Saturday, May 3, 2008

On "Java, Improve or Renew", Java Posse #182

So, the "dual post" from me and Christian has not yet been produced (we need to drink coffee together more often!). It's not really so that my fingers are itching - on the contrary; I've been writing on my thesis report like never before. The part I'm writing right now is about Domain Specific Languages, where focus is on writing internal DLS in Java. But right now, I feel like writing something short about something else..

I listened to on of the Java Posse podcasts yesterday. Its title was "Java, Improve or Renew?". I've actually started to listen to a lot of other podcasts (other than Hanselminutes and .NET Rocks), a complete list with reviews will come in a later post. I like the "Roundup" discussions of Java Posse a lot! There's so much people in the room, it sometimes impossible to identify who's who, but the good thing is that their opinions about the specific topics are so strong!

So, the "Improve or Renew" podcast was very interesting because to me it seemed like a quick (72 minutes) summary of what challenges Java has right now, to meet the future demands on the language. And not only about the language, but of course also about the bytecode itself.

There's so many factors to put into account for when changing/improving a language! Breaking backward compatibility is very painful when you're the biggest business software language in town, so that should be avoided as long as possible. When listening to the show, it seems like the question of closures in Java 7 really boils down to the implementation of generics in Java. .NET and Java has different implementations of generics and this is the reason that .NET had no problems putting lambdas (anonymous delegates) in - the generics implementation is the whole way down to the IL. It will indeed be very interesting to follow how the Java community deals with this question, but (as someone in the Java Posse panel points out) closures without a proper generics implementation will be a "half-baked" concept that's likely to make people disappointed.

Notice that I'm not trying to say that .NET is better choice than Java in any way! There's a lot more things to a language than small (but important) language details. Certainly no companies (or at least very few) are going to abandon Java in favor of .NET just because of the differences of generics, but it adds one more argument to the discussion of choosing framework (for you who have that choice).

But honestly, I must admit that Wes Dyer's post on monads in C# ("The Marvel of Monads") was really inspiring stuff. Monads in Java seems to be a far away, but I guess it's up to the Java community to decide on the "whats, whys and whens" on creating the proper building blocks for enabling them.

No comments: