Ehrm… ok got to admit it… this blog has been pretty much brain dead over the past few months. But, according to tradition I got to cover at least a few conferences a year and since its JavaPolis time again that is just barely enough of an excuse to bring this blog back to life. So here we go…
As usual a week of JavaPolis kicks off with two days of University. During the day 3 hour sessions offer an in-depth glance at a selection of popular topics. For today I picked a session titled “Open Source ESBs” for the morning session and “Google APIs” as the afternoon pick. Afterwards there were some 30 minutes short sessions that were mainly sponsored sessions.
The Open Source ESBs session was presented by Jos Dirksen and Tijs Rademakers, both familiar to me because we had them to talk at NLJUG events as well. During the talk they covered Mule, Apache ServiceMix and OpenESB. Mule wasn’t entirely new to me. I have been experimenting somewhat with a previous version earlier this year. I had never looked at ServiceMix before. At JavaOne I attended some talks on OpenESB and JBI but since I had done nothing with it afterwards it kind of slipped from my mind. All in all this session was a very interesting teaser to all three solutions. Both Mule and ServiceMix involve a lot of XML and no (graphical) tooling to help out in defining transformations and mediations. OpenESB on the other hand has a lot of backing from the various NetBeans plug-ins that come with it. My ESB experience besides Mule consists of WebSphere ESB and tools like WebSphere Integration Developer and from what I saw of OpenESB and NetBeans it was quite comparable.
The session was pretty much Spring infested with Mule 2 being Spring based and ServiceMix offering Spring bindings that were used in the various code examples that Jos and Tijs showed. If you happen to know me, or have read earlier installments of this blog it might not be a total surprise that I’m not really a fan of Spring. So in fact I got to admit I was quite relieved when I first saw the JavaPolis schedule and noticed there were hardly any Spring talks on it. Unfortunately, it seems as if Spring has gone underground and pops up in totally unrelated talks. Just to annoy me 😉
Anyway, besides the Spring nightmare it was actually a very interesting session with just enough information on all three solutions to get the high level picture. On first impression I think I like the Mule approach best. Just a plain EAI solution without all of the fancy and formal stuff around it.
After lunch I attended the Google APIs session. One of the reasons for picking this session was the session abstract that talked about the OpenSocial Container and I’m quite curious about that. Not in the least since I’m still planning on becoming a billionaire by inventing the next piece of fantastic social networking software that everybody wants to use. Unfortunately my dreams were shattered by the fact that Dick Wall – who presented the session – mentioned that the abstract was mixing up this talked and the talk on Google Web Toolkit and the OpenSocial Container on Wednesday. Luckily there was enough of other interesting stuff happening at Google to talk about so I decided to stay. What the session did feature were actually two separate API level talks on the Google Collections API and the Google Data APIs. Interesting stuff that tickles your imagination! Dick did a typical developer-to-developer kind of talk and did a good job evangelizing these APIs.
Late afternoon was covered by two shorter sessions of 30 minutes each. I picked two with the common theme of Continuous Integration. The first was about Atlassian Bamboo – a product they recently acquired from Cenqua – and the second session was about Hudson. Funny how two tools that share a common theme can be so totally different. A lot of that is perception caused by the presenter. The guy from Atlassian that did the Bamboo presentation was either terribly jet-lagged or lacked presentation skills at all. I admit his way of presenting does not really encourage my feelings about the product. A few weeks back I got an e-mail from Atlassian – since I’m the technical contact for Jira within my company – trying to convince me to take a look at Bamboo. As I had found no time to do so this was my first experience with it. My first impression is that it features some very nice functionality for build telemetry and trend analysis but as usual with Atlassian products it is very much centered on their other products. What I am actually looking for is a more open approach to build information and trend statistics since our Software Factory “Endeavour” uses its own project portal and data warehouse. With Bamboo I’m under the impression that it is possible to get anything IN but getting stuff OUT seems impossible. I think I’m going to stop by their booth anytime this week to ask them a few questions about it. Hudson on the other end does seem to promise a far more open solution since it’s entirely built to be extensible. In fact everything – even the basic functionality of Hudson – is a plug-in. Also the strong Maven integration sounds promising. This is definitely a topic to look into after the conference. To be continued.
Another highlight of the day was the Parleys Newspaper that was handed out to all attendees. It featured an article on the NLJUG and how we are trying to reach out to the student community. The article talked about the project that I ran at the AVANS Hogeschool in the Netherlands where a team of 25 students built us a new administration system.
By the way, I’m not the only one from Info Support attending JavaPolis. I’m accompanied by Paul Kramer – a trainer at our Knowledge Center – and Paul Bakker – who also has his own blog.
In summary this first day was definitely alright and I’m really looking forward to tomorrow. Not in the least since I’m invited to the JavaPolis Speakers and JUG Leaders diner!
*update* fixed the link to Paul's blog which appeared to be totally different that mine. Thank to my faithful reader Wouter for pointing it out 🙂