Please share:
Just teasing. Hehe.
No seriously, for whatever reason there are those few things that come around (technically) every once and a while that just make me giddy. It was embedded systems, then it QNX, then it was Integrity, then is was ActiveMQ/ServiceMix, throw Ruby in there some where - and now it is OSGi (again).
I have a couple of cool project spinning these days - and totally divergent. One is in the “enterprisy” space and the other is in pervasive devices/ubiquitous computing. The common denominator is OSGi, and there’s no irony here. Unfortunately, I can’t talk at all about the latter project - but the former probably can leak a few details here and there. One of the things that is really interesting about this project is that its current form is seriously broken. Not just broken functionally, but from a developmental perspective too (code, compile, test, deply, test, code, blah, blah). It is just too damn hard to build in modularity where it is needed, and integrate external and internal interfaces, while maintaining simplicity. The needs are not great, nor complex. But, the implementation dwarfs the needs, and the complexity went through the roof. And we are just now considering persistence. Soooo.
Enter the need for an off-the-shelf framework. Grails? Yes, because we are a predominantly customer-driven, and agile environment; the customer prescribes Java. Grails affords us the potential support many different external interfaces including ReST (default), WS/SOAP, and JMS. Nevermind the beauty of GORM’s abstraction to Hibernate.
Behind Grails comes OSGi. Not beneath, or around - but behind. We need a service infrastructure to house extensible algorithms. Sure we could go the strange world of ESBs (and we do have the JBI expertise in house). But, we don’t need that overhead. OSGi allows us to provide a nice package around a simple API pushing the extensibility with absolute modularity and scalability (with cool stuff like hot-deploy).
Both Grails and OSGi are well-blogged topics and I don’t intend to rave and praise them. Over the next couple of days I’ll drop a bit more, in detail, about how I’m going about integrating the too. I suppose I should state that I have given zero thought to getting Grails OSGI bundle-ized. That’s just not something at this point that I really want to think about. But, is an interesting notion - and does make sense because it would afford us a bit of independence from any one OSGi framework. We are using Felix, really just because it is an Apache project and it is a little less coupled to any major project like Equinox is to Eclipse.
More to come…
Please comment: