Netuality

It you have ever written even the smallest standalone app using SWT and Jface library, then you must know (and hate!) org.eclipse.core.internal.runtime.AssertionFailedException. This exception has the very bad habit of substituting the initial one without keeping any trace of it. The official explanation is the following :

“Jface is not guaranteed to work in standalone mode. As a matter of fact, Jface tries to report different errors using the Eclipse plugin mechanism, thus provoking a fatal assertion failure which is the only exception being reported.”

Ryan Lowe suggests wrapping [the code] in a try/catch block temporarily. Yeah sure this might just work if you know WHAT to wrap. What do you do when, after multiple commits and updates of the same source file modified by different people, you have the failed assertion ? Wrap in a try/catch every slice of code committed in the last few hours ? Hmmm … don't think so.

An interesting solution would be (as proposed by Ryan) to use aspects in order to wrap automatically each and every SWT call. But this might bring serious performance issues and I also have some doubts that it will catch everything.

But boy, aren't we lucky that Eclipse is opensourced ? Just by looking at the stack trace (which is the same over and over no matter where your real error is) you can spot the guilty : org.eclipse.core.internal.runtime.InternalPlatform (found in runtime.jar/runtimesrc.zip). The little bugger is Mr. private static boolean initialized. Putting it on true involves starting the platform, which is a quite complex process judging by the loaderStartup(URL[] pluginPath, String locationString, Properties bootOptions, String[] args, Runnable handler) throws CoreException method. Unfortunately, being a private variable means that reflection won't work and dynamic proxying won't work, either*. AKA Out Of Ideas (TM)

So, it's time to stop trying to be smart. I'll just be a rude dumb programmer and simply modify the source, recompile and replace the .class files in the runtime.jar. It's extremely simple yet sharply efficient:

• initialize “initialized” on true
• in the method private static void handleException(ISafeRunnable code, Throwable e) replace all the content with the following line : e.printStackTrace();code.handleException(e);. A stack in the console is enough for me at this stage, but of course you may use your favorite logging infrastructure classes to report the exception. The code.handleException call has (at least in my app) the effect of displaying an innocuous dialog box telling the user that something went wrong. Just the right dose of details aka nothing at all (don't scare the poor user, please).

Well, what's more to say ? This stupid trick simply works.

On second thought, I might use RCP. Maybe, in a future episode.

*AFAIK Please correct me if I'm mistaking here.

PS That was quick “Setting the accessible flag in a reflected object of java.lang.reflect.AccessibleObject class allows reading/modifying a private variable that normally wouldn't have permission to. However, the access check can only be supressed if approved by installed SecurityManager.”. Humm, ok, so there IS a smarter solution after all. Left as an exercise for the reader

According to David Heinemeier Hansson, the RAILS 1KLOC (!) web framework written in Ruby was used to develop Basecamp, a mildly complex project management webapp. What's fascinating is the Ruby code used to develop Basecamp is only 4KLOC (according to the RAILS document) and was developed in less than 2man*month (including 212 test cases and all the bells and gingles). Although I strongly suspect that there was a lot of templating going on behind the hoods and I doubt that template development time was included, the efficiency is amazing, especially if you consider that they have started almost from scratch. Makes you wonder what would be possible to perform in Ruby with a comprehensive library such as PHP's PEAR (well, duh, ignoring the fact that it should be called REAR which is a very very very nasty name) ?

My first contact with Tapestry was more than 18 months ago. Back then, I was interested to find a web framework for integration with our custom Avalon-based (using the now-obsoleted) Phoenix server*. The web interface was ment to be backoffice stuff, for simple administration tasks as well as statistics reports. Given that the data access and bussines logic were already developed, we were looking for something simple to plug into a no-frills servlet container such as Jetty. we managed very easily to integrate Jetty as a Phoenix service and pass data through the engine context. But when we finally integrated Tapestry [into Jetty [inside Phoenix]] and make it display some aggregated statistics, the project funding was cut and the startup went south. But, that’s another story and rather uninteresting one.

Meanwhile, things have changed a bit. Tapestry had become a firsthand Apache Jakarta project, the Tapestry users list is more and more crowded, and again I see it used in my day work (by Teodor Danciu, one of my coworkers and incidentally author of Jasper Reports) and doing some moonlighting by myself for an older web project idea. And there is exceptional Eclipse support via Spindle plugin. While the ‘buzzword impact’ on Tapestry on a Java developer CV doesn’t yet measure up with Struts, this framework has obviously gained a lot of attention lately.

So, what’s so special about it ? If I’d have to choose only one small phrase I’d quote Howard Lewis Ship, Tapestry lead developer, from the preface of his book ‘Tapestry in Action’:

The central goal of Tapestry is to make the easiest choice the correct choice.

In my opinion this is the weight conceptual center of the framework. Everything, from the template system which has only the bare minimum scripting power, passing through the componentized model, up to the precise detailed error-reporting (quite unique feature in the opensource frameworks world) gently pushes you (the developer) to Do The Right Thing. To: put logic where it belongs (classes not templates), organize repetitive code in components, ignore the HTTP plumbing and use a real, consistent, MVC model in your apps (forms are readable and writable components of your pages). You don’t need to be Harry Tuttle to make a good Tapestry webapp, just a decent Java developer is enough. That’s more than I can tell about Struts …

Coming from a classic JSP-based webapp world, Tapestry is really a culture shock. The most appropriate way to visualise the difference is to imagine a pure C programmer abruptly passing to C++, into the objects world**. For a while, he will try to emulate the ‘old’ way of work, but soon enough he’ll give up and start coding his own classes. However, this C programmer will have to make some serious efforts, not necessarily because OOP is hard to leard, but in order to break his/her old habits.

“Tapestry in Action” is your exit route from the ugly world of HTTP stateless pages and spaghetti HTML intertwingled with Java code and various macros. It’ one of the best JSP detoxification pills available on the market right now.

The first part of the book (‘Using basic Tapestry components’) is nothing to brag about. It’s basically an updated and nicely organized version of the various tutorials already available via the Tapestry site, excepting probably some sections in chapter 5 (‘Form input validation’). By the way, the chapter 5 is freely downloadable on the Manning site and is a perfect read if you want a glimpse of the fundamental differences between Tapestry and a classic web framework (form validation being an essential part of any dynamic site). However, if you want to go over the ‘Hangman’*** phase you really need to dig into the next two book sections.

The second section ‘Creating Tapestry components’ is less covered by the documentation and tutorials. I’m specifically pointing here to the subsections ‘Tapestry under the hood’ (juicy details about pages and components lifecycle) and ‘Advanced techniques’ (there’s even an ‘Integrating with JSP’ chapter !). While it is true that any point from this chapter will generally be revealed by a search on Tapestry user list or (if you’re patient) by a kind soul answering your question on this same list, it’s nethertheless a good thing to have all the answers nicely organized on the corner of your desk.

The third and last chapter (‘Building complete Tapestry application’) is a complete novelty for Tapestry fans. It’s basically a thorough description of how to build a web application (a ‘virtual library’) from scratch using Tapestry. While the Jboss-EJB combination chosen by the author is not exactly my cup of tea (I’m rather into the Jetty+Picocontainer+Hibernate stuff) I can understand the strong appeal that it is suppposed to have among the J2EE jocks. Anyway, given the componentized nature of Tapestry, I should be able to migrate it relatively easily if I feel the need for it. The example app is contained in a hefty 1Meg downloadable archive of sources, build and deployment scripts included.

To conclude, ‘Tapestry in Action’ is a great book about how to change the way you are developing web applications. The steep learning courve is a little price to pay for a two or three-fold improvement in overall productivity. And this book should get you started really quick.

*Which AFAIK is still used at our former customer.
**There were some posts on Tapestry user list on about a certain conceptual ressemblance with Apple’s WebObjects. I can’t really pronounce upon this because I do not know WebObjects, but the name in itself is an interesting clue.
***’Hangman’ is the Tapestry ‘Petshop’ (although there is also a ‘real’ Tapestry Petshop referenced in the Wiki).

One of my current responsibilities is to maintain an internally developed plugin, used by various members of the team to generate code from the analysis model. As far as I can tell by the webstats of the update site, every version is downloaded by 18 people, a small but heterogeneous user base.

My biggest problem is the Eclipse version. The analysts are not exactly Java geeks waiting anxiously for nightly builds of Eclipse, they use a 'standard' 2.1.2, mainly because it's stable and well internationalized. Things go wilder in the programmers team : versions ranging from conservative (2.1.x) to liberal (3.0M4) and even the occasional dumbass with the latest integration build (that would be me, of course).

The 'enhanced binary compatibility' in 3.0M7 came as a relief, diminishing the need to switch between Eclipse versions in order to develop the plugin or work on other tasks. Well, I still have to briefly test the damn thing on Eclipse 2.1.x before releases. However, running simultaneously two or three Eclipse instances is no piece of cake for my 512Mb laptop (I still haven't found who I have to kill here in order to be awarded a memory upgrade). Unfortunately, checking out the plugin source into M7 has shown the invisible ugly face of 'binary compatibility': the plugin doesn't compile.

There are just a handful of lines of code, some emphasizing differences in Eclipse API which are somehow hidden in 'compatibility mode', some effectively showing small bugs in plugin behavior. But the real issue here is that I cannot really develop the plugin in M7 until I manage somehow to compile it, while not losing downwards compatibility.

Let's dissect one of the compilation issues. The bummer concerns automatic opening of an editor (or focus if already opened) when clicking on its reference (somehow similar to what happens when you Ctrl+click on a class name in JDT). In the older API it was a question of page.openEditor(file); where page is a IWorkbenchPage and file is an IFile. This simple stuff worked well until 3.0M4, then (M5) things changed to page.openEditor(new org.eclipse.ui.part.FileEditorInput(file),editorId); where FileEditorInput implements (among others) an IEditorInput. While this is certainly nice because you may directly link editors to something other than files***, obviously the old code does not compile under M7.

Maintaining different projects for 'old' and 'new' style projects for 10 or so lines of code is obviously overkill. Second solution – via reflection, but it would mean more than few lines of code and the result would not exactly be comprehensible nor maintainable. Only thing left : use a scripting language.

Of course I could have taken any decent scripting language embedded in Java. Decision to go with Groovy was taken mainly because of its coolness factor, but I am sure the idea will apply easily with Jython (big favorite of mine) or the performance-aware Pnuts, for instance.

In a nutshell, you have to execute a line of code depending of the current Eclipse version (it's a little bit trickier, but we'll discuss later about it).

groovy.lang.Binding binding = new Binding();

binding.setVariable("page", page);

binding.setVariable("file", file);

groovy.lang.GroovyShell groovyShell = new GroovyShell(getClass().getClassLoader(), binding);

if (newPlatform)

{

	return groovyShell.evaluate("page.openEditor(new org.eclipse.ui.part.FileEditorInput(file),editorId);", someExpressionId);

}

else

{

	return groovyShell.evaluate("page.openEditor(file);", someExpressionId);

}

It's basically a vanilla flavored ripoff of the Groovy embedding example from the docs. The boring part : caching the binding, hiding everything behind a nice facade, is left as an exercise for the [interested] reader. Remember to pass the classLoader of the current class, do not create a GroovyClassLoader out of nowhere or you'll end up dealing with Eclipse own class loader, which means trouble for simple tasks like these.

How do we know that the Eclipse version is the 'new' or the 'old' one is not that simple because remember : 'old' means 2.x up to 3.0M4. So finding out Eclipse SDK version is not enough, you have to find out another discriminant which in our case is the 'org.eclipse.ui.ide' plugin. Result:

boolean newPlatform;

//find out if we are inside a new or an old platform

PluginVersionIdentifier pvi = Platform.getPluginRegistry().getPluginDescriptor("org.eclipse.platform").getVersionIdentifier();

newPlatform = pvi.getMajorComponent() >= 3 && Platform.getPluginRegistry().getPluginDescriptor("org.eclipse.ui.ide") != null;

No, we are not ready to deploy yet. A small trick has to be performed or the plugin won't start under older versions of Eclipse. We had to add some new plugins to dependencies (in the pligin descriptor) such as the aforementioned 'org.eclipse.ui.ide', obviously the older versions of Eclipse will not find it, hence block our plugin activation on startup. In order to overcome this, you have to add (by hand !) a lesser known attribute ('optional') in the corresponding tag from the plugin.xml file : <import plugin=”org.eclipse.ui.ide”/> becomes <import plugin=”org.eclipse.ui.ide” optional=”true”/>. Now, the plugin is ready to be deployed.

For those brave enough to dare distributing such a plugin via an update site remember to 'cheat' by not allowing new plugins such as 'org.eclipse.ui.ide' in the feature.xml file (again, delete by hand). The 'optional' attribute doesn't help in this case. Go figure…

I hope that some of you will find useful this recipe for maintaining compatibility between different Eclipse versions with minimum of fuss. However, please note the specific prerequisites for this type of solution :

• there are only simple 'few-lines' modifications
• the code is not expected to evolve a lot in the 'affected' areas
• the evaluated code is not in a performance-sensitive area

***Interesting enough, this was one of the reasons I recommended against adopting RCP in one of our apps, a few weeks ago. It's nice to see that – now – the mechanism linking editors and resources is MUCH more flexible. Anyway, this won't probably change the decision of not using RCP because the main issue it's the volume of code we have to change. Development of one of the app modules started almost a year ago and the animal it's already sold and deployed on different production sites: upgrading would be a real nightmare. Maintaining a fork of the app is not an option either. Well, I guess we'll just have to cope with 'plain old' Jface and SWT.

PS After some days of 'silence', I have noticed from the logs that most popular posts on my blog are those concerning Eclipse plugins and Manning books (I seem to have a nice Google ranking on these topics). So, expect more of these (I am reading the MEAP of 'Tapestry in action' – a review should be up shortly).

It happened also on older versions but it does happen more frequently on the “final” 2.1 version. FYI : Gentoo Linux, Eclipse gtk, seems to be related somehow with bug 33138 (don't have the time to dig further).
Sometimes the monster simply hangs during a [take your pick : refactoring, new class generation] with an empty progress bar in the dialog box and a completely useless “Cancel” button. Been there, done that : kill -9 …
Then, trying to re-start Eclipse leads to a deadlock – while recovering workspace : dialog box, empty progress bar and useless “Cancel”. Freezed !
I have a lot of settings and projects so deleting the whole .metadata directory is just too painful. Therefore, I had to find out a smaller workaround : just delete the file .metadata/.plugins/org.eclipse.ui.workbench/workbench.xml and Eclipse restarts with a clean workspace. Some adjustments are lost but hey – my metadata is still there.