If you look at Haskell, Erlang, Scala and F#, what do you see? If you look at Haskell, you see something which, within its context, within the little framework it sets up for its sandbox, it’s very consistent, it’s very beautiful. You look at Erlang, it kind of fits, the bits fit together nicely. Of course, they don’t fit together nicely with the JVM or with .NET or anything like that. If you want to use all the nice things that are there, you can’t use them, or you can use them, but it’s difficult. So the other approach is, you say “Let’s use the JVM and target lots of different languages, so that the different languages can use each other” or you can do that within the .NET framework, you get Scala and you get F#.
The benefit there is you can use all these other things that are available, but in order to do them, you have to break them, massively corrupt and break these abstraction boundaries and I don’t like that. I think you are breaking abstraction boundaries in the wrong place. How I would like to see systems built is through communicating back boxes. And I would like to see the type systems applied to the definition of the protocols themselves and I haven’t seen that done.Joe Armstrong from http://www.infoq.com/interviews/armstrong-peyton-jones-erlang-haskell
ps: reminds me of “And you get all the VM libraries for free! Is it actually what I want when I switch languages?“
