices.

Another step backwards was the migration to Bugzilla. In 2014 (and earlier), JavaFX was a "newcomer" and announced as Swing replacement. The development process at Oracle was very open and you had a "direct connection" to the development team, or something like that. It was very simple to report issues and to get feedback from the developers. It was fantastic to see and feel the progress. This is different nowadays, because the ticketing system moved from JIRA to Bugzilla and it's more or less readonly for standard developers.

One problem with JavaFX is that Oracle doesn't tell us the strategy behind JavaFX. The investment of Oracle in JavaFX was high, but what will be the next step? It's hard to bet on JavaFX without definitive commitment from Oracle.

But we still hope that JavaFX will be the one and only UI toolkit for Java.

JavaFX for Android and iOS

In summer, we moved forward and tried to use JavaFXPorts. It makes it possible to use your JavaFX application as native Android and iOS apps. Our JavaFX UI worked with some smaller changes out-of-the-box. It was awesome to see our JVx on Android or iOS devices! It was a milestone.

But things have changed since summer because JavaFXPorts has a depency on RoboVM. It was an OpenSource project and was a big win for the community. But now it's a commercial tool (read more). This has no impact for us, but means that we can't offer an out-of-the-box OpenSource solution.

JavaFXPorts also works without RoboVM and there are plans for a JVM running on mobile devices, but this won't solve the problem that you need access to device features/APIs. But we'll see what happens in the next few months.

We had some fun in the last few weeks because we tuned our application frame for mobile devices. A mobile application has some specific characteristics and isn't a 1:1 desktop application. A desktop application usually has a menubar and a toolbar. The same app on a mobile device shouldn't use a menubar or toolbar. It should offer a hamburger button which offers a custom menu. Also the style of the controls should be different on mobile devices, e.g. Tabsets.

There are two interesting projects which offer custom mobile controls. The first is a commercial product, from Gluon and the second one is JFoenix and it's OpenSource (LGPL). We took JFoenix because it met our requirements and JVx is OpenSource.

We have some screenshots in older posts:

Resume

JavaFX has potential to rule the Java UI world, but it needs an official commitment/statement from Oracle. Now it's unclear if JavaFX is the right UI technology for the future.

We hope that JavaFX is the right toolkit, but we aren't sure. Our investment was risky and high, but no risk - no fun. The results are awesome!

Thanks to JVx the UI technology is not important for us, because if JavaFX won't make the race, we still have Swing or vaadin. It's important to be UI technology independent because it saves time and money and is future save.

VaadinUI without web.xml

Post to Twitter

We've added some annotations to our vaadin UI to support UI configuration without web.xml (deployment descriptor).

Check this:

@Push
@Widgetset(value = "com.sibvisions.rad.ui.vaadin.ext.ui.Widgetset")
@Configuration({@Parameter(name = "main", value = "MinimalApplication$MyApplication")})
@SuppressWarnings("serial")
public class MinimalApplication extends VaadinUI
{
    @WebServlet(value = {"/minimal/*", "/VAADIN/*"}, asyncSupported = true)
    @VaadinServletConfiguration(productionMode = false, ui = MinimalApplication.class)
    public static class Servlet extends VaadinServlet
    {}

    public static class MyApplication extends Application
    {
        public MyApplication(UILauncher pLauncher)
        {
            super(pLauncher);
           
            setName("JVx application");
            setLayout(new UIBorderLayout());
           
            add(new UILabel("Hello JVx' vaadinUI"), UIBorderLayout.CENTER);
        }        
    }
}

The example demonstrates a minimalistic JVx application with vaadin UI. It's like a standard vaadin application with one additional annotation:

@Configuration({@Parameter(name = "main", value = "MinimalApplication$MyApplication")})

The main parameter defines the JVx application class. It's the inner class MyApplication, in our example.

It's that easy!

Application Coder

Post to Twitter

This article is about our new Research project: Application Coder.

The application is a very simple Java Code Editor. It shows a tree with Java files and has a Code Editor for modifying files. The code editor is the popular Ace Editor and we use Eclipse JDT for java compilation. The application was written as standard JVx application with vaadin UI.

We made tests as single-page vaadin application, embedded with iframes and embedded with div areas. All versions work without problems but the last one is preferred. If you embedd a vaadin application with divs, it's a little bit tricky because you have to configure the client-side on your own, but you get full access to the whole html page. This wouldn't be possible if you use an iframe because you can't access the main html page.

Some cool features of our editor are: error annotations and error markers. And the best feature is our LIVE Preview of code changes!

Annotations and marker

Annotations and marker


Watch this video:

Application coder


The application coder is not only a Java code editor, it was designed especially for JVx applications, because it groups client code and business logic. The preview also starts a JVx application and LIVE preview reloads the application.

Our coder application has a push mechanism and reloads every preview window automatically and immediate after compilation.

Above video shows multiple instances of our coder application, embedded in divs and it's possible to drag around and resize the applications (thanks to jquery-ui).

The application is a perfect showcase of JVx because it's not a database application as many other JVx applications. We've used vaadin, jquery-ui and vaadin-addons to create a great UX. JVx is technology independent and open for other technologies.

Style vaadin menubar popups

Post to Twitter

Our new application frame has two different vaadin menubars. One for the application menu and one as "hidden" menubar for the user options.

Here's a merged screenshot of both menubars:

Menubar popup

Menubar popup

As you can see, the left popup is dark and the right popup is bright. Usually, vaadin doesn't allow custom "additional" style definitions for menubar popups. The default style name is:

v-menubar-popup

The name is defined in the class com.vaadin.shared.ui.menubar.MenuBarState as

primaryStyleName = "v-menubar"

If you want to add an additional style class, it's simply not possible because the client implementation (com.vaadin.client.ui.VMenuBar) sets the primary style name, from the menubar, as style name for the popup overlay. The only thing you can do is, to change the primary stylename of your menubar. The problem is that you have to define all styles for the menubar-popup and menuitems, submenus, ... again in your style file. We tried it but it was horrible because you can't easily change the theme.

We tried to find a solution without changing vaadin client. We tried to set a primary style name as e.g. customstylepopup v-window-popup. But the client menu implementation has following code in getPrimaryStyleName:

protected static String getStylePrimaryName(Element elem) {
  String fullClassName = getStyleName(elem);

  // The primary style name is always the first token of the full CSS class
  // name. There can be no leading whitespace in the class name, so it's not
  // necessary to trim() it.
  int spaceIdx = fullClassName.indexOf(' ');
  if (spaceIdx >= 0) {
    return fullClassName.substring(0, spaceIdx);
  }
  return fullClassName;
}

(it makes sense to do this!)

So, it wasn't possible with simple tricks. The only solution - we found - was an extension of VMenuBar to override getPrimaryStyleName and return a custom style class concatenated with the default style class, something like this:

@Override
public String getStylePrimaryName()
{
    if (popupStyle != null)
    {
        return popupStyle + " " + super.getStylePrimaryName();
    }
       
    return super.getStylePrimaryName();
}

It wasn't our preferred solution, but we had another hack for setting ids of menu items. So it wasn't hard to add another hack for the style name :)

With our modification, it'll be possible to add custom style classes to menubar popups:

Custom style names

Custom style names

The source code is available on sourceforge.

From Swing to Vaadin?

Post to Twitter

Some days ago, vaadin released a Tutorial for Swing developers. It's a hitchhiker's guide to convert a Swing app to modern web app. It's a must-read if you plan to replace/migrate or modernize your Swing application.

We were mentioned in the last paragraph with our JVx framework, as possible conversion strategy. Thanks for that!

I want to hook in at this paragraph, because I totally agree with the rest of the tutorial.

It's true that a wrapper has pros, cons and limitations. You can't wrap everything. Sure you could try, but it needs so many developers and doesn't make sense because a wrapper shouldn't copy the underlying technology. The more features the wrapper has, the more problems will occur with new (different) technologies. The wrapper should be a subset of all technologies. But a subset is limited in functionality!

A wrapper should be focused on a specific domain, e.g. database/data driven applications or game development. A wrapper for multiple domains will fail!

I don't know many working wrappers. There were many attempts to create (UI) wrappers, in the past, but most were stopped because of complexity or the developers had other interests (if project was open source).

JVx is one working solution and in my opinion the most complete one because it contains an UI wrapper, has implementations for different technologies like Swing, JavaFX, Headless and Vaadin. The APIs are bulletproof and there are native applications for Android and iOS. JVx is a generic solution and doesn't generate additional source code.
But it has more than that, because it's full-stack and comes with different application frames for desktop, web and mobile applications.

But what is your advantage if you're using a wrapper?

You're (GUI) technology independent.

An example:

Your current business application is a Swing application and you plan the migration to a modern technology.
Your first migration decision should be: Desktop or Web
Next decision: which UI framework
Optional: Mobile support?

If your platform decision was: Desktop, then it's very simple to find the right UI framework: JavaFX and try JavaFXPorts for mobile support.
Fact: No real web and possible problems with mobile support

If your platform decision was: Web, then it's not an easy task to find the right UI framework, but vaadin should be the first choice because it's comparable to Swing and hides web technology for you!
Fact: No desktop but mobile support

Every decision has pros and cons. If you bet on one technology stack, you're fixed to this technology stack. In our example it was JavaFX or vaadin. And what will be the next preferred UI technology after JavaFX or vaadin?
You'll have the same problems again and it's never easy to migrate a (business) application.

You should bet on a technology independent solution, to be prepared for the future!
Means, you should use a wrapper. But don't use a wrapper which hides the technology from you. It should be possible to access the technology directly - if needed or if it's not important to be technology independent.

Sometimes it's not possible to be technology independent, e.g. some custom controls aren't available for all technologies.

The wrapper should allow technology dependent and independent development without any limitations!

Does it make sense to use the same application with different technologies?

Yes, but...

It's not a good idea to use e.g. Swing AND JavaFX because both technologies are desktop toolkits. But it makes sense to use JavaFX for your backend application and vaadin for your frontend or your mobile devices.

It's also a good idea to create only one application that works as desktop, web and mobile application - the same application. But show different screens/views on different platforms.

There's no big difference between desktop and a standard browser because resolution of a desktop pc is the same. A mobile browser has limited space and you shouldn't use the same screen/view on a mobile device as on the desktop pc.

Example
We use an ERP backend application to manage vacations. The appliation has about 10 screens for resource management, master data management and accounting. The application runs as desktop application with JavaFX. The same application runs in desktop browsers with 3 screens because the web frontend doesn't offer master data management and accounting. The same application runs on mobile devices with only 1 screen because mobile devices are used from employees to enter and view vacations.

We have only one application, started with different UI technologies and with different screens/views.

It would be possible to create 3 different applications with different screens and with dependencies between the applications and ... (complex to maintain, 3 different projects, application frame x 3).

If you'll create a "native" vaadin application and a JavaFX application you'll need different development teams with different know-how.

Don't waste time and resources, focus on the real problems of your application. A wrapper hides technology problems and allows fast development with few developers: Win-win situation!

Vaadin application frame design

Post to Twitter

Many of you probably know our application frame, for web applications: One application, different styles

We had a MDI variant for legacy applications with internal frames for screens. This is useful for applications with many screens and many records. The internal frames help to compare records from one screen with records from other screens.

For modern user interfaces, we use the SDI option. It looks fresh and simple and is great for applications with a small number of screens, because a scrolling menu is not nice - sure it will scroll if needed.

But what about application with e.g. 60 screens. Is MDI the only option? Yes and No, because we didn't have a variant for that kind of applications.... but now we have.

It's a mix of MDI and SDI, with a little bit CI:

Application frame SDI with menu

Application frame SDI with menu

The application frame is still the same as before, but with a different option. The name of the option is corporation, because the UI is meant for applications with many screens. It's still SDI with modal popups, because it shouldn't be too legacy!

Oh, do you see the small buttons on top/right? We use the FontAwesome feature of vaadin 7.2. We're already using vaadin 7.4.2 for our JVx vaadin UI!

We now have one application frame for many different application styles (I don't talk about visual attributes). The application frame is responsive and works well on mobile devices.

JVx and JavaFX (fasten your seat belts)

Post to Twitter

Today is a great day, because I'm able to write about our latest Research Project.
It has to do with JavaFX and JVx - what else.

If you remember one of my blog postings in 2013, you know that we tried to implement our JVx UI with JavaFX. Long story, short: Not possible with one master thesis.

But... we always try to reach our goals and it's an honor to tell you that we're back on the track.

Our JavaFX UI is one of the coolest things on this IT planet.

It's cool because of JavaFX. It's APIs are still not perfect and the overall performance could be better but it has so many awesome things. Two of them are styling via CSS and animations. JavaFX is the official replacement for Swing and it has potential to bring back "write once run anywhere". We don't know if JavaFX will be the holy grail for desktop application development but there's nothing comparable in the Java world.
To be honest, it's not a problem for us because our JVx is (UI) technology independent and if JavaFX won't do it, we'll implement the next UI toolkit for JVx.

... but back to the topic

You should already know that JVx is a Full-stack application framework for building database (ERP) software. It's GUI technology independent and we've different UI implementations. The most important ones are based on Swing and vaadin. We also have native Android and iOS support and our applications run on embedded devices as well. Our applications don't need an application server to work properly because we have a plain socket server as well (IoT ready).

Our preferred Java UI technology was and is still Swing, because it's feature rich and you can build every kind of application without bigger problems. The problem with swing is that it's old and not maintained from Oracle. Sure it will work with future Java releases without bigger problems (hopefully) but you can't use modern development techniques or create fancy and fresh UIs with less effort. A standard Swing application looks boring and cold. Sure, it's possible to pimp a swing application with lots of pictures or nice looking LaFs, but it's not fun.

Swing development is like software development in the stone-age, compared to state-of-the-art web development. If you do web development, you have CSS3 and HTML5 and some nice designed UI frameworks. I won't say that you're faster or that web development is better but it's easier to build nice looking UIs. I'm not a big fan of classical web development because most applications are horrible to maintain, there are soo many different frameworks and you can't be sure that your investment is future-proof because many frameworks don't survive the first year. But everything depends on knowledge and the right framework - for sure.

IMO it's better to use one framework that solves most of application relevant problems - if it exists. If we talk about web applications, I recommend vaadin because it's feature rich, just works, is very polished and is licensed under Apache 2.0. The community is active and great support is available!
Sure, there are some other frameworks like GXT or extJS from Sencha, but the license could be a problem if you build commercial applications. Don't forget RAP, but it's not trivial to start with the whole Eclipse ecosystem.

I'm talking about (enterprise) application development and not about webpage or webservice development. I would recommend pure HTML5 and CSS3 with one or two smaller javascript libs for developing fancy "webpages", or php instead of Java. If Java and "webpages", PrimeFaces is a good decision.

Soo, if Swing is "dead" and web development is hard, what should we use for (enterprise) application development?

If you develop for the desktop, try to start with JavaFX becasue it's maintained and has potential for mobile devices.
If you develop web applications, use vaadin.

(This is my personal opinion, mixed with my technology and project experience)

And why should you use JVx?

JVx bundles all together in one independent framework (it's still small and easy to learn). You'll be independent of the underlying UI technology. Write your application once and start it with different UI technologies or use it on different devices without problems. Your investment is future-proof. JVx is open source and licensed under Apache 2.0. JVx doesn't hide the underlying technology. It's always possible to use features from the technology if you want. If you use a UI without abstraction, you can't switch to another UI easily without rewriting most parts of the UI. If you start with e.g. JavaFX and if Oracle will abandon JavaFX, will you rewrite the application?

You can compare the concept of JVx with SLF4J, but with a different use-case: (enterprise) application development

And what about JavaFX?

I like JavaFX because of, IMO, 5 great things: scene graph, CSS styling, Animations, Web engine and the community. All other things are important and useful but not essential for real-world applications (vs. Swing). It's not possible to compare Swing with JavaFX because JavaFX was written from scratch and Swing was based on AWT. Sure, both toolkits are UI toolkits :) but with different concepts.

JavaFX bridges the gap between desktop and web development. It's still a toolkit for desktop applications but with most advantages of web development. It's possible to create fancy applications without headache. There are cool tools like Scene Builder or ScenicView. IDE support is getting better and developing controls is not too hard.

Sounds too good? JavaFX is on the right track and ready for real-world applications.

That was enough for us to start a JavaFX UI for JVx. It's the only framework which allows switching between Swing, Web, JavaFX or native mobile. Sounds unbelievable? Here's is our first video with our JavaFX implementation. It demonstrates the full power of JVx and JavaFX.

The video starts with our good old Swing UI. The same application will be shown as web application in a Chrome browser. And finally it will run as JavaFX application. You should carefully check the JavaFX application because it has an embedded browser that shows the same application, with a different UI technology - at the same time. Don't miss the animations :)

It's a world premiere!

JVx and JavaFX



We know that the styling is not really cool but we're currently in the development phase and aren't finished... be tolerant :)

IoT: Window shutter control with JVx, Vaadin and RaspberryPi

Post to Twitter

It's about two years ago that I wrote about controlling window shutters. The system run for 2 years without problems but it was time to change "something" because it was vacation time and I thought it would be useful to showcase our vaadin UI :)

The old system was controled via (good old) SMS - without GUI. It had a lot of commands but only en/disable up/down were heavily used because the integrated public holiday and weekend automatism solved most use cases. The SMS contol was great and worked without problems but it was a problem if you didn't know the commands and syntax by heart!

Another consideration was a simple real-world showcase for our vaadin UI, the responsive feature and push support. Our application frame was already responsive and worked on desktop, tablet and smartphone browsers. I thought about using AngularJS or different "hypeware" but I didn't have weeks to research and implement. My new GUI had to be finished in few hours. And it should be maintainable for the next two (or more) years without problems.

I love researching hypeware... but simply read some articles if you want:

There are too many frameworks, everyone is different - no real standard APIs (didn't count HTML and CSS). You can't be sure that the framework is future-proof because developers love to change software to be incompatible with previous versions just to be state-of-the-art.
And why should a developer use 5 or more frameworks to create a simple application for desktop and mobile? You'll need more libs if your application needs DB and CRUD support.
C'mon we need simple and small frameworks and not what we have right now! The world is talking about microservices and what about "micro frameworks"?

Conclusio: I didn't find something comparable to JVx. Of course, it's my opinion and I'm JVx developer, but I'm a researcher and the technology isn't a problem if it makes sense to use it.

To be honest: Our Vaadin UI in combination with other JVx features is unbeatable for most application types. Of course, it's not recommended for web pages with custom layout requirements. If you need an "application style" solution with menu and screen/work area or form based - there's nothing comparable.

After I knew that my new GUI would be a JVx' vaadin solution, I had to think about the details. The whole application should be responsive to support every modern device and desktop browsers as well. The application server should be Jetty or Tomcat and it should run on the RasPi. The performance is not critical because the hardware is limited (but cheap).

Software stack: clear
Hardware: clear
Deployment type: clear

Next step was development. This was easy because I re-used existing classes and added some get/set methods to access internal members. The most important change was an event listener because every vaadin client should be updated immediate via push. It wasn't possible without notification mechanism. I thought about an event bus and played around with mbassador. It's a nice and small lib, fast and just works (recommendable).
To be honest: I didn't need it because I didn't need more than one listener and an event bus is not much more than a listener abstraction (with some extras).

After some hours developing and refactoring, everything was ready for a test.... The next decission: Jetty or Tomcat?
I worked with Tomcat for many years and didn't know much about Jetty, only that it's the number one for embedding into applications, supports SPDY and has websocket support. I played around with Jetty and got it running. The installation of my GUI worked like a charme and everything worked except push. I found solutions for embedded run mode but not for standalone Jetty. I'm pretty sure that I missed something, but I wasn't very patient and switched to Tomcat 8.
Not only because of my push problem but Jetty didn't handle requests as expected:

http://localhost/house/ didn't redirect to http://localhost/house/index.html automatically. I didn't have an index.html file in the public file area, anyway the request should be executed. Such requests work with Tomcat without additional configuration.

I'm not happy that my Jetty server didn't work but I'll retry it in the next days :)

Here's a simple screencast of my solution because it's easier to understand:

Window shutter web control


It took me less than one day to finish my GUI and the best: It's family safe.
I tried the GUI with PS4 and WiiU integrated browser and it was great to see shutters moving.

Summary

I've used a RaspberryPi, installed Tomcat 8 with Java8 and deployed my application as war file. The GUI is responsive and supports push via websockets and runs on every device in my house. It wasn't planned using gaming consoles :)

Oh, I forgot to mention: During my tests I found out that Android and iOS support custom Bookmark images for Desktop links but this needs additional tags in index.html (works without, but it's better to include the tags). Our VaadinUI now has built-in support for such images :)

The switch component is available in the Vaadin Directory.

Stress testing a Vaadin application with Selenium Grid, a summary

Post to Twitter

For quite some time there has been the idea in our heads to run a stress test against a JVx/Vaadin installation to see if there are any major problems with performance and multithreading. We knew that Tomcat, Vaadin and JVx could serve a lot of users simultaneously, that's what all three were designed to do, that's what all three do well in many different environments, but we never really tried to stress it out. So we set out to change that.

The plan

Some weeks ago we started discussing and outlining how to perform such a stress test. After some research we came up with a plan:

  1. Design a test case using Selenium for the DemoERP application
  2. If necessary, design and create solutions to allow that
  3. Set up a Selenium Grid which can be used for testing
  4. Perform the actual test and fix anything that needs to be fixed

As simple as it sounds, it had quite some gotchas on the way and taught us some valuable lessons.

Name all the things

The first thing you need for automated testing are names, components need names so that they are easily identifiable by whatever you're using for testing. Already back in September we started to name components in a unique and easily recognizable fashion, our efforts are covered by ticket #1103. Of course you can still set your own names, if you want that, as the default names will only be applied if no other name is set.

In short, components are now receiving a name based on their position in the component tree. For example take this simple workscreen:

Class                 Name

WorkScreen:           WS-XX
    SplitPanel:       WS-XX_SP1
        Panel:        WS-XX_SP1_P1
            Label:    WS-XX_SP1_P1_L1
            Editor:   WS-XX_E_COLUMN
        Panel:        WS-XX_SP1_P2
            Label:    WS-XX_SP1_P2_L1
            Button:   WS-XX_B_DOPRINT

The naming starts with the "root" component and every child component does append it's name. So you can easily guess where a component is assigned to, simply from its name. Every name is guaranteed to be unique, either by its position in the tree or by incrementing a counter.

Additionally there are "special snowflakes" which are getting better names, like the name of a button which is only prefixed with the name of the workscreen and a "B" and then followed by the actions which are assigned to it, e.g.

public class DashboardWorkScreen extends ...
{
    ...
    public void initializeUI()
    {
        ...
        UIButton butUp = new UIButton("UP");
        butUp.setSize(80, 80);
        butUp.setStyle(new Style("up"));
        butUp.eventAction().addListener(this, "doUp");
    }
}

The button "butUp" will get the name Das-WV_B_DOUP. The prefix Das is the short name for the DashboardWorkScreen, followed by -WV which is like a prefix checksum/number. The letter B stands for Button and the action name DOUP is the last part.

With such a naming scheme in place, it's easy to create automated UI tests as every component is uniquely identifiable in the component tree. In the case of Vaadin the names are also used as ID's on the elements

<button type="button" class="v-nativebutton v-widget up v-nativebutton-up daswv_b_doup v-nativebutton-daswv_b_doup default-padding v-nativebutton-default-padding v-has-width v-has-height" id="Das-WV_B_DOUP" tabindex="0" style="width: 80px; height: 80px;">
  <span class="v-nativebutton-caption">UP</span>
</button>

The test case

For everyone who does not know our DemoERP application, it is a simple ERP application which features customer, product, offer and order management together with a nice statistic screen. All this can either be run as Swing application or (powered by a Tomcat Server) as Vaadin application directly in the browser.

The test case that we designed is rather simple:

  1. Load the application website and perform the initial login
  2. Open the customer management screen and add a new customer
  3. Open the offer management screen and search for an offer
  4. Switch to the order of said offer and edit it
  5. Open the statistic screen
  6. Logout

This gives a nice workload for testing, as it touches close to all areas of the application, changes values nearly simultaneous and also adds data with every run. Also note that the testcase would perform these steps as fast as possible, without any artificial waiting in between.

Thanks to the Selenium IDE the test case was recorded within a matter of minutes, and quite fast copied into a unit test. The first round of testing happened with the Chromium WebDriver, which allowed us to watch the test case do its work and make tweaks wherever necessary to allow the test to succeed.

Setting up Selenium Grid

Selenium Grid (or Grid2) is a simple, scalable framework for distributed Selenium tests, allowing to control multiple browsers on different platforms. The central server is called a "hub" and the clients are called "nodes". The nodes are registering themselves at the hub, and from there a client (for example a unit test) can send a request to the hub and it will receive a node matching the requested requirements. On the client side this is happening by simply creating a new instance of a RemoteWebDriver provided by Selenium. It will automatically contact the hub, wait for a node to be ready for it and then forward all commands to that node.

This is useful if you want quickly test an application in multiple browsers on multiple platforms without leaving your IDE or to perform a simple stress test.

Setting it up is as easy as starting the hub on one machine, and starting the nodes on any other machines. We decided to go with PhantomJS as browser, as we intended to run many instances of said browser on one node and we also had a wide variety of machines, some without a GUI. PhantomJS does already have it's own built-in node, so we just needed to start PhantomJS on the machines and tell it where it can find the hub.

At the end of day we had 12 physical machines in place and we maxed them out to get a total of 82 nodes which can be used for our stress testing.

Sessions

One thing we noticed very quickly was that lingering sessions can be a real issue and the timeouts really need to be tailored to your environment. While testing we quickly noticed that sessions kept lingering on, this was less an issue for the server and Tomcat, as more for our MariaDB installation because at some point it simply stopped accepting new connections. So we needed to reduce the amount of time a session would linger on to not clog our test server.

There are two important settings controlling the amount of time a session will linger on in such a setup:

  1. The timeout setting of the Tomcat server
  2. The heartbeat interval and closing of idle sessions in Vaadin

As said, these settings should be tailored to whatever environment the server is running in. In our test environment it didn't make any sense to keep sessions longer than a few minutes, because if the test case ended or was interrupted for some reason, the session would not resume. Armed with that knowledge we reduced the amount of time a session could linger on and we could finally start the real test.

Running the test

The server in our case would be a rather unspectacular laptop, with a Core-i7 (1.80GHz) and 16 GB of RAM. Tomcat 6 would deliver the application and the data would be provided by a local MariaDB/MySQL database (with an InnoDB engine).

The test itself, which ran several times for several hours, was rather boring to be honest. 82 clients swarmed the server as fast as it could answer requests without any noticeable effect on it's performance. The laptop, despite not being the typical server machine and for sure not intended as such, could easily handle the load that 82 clients produced. We did expect the application to stay in a usable state, however we did not expect that it wouldn't have any noticeable effect at all from a users perspective.

Here is a short summary of the data we gathered (using jvisualvm and ProcessHacker):

  • CPU peaked at 60%, but stood most of the time between 40% and 50%
  • Tomcat peaked at 4GB RAM, but would have been able to work with as little as 2GB
  • Over 21.000 datasets were created during the test run
  • The GUI was not noticeably slower during the test

Our setup could handle that amount of requests with ease. Unfortunately we could not get a hold of more machines to add further nodes for now.

We broke it, we fixed it

During the tests two nasty looking issues stood out, which kept happening in roughly 1% of all test runs with seemingly no pattern behind them.

The first was a simple NullPointerException which kept happening in UIImage, but that was fixed rather easily as certain methods where not synchronized and so destined to fail.

The second was a set of Exceptions which send us to our toes: NullPointerExceptions, "Object 'offer' was not found" and "Remote storage returned 15 value(s) but 22 were expected". Clearly there was something very wrong, and with the low volume with which these exceptions occurred, it could only be a race condition somewhere. Looking for possible causes send us to all corners of the codebase without any clue in sight what might cause them.

It only dawned upon us when we got a hold of the map which did not contain the object "offer", it only contained the objects "v_statistic_order_offer_year" and "v_statistic_order_offer_month". Looking at the lifecycle objects immediately showed the problem, the session did have a completely different lifecycle object. That did also explain the wrong amount of values, as two lifecycle objects happened to contain identical named objects. But how was that possible? We immediately checked the complete source code associated with creating a session, and in AbstractSession we found the culprit. Even though the complete code for creating a session was properly synchronized where necessary, one simple statement had slipped past:

private static long lSessionCount = 0;
...
private final Long lObjectId = Long.valueOf(++lSessionCount);

For everyone not seeing the problem, here is a simplified explanation. If an object is instantiated, the fields are initialized even before the constructor is run, this is not and can't be directly synchronized. A simple increment operation (++) is compiled to six operations:

  1. Load a reference to a variable onto the stack
  2. Put into that reference the value of the static field (in our case lSessionCount)
  3. Put the value "1" onto the stack
  4. Add both values on the stack together
  5. Duplicate the values
  6. Put the new value into the static field

Six instructions, a lot of time for two threads to overtake each other and cause problems. Instantiating two sessions at once had the potential to give both the same ID. Of course we immediately fixed it and applied some other optimizations.

Another run verified that this had fixed all three issues at once and the tests were running without any further incidents for hours.

Conclusion

We were not even close to pushing the limits of what a JVx application/server can handle, and judging from the performance of the server I can only call our test setup "humble". Still it gave us a good measure of what is for sure possible and showed us that we need a bigger setup to really push the limits.

During the tests we also created various utilities to aid us with testing, unfortunately they are in a rather barebone state and can not be released by now. Also there is now a more important project awaiting, one that we think you've been waiting for for far too long.