Mylar and the Way of the Task

September 23, 2005

If you’re an Eclipse jockey and you have lots of different tasks on the go at the same time like me, especially if they span multiple projects, you may well need some special mylar care and attention. Mylar is a cool new Eclipse project that monitors your interaction with Eclipse creating a context specific to the current task. Switching from task A to task B closes everything for A and reopens all the stuff you were using for B. There are a bunch of other features like color coordinated task items. Mylar is a work in progress right now – the current 0.4 release seems to have made my rename refactoring very slow and broken a few other things, so its probably a bit early for most people. Could be great though.

Tools are a vital part of our lives. One of the tenants of good bush craft is build the tool first, then use it to do the work. Most outstanding enterprise software products feature outstanding tools. I’m a big fan of good tooling, and a particular fan of Eclipse.

If you are considering developing tools for the enterprise space look first at Eclipse. You would really need to have a very strong reason not to build on Eclipse – leveraging the various sub-projects and the tooling that is already out there should enable you to build functionally impressive and slick looking enterprise tools fast. Unless you are doing “end user” style browser based tools, Eclipse is the business.

I find it astonishing that it took Borland so long to drop their own IDE platform in favour of Eclipse – as a Java developer I dropped JBuilder years ago, Swing was (and despite its advances really still is) too clunky on the Windows desktop. And look at BEA and BEA Workshop, they put a lot of effort into their own IDE framework, and it blew up in their faces. When Weblogic 8.0 launched the Eclipse juggernaut was already unstoppable. BEA had to rethink and back Eclipse. Just recently they bought M7 of NitroX studio fame. It is a pity those good folks over at JetBrains keep putting effort into their tooling platform – if they could just take the best out of IntelliJ IDEA, marry it with Eclipse, and switch all that talent to higher level value add, now that would be something. What can I say about Sun and NetBeans? If only Sun could get past that name, and do a little embrace and extend. Actually, I think Sun is a great company and you don’t need me to tell you there is engineering talent to burn there, but sometimes that talent is misguided. But who knows, if NetBeans is the only competition left for Eclipse, then maybe its a good thing. Competition breeds innovation, and Eclipsers need to keep innovating. Thankfully the healthy development community involved with Eclipse, its subprojects etc., are showing every sign of doing so.

One of the things I love about Eclipse – apart from all the (growing list of) tooling goodies you can reuse and extend – is SWT. In fact I’m a total convert to SWT – I would rarely now choose Swing over it for rich client applications.

Eclipse is one of those things – like Java, JavaEE (well, in some form anyway) and XML – that will most probably span many computing generations.

Test Perfect Development

September 5, 2005

Write a failing test, make it pass. Test Driven Development. Beginners Tip: write the test to the “perfect” test API – don’t even think about how the test is going to work. Then go down a level, code the details to another perfect API. At some point you’ll feel you cannot go any further. Then worry about implementing the API.