Basically, my question is if the equivalent of this C++ code possible in Scala?
int x = 1;
int& rx = x;
rx++; // x = 2 now
std::cout << x;
I.e. to have two different variables both point to the same thing (and both change, when one of them is changed)?
Also, if it is, can a scala function return such a reference?
Well you can't. In Scala, Int, Double, Float, Boolean etc are AnyVal, you can't pass them by reference. You can pass references of AnyRef i.e. String, List, Set, Any user defined object etc.
You can sort of emulate it:
class Ref[A](var value: A)
object Ref { def apply[A](a: A) = new Ref(a) }
val a = Ref(1)
val b = a
b.value += 1
println(a.value) // 2
Or even:
class Ref[A](get: => A)(set: A => Unit) {
def value = get
def value_=(a: A): Unit = set(a)
}
var a = 1
val b = new Ref(a)(a = _)
b.value += 1
println(a) // 2
You could probably write a macro to make constructing a Ref a bit prettier in this case (e.g. Ref(a) and the macro generates the appropriate closures).
The usual workaround in Java/Scala that should seem somewhat familiar to the C/C++ programmers is the following:
val x = Array[Int](41)
val rx = x
rx(0) += 1
println(x(0))
You cannot have references or pointers to primitive integer variables, but arrays kind of contain information equivalent to a pointer, so if you declare x to be an Array[Int] with a single element instead of an Int, you get roughly the same behavior: the above example will print 42.
You can think of it like this:
Array[X] roughly corresponds to X[] in C, which in turn roughly corresponds to X*x(0) corresponds to accessing the value at the "pointer" x, so it's something like *x in C.The crucial difference is that you can only declare an array right from the beginning. There is no way to get a "pointer" to an arbitrary location in memory.
Note however that the array will reside on the heap (usually, unless it is eliminated by some escape-analysis), you cannot keep it in the stack frame of a function.
