Question

I am manually converting code from Java (1.6) to C# and finding some difficulty with the behaviour of primitives (int and double). In C# it appears that almost all conversions happen automatically

        List<double> list1 = new List<double>();  // legal, C#
        double d0 = 3.0;
        list1.Add(d0);                            // legal, C#
        Double dd = 2.3f;                         // legal, C#
        list1.Add(dd);                            // legal, C#
        List<Double> list2 = new List<Double>();  // legal, C#
        double d1 = 3.0;
        list2.Add(d1);                            // legal, C#
        list2.Add(2.0);                           // legal, C#
        double d2 = list2.get(0);                 // legal, C#

but in Java only some are allowed

        List<double> list1 = new ArrayList<double>();  // illegal, Java
        List<Double> list2 = new ArrayList<Double>();  // legal, Java
        double d1 = 3.0;
        list2.add(d1);                                 // legal, Java
        list2.add(2.0);                                // legal, Java
        double d2 = list2.get(0);                      // legal, Java

I’d be grateful for a systematic analysis of the differences and any underlying rationale.

Answer 1

In your C# example there is no boxing or unboxing (and autoboxing) happening. double is just an alias for the struct Double.

In Java, the boxing is necessary. Because of type erasure, you can’t create a List<double>, only List<Double>. At compile time, List<?> becomes List<Object> and boxing/unboxing will need to take place so you can add a native type variable to a List<Object> or assign a native variable to an element of the List.