Being able to develop handheld applications with the spring approach, using techniques like dependency injection also for the JavaMe environment, looks promising.

I’m curious if and when approaches like this gain more momentum, it could be important for the development of mobile applications. I can imagine that this and Spring Roo together could make a powerful duo for the mobile app world.

I have to fiddle around with webservices during my work, we use JAX WS for generating the glue code. I wanted to create a HelloWorld example (again), only to realize that everything is already online (again).

I want to combine maven and JAX-WS. I’ve found the web page for Jax WS here, and more importantly, an example pom which contains everything you need for jaxws here.

One problem of the average Bob D. Veloper is that information about a specific problem may be available, but instead of solving your specific problem it raises again more questions. For example, the referenced pom above uses the antrun plugin to download a wsdl, which is no problem at all – if you look at the code and uncomment the lines where the wsdl is retrieved from the web. In order to make it work you should consider to remove the comment in the antrun configuration section.

Another problem is that you should know how to invoke maven properly in order to generate your artifacts. As always, it is easier as I thought initially, a simple

mvn package

or

mvn compile

first downloads the wsdl and secondly creates your webservice clients in the target directory.

As you surely know, it is best practice to generate the glue code from the wsdl; by placing the classes in the target directory, you are typically not tempted to change them (for whatever reason). As always, modify the source, not the artifacts. DRY you know.

Anyway, with a few commands you’ll get a ready to go implementation for an arbitrary webservice, both client and server implementation is generated. The only thing you now have to do is to fill out parameters and hope for the best.

There are many websites which provide an api for their services – if a wsdl is available, you are ready to go.

Another acronym popping up “recently” (first draft 2005   is WADL. Oh man.

O3D is an open-source JavaScript API for creating interactive 3D graphics applications that run in a browser window—games, ads, 3D model viewers, product demos, virtual worlds.

….

At last! A major player like google tries to resurrect the idea of 3D in the web. Aeons ago I did some VRMLing which was a very fustrating experience; (for me, not my clients ) Of course it supported javascript then too. It was a nightmare.

Anyway, maybe O3D does it. Several companies tried to establish a de facto standard for Web and 3D. I have the impression that none succeeded really.

At least google shows again a strong commitment to push the browser as platform and the importance of their Javascript engine which empowers Chome.

What is still missing (maybe someone can correct me here) is a standard API for audio. This will then eliminate Flash and establish even more Javascript as a systems language. In the meantime, it is still necessary to do some tricks like wrapping flash to acheive sound effects, native support would be great.

Recently SpringSource released an interesting … hm … what is it?

Roo?

In short, its a bootstrapping device for new projects and a code generator for common tasks, tailored for the spring stack.

(and surely much more I’m currently not aware of)

As far as I see, it is a missing link for Spring development. Most developers are very lazy and the barrier to plunge into new technologies is quite high if the given introductory example  doesn’t work out of the box.  Another problem domain tackled by ROO is the boring stuff which has to be done frequently – like adding Controllers, Views to some web project, creating domain objects and so on.

Everybody has his tricks, ranging from keyboard shortcuts to code templates. Roo looks like a powerful tool designed for boring tasks. Or rather, to avoid boring tasks.

Ok folks, stop writing abstractions which reach the end of their lifecycle before the next homemade nullpointer, don’t hassle with your favorite code generator – Let others do the dirty work. Concentrate on your business.

Scaffolding, prototyping, being productive … what else do you want from an average developer, always underpaid? Of course a tool which does “everything” for you.

ROO is something like that, well almost. It won’t get paid (that is your job) and it is even opensource!

It even generates some weired AspectJ stuff!, well don’t mind about that. Aspect oriented programming doesn’t get the momentum it should, maybe because programmers are just too dumb for using it?

Part of the problem is that an average Java developer not only has to master the language itself, which got considerably more complex since the addition of generics, but even more has to integrate lots of libraries in order to get to a simple CRUD application. The freedom of choice is not a bad thing per se, it can lead to considerable fustration sometimes on the other hand.

Roo is clearly aimed at the ‘experienced’ hacker who remembers the time of command line applications, seems like they’ve invested some thought into what developers really need and really like. At first glance, skimming through readme and examples, it looks quite promising. By exploiting its scripting functionalities, we’ll soon see many Meta Meta Meta Generators. I’m curious.

Alors, go and try it yourself.

Disclaimer: Using ROO won’t save you from not needing to understand the involved technologies.

Here is an eclipse workspace (zipped) to jumpstart in with maven. Very rough but works for me!

The XSLT approach had some drawbacks, however – some of them were a nuisance for me, therefore I’ve rewrote the functionality in Java.

This led to a little parser for freemind files, which could serve also for other use cases. My first thought was to reuse the java implementation, but I gave up after skimmig through the code – as much as i like the end functionality of freemind, I couldn’t really embrace the internals of the freemind implementation – well maybe in the meantime it got better.

Anyway, latexbeamer is a powerful tool to create nice presentations, and the source language is quite simple and easy to understand. Moreover, presentations and Mindmaps match nicely together since presentations should be very easy – less is more in this case.

In contrast to the XSLT approach here and here the Java solution supports more features as listed below:

• Support for highlighting of text (bold/italic/red)
• Support for boxes
• Pause support (elements are shown one after another on the same slide)
• Different themes

In the source code you’ll find experimental support for tikz/pgf generation too. I’ve discovered this very nice technology only recently, it has a nice mindmap generation support and I think Freemind would benefit greatly by exporting also in this format. Have a look here what the mindmap module it is able to do.

Below is an example for the latex beamer and tikz output.

Here is the source. (LGPL)

At the moment, it is a command line application, which is fine with me. I plan to write a GUI using JavaFX, so stay tuned.

….

Ps: Just discovered the project JCommons which implements utility methods for mindmap manipulation. Looks interesting! Although I don’t know why it is called JCommons.

Btw: The original XSLT approach was pimped. Cool.

*rant start*

• I see a .classpath, .project file, which tells me that eclipse should be used.
• Well, they should use also subversion, not cvs
• an ant script is here however … on command line, use ‘ant -p’ to show the most important build options. An alternative is to use eclipse and its ant integration -  Robert J. Alexander has some tipps (remarkably old anyway)
• the main directory is cluttered with various resource files – the packaging could use some brush – up.
• A quick try yields a successful build, creating a dist dir with all compiled classes. So far so good. Anyway, I would use maven for the job, just because it is cooler.
• In the ant project help the main goal for running is missing – use ant runWithoutPreferences . It working out of he box for me!  Thats great. Typically that is never the case
• a freemind.bat file is here, this should start freemind without 3rd party tools. Doesn’t work for me.
• OUCH! just discovered that the beformentioned ant runWithoutPreferences creates a directory directly in my eclipse workspace. thats not so good. The paths should be corrected accordingly in build.xml. Those values should be put in a separate property file.
• Consequently, the freemind.bat also fails.
• For my taste, freemind.bat should be replaced by a maven exec command – this could reduce the need for a bat and a shellscript file.
• …
• …
• wtf???

*rant end since …*

Well, I could complain about various things, but it would lead exactly to NOWHERE. Much more productive is to submit patches, as suggested on the project site. Again. Just do it.

Ok, so what is necessary to contribute to freemind? The well structured wiki is certainly a strength of the freemind development. It is very end user friendly and really motivates to contribute. This is one aspect which is often under estimated by open source projects in my view. Whoever is responsible for the nice freemind website, kudos for him.

…

ok … after some surfing, I discovered that one of the main developer did a fork on freemind, called freeplane. Obviously he got tired of the codebase and wants to make a complete refactoring, mentioned OSGi and stuff … will be interesting to see where this development is going to – end.

As a side effect, it also prooves my theory that the code is at is end of life cycle and it needs a major effort to get it maintainable again. Qed? I would very much regret if freemind development would stop or the core team breaks apart.

Remember: Technological problems can be always repaired, the important thing is that the involved team shares a common vision.

Didn’t try the mentioned  XMind yet. Dick Wall did, however . As an end user, I’m very happy with freemind. Seen from a developer’s perspective, I’m not. But if I’m not contributing, I have to live with it. I have also to accept if freemind development won’t continue. I would very much regret that. But do i regret it so much as to invest time and energy?

This is the key question for open source projects – bottom line.

Freemind is an example for a product which gives a high value to end users, but looking at the code is a pain and relatively isolated. It is sexier to contribute to an apache project, use a cool new language or some kind of middleware; is it of more value to learn EJB 3.1 or struggle with ‘customers’ (end users) who are ignorant of  architectual problems in the code? Whoa. What a dejá vue. It reminds me of the problems I face on a daily basis, just like every sw engineer. I don’t want to cope with such problems also in my spare time.

Alas!

Open Source Projects are hard work, the people involved just do it for the respect of the users. For their own satisfaction. Much of the motivation is described by the thoughts of this man. Remember: everything is better than to watch tv and waste your time. Everybody can contribute, no matter which skills. Energy is a valuable thing. Think twice before you invest it.

Coming back to freemind vs. freeplane – I don’t know what I should think about it. The vision sounds very ambitious, if not too abstract? Dimitry selected his main points for refactoring freemind here. I miss very basic things like source code management, continuous build system, tests … to start with. It is essential to give new developers a jump start into the project, this means that checking out provides all things in order to get the tests (seems like coverage is way below it should be – the follow up question: are the classes testable at all?) running.

My 0.02 cents.

I mention only the maven support which feels more mature than in Eclipse.  Source navigation and the refactoring support for Java lacks some features though. Well, I won’t bore you with details, there are plenty of blogs which do a brilliant job at commenting, comparing, explaining and ranting about almost all aspects of software engineering  – you aren’t interested in another one.

A killer feature for the productivity in Eclipse is the Mylyn approach. Netbeans pendant is Cube°n – but I didn’t look into that so far.

Some people have asked me to post more of stuff I’m currently interested in, since ” … it would be interesting …” (?).

Well here is a small list:

• scala
• eclipse
• netbeans
• javaFX
• tech related podcasts like javaposse, se-radio

As for the usage of generics in java, I liked Joshua Blochs talk about Effective Java Reloaded on parleys.com. If you follow his recommendations you make your developer life much easier.

On the aforementioned website you’ll find also talks from other computer superstars like Mr. Odersky. Watch it!

I am, however, disappointed about the (still) missing feature (?) to deploy on ‘real’ mobile devices, not just the emulator. This will be fixed soon I’m sure. There are some blogs which complain about the missing authoring framework, but I’m sure the engineers at Sun do their job perfectly well. A major contribution to the success story of JavaFX will be the modularization of the Java runtime distribution, using profiles for target systems.

Anyway, all competing technologies (Flash, Silverlight … ) converge to a common denominator, from a developer’s perspective it won’t make much of a difference.

Anyway, this will improve Sun’s perspective on the RIA market and finally provide an enterprise tool chain with decent tool support.

Of course I created a nice looking JavaFX Hello World demo which demonstrates Timelines, some graphic primitives, fonts and a home made gradient effect   using the Java Network Launch Protocol (JNLP) – here is the source.