Category Archives: Eclipse

Eclipse is one of the greates platforms I ever worked with. I’m a committing member of the Eclipse Foundation. This category will collect everything about Eclipse, the Rich Client Platform, SWT, GEF and other related projects.

On my way to JavaOne 2012

For the first time ever I’ll be attending JavaOne next week. I’ll speak together with Shaun Smith from Oracle about Polyglot Persistence with EclipseLink JPA.

Polyglot Persistence: EclipseLink JPA for NoSQL, Relational, and Beyond
(Tuesday, Oct 2, 3:00 PM – 4:00 PM – Parc 55 – Cyril Magnin I) 

Yes, there will be a demo running EclipseLink on Gyrex OSGi connecting to MongoDB. 🙂

I’m currently in Frankfurt  waiting at the gate. I’ll arrive Friday night in San Francisco and will stay till next Friday (Oct. 5th). If you are around, interested in a chat (about runtimes, OSGi, other stuff that matters) and/or want to grab a beer please ping me!

Eclipse 4 – Uh Oh! Trouble’s brewin?

Recently I was approached by someone I worked with in the past about Eclipse 4. Apparently, he read Andrew’s blog (Something is really broken with Eclipse foundation) and was really concerned. I’ve been very quiet in the whole debate that was going on but I had my thoughts. So I crafted a response that I’d also like to share with the public in this blog.

It’s no secret that IBM’s investment into the core platform is declining. From looking at the stats today, they still have 80+ active Eclipse committers. However, only three (according to this mail) are working on the core platform (Platform UI that is).

Another issue is attitude. There are a lot companies out there using and building on the core platform. However, most just take the investment by IBM for granted. A lot also simply underestimated the effect of a new major release. Eclipse 4 is a complete rewrite in terms of Platform UI. Of course it’s not as stable as 3.x yet. However, I heard statements like “oh it (any Eclipse release) always worked for us“. So some just didn’t care a lot about testing on 4.x. That valuable testing feedback is painfully missed these days.

Frankly, I’m curious myself how this will turn out because developer tools are still necessary. I just can’t believe that no one of the large companies with hundreds and thousands of developers realizes that. I actually thought about making a business out of that, i.e. getting paid for working on the core of Eclipse by selling maintenance/support contracts/features. However, I also think that this will be a challenging business to develop. It’s hard to convince companies of certain sizes to spend money on something they either get for free or invest money into someone that is too small for their business.

BTW, I know of at least a few companies in the Eclipse ecosystem that tried. But they never became a significant contributor to the core platform. Not sure if it was a matter of lack of business or lack of focus. They are making money with Eclipse, though. Their business is also based on the core platform. Wondering what they will do.

It’s also hard to convince members who are already paying membership fees to spend additional money for developing Eclipse.  Paying twice sounds unfair. So another intriguing option is to change the Foundation by-laws so that it will be allowed to hire rock star core platform developers. Maybe they just need to hire a complement to Linus for Eclipse. Well that would be me or course. 😉

I also do have an issue with people complaining very loudly about something. It’s not that there weren’t any previous chances to chime in and speak up when previous decisions were made. But those complaining the loudest do not necessarily represent a majority. Note, I’m not denying any of the Eclipse 4 problems here. Just saying something I observed during my excursions in politics.  It’s not what you say, but how you say it.

Luckily, Google not only noticed that discussion but carefully paid attention for an immediate need. Their Open Source office donated $20.000 to buy new performance testing hardware. Thank you Google! 

BTW, Eclipse 4.2 runs fine on my new notebook. I did not observe the performance issues so far. However, I remember I had issues with the new animations on my old notebook. I showed them to Eric at ECE 2011 . But that was on a development build and turning off animations solved my performance issues back then.

I really wonder how this will turn out. I recently learned that SAP is the 3rd largest contributor to Eclipse these days. I hope they will start investing into the core platform. The guys in the SAP Open Source office (although not officially named as such) are smart. But it would mean for IBM to give up control and let others play an important role in the strategic future of the core platform.

Eclipse at FOSDOM 2012

It’s my first out of two days at FOSDEM 2012. It took us quite a ride to get from the hotel to the ULB. We tried to order a taxi but the people at the reception told us that it would take at least 1.5 hours till a taxi arrives. Luckily, Mike and Andrew know someone who has been at FOSDEM a couple of times before. He guided us safely to ULB using a combination of walking, metro, tram and more walking.

We quickly setup an Eclipse stand over there and Mike, Andrew and myself are showing demos and talking to people. BTW, thanks to the FOSDEM organizer to have it well prepared so that we just needed to setup our banner and our notebooks for the demos.

So far we have a great mixture of questions from developers using Eclipse for their day-to-day work, programming questions of Eclipse plug-in developers and people interested in Orion. There are also people stepping by that have no questions – they introduce themselves as happy Eclipse users and appreciate what the committers of the various projects have built over time. Thanks for those kind words folks!


Use Target Definition Files in PDE Build

I’m not hip when it comes to building Eclipse based products. I still use plain vanilla PDE Build. The cool guys all switched to Tycho. Well, you’re allowed to call me dinosaur then. Anyway, I’ve been using .target files as a source for dependency management for quite a while now. They are very useful in Eclipse. All dependencies for all the bundles are defined in one place. You simply open one and Eclipse starts downloading one half of the Internet. Maven already did the other half, I’ve heard.

I also use them to drive PDE Build. Instead of specifying stuff to fetch for a build again I simply point PDE Build to the same .target file I use in the IDE. Works like a charm as long as you only use “Software Sites” (p2 repositories) which are accessible via URL by any developers Eclipse instance as well as the build machine.

Previously, this was implemented by transforming the .target file XML using XSL into an Ant script. The Ant script did a bunch of p2 mirror calls to download the bytes. I recently upgrade to a newer version of the builder containing a Juno version of the org.eclipse.pde.core bundle. Starting with 3.8 this bundle defines a nice Ant task that does not require the XSLT magic.

<target name="buildTargetPlatform" unless="skipTargetDefinition">
  <!-- note: this requires pde.core (from Juno) in the base builder -->