posted 14 years ago
When I look back to ten years ago when OSGi was very new and think about how we prepared for the future (we now call it the present!), here's what comes to mind: We didn't bet our farm on some newfangled framework called OSGi. But we did bet that making our applications modular and relying on a clean separation between our business logic and our dependency injection details was a good idea. This meant that as OSGi evolved, porting our applications was pretty easy. If OSGi had withered up and died, we could have moved our well-structured application to some other more successful framework, or we could have rolled our own.
I thin the same holds true now as we contemplate the next ten years. Don't bet your farm on OSGi. Bet it on well-structured, highly modular, and decoupled architecture. Then OSGi (or Jigsaw or SpringDM or whatever) becomes more of a deployment detail. I write code all day long that runs on OSGi. But I spend a very very tiny portion of my time actually being concerned with OSGi per se.