Well, take this Clojure tutorial, for instance, comparing a simple DI implementation in Java with a single function in Clojure. The latter bears almost no resemblance to the former, other than achieving the desired effect. Is it fair to say that the user of a closure is implementing the Dependency Injection pattern, or is it a positive side effect of the fundamental properties of first class functions and lexical scope?
IMHO it's largely a side effect of first class functions. In Java in order to pass a function you need to pass an object that has the function, or an interface with the function, so it leads to a lot of line noise and thus people don't do it much. DI doesn't look impressive in a functional language because it's so well supported, but if you look at something like a global accumulator in a functional language you'll need a monad because you need state.
eg in Java it's easier to write:
int accum = 0;
for(int i : collection){
accum +=i;
}
rather than
collection.reduce(new IReduce<int> {
int reduce(int accum, int i){
return i + accum;
}
})
I'm a bit rusty on Java so the syntax may be off but you get the idea, where as in a functional language (F#) that code is reduced to:
collection
|> Seq.reduce +
Or in an imperative language that supports function passing (C#)
http://vimeo.com/10368175