Question

Java doesn’t allow you to use an variable that may not have been initialized within a method scope. An uninitialized variable within a class scope may still be returned by a class method, and the value defaults to null.

Why the different treatment of the two different scopes?

public class TestClass {

    Integer i;
    Double d;

    public TestClass() {
        d = 1d;
    }

    public Double getD() {
        return d;
    }

    public Integer getI() {
        return i;
    }

//  public Integer getSomeInt() {
//      Integer i;
//      return i;
//  }

    public static void main(String[] args) {
        TestClass myClass = new TestClass();
        System.out.println(myClass.getI().getClass());
    }
}

This results in a NullPointerException, but returning i within getSomeInt() is a compiler error because “the variable may not have been initialized”.

Answer 1

The reason behind this are the limits of Java’s static code analysis. The compiler is able to prove beyond doubt that you will not read a stack-allocated local var before initializing it. This is impossible to do for heap-allocated memory and therefore Java mandates that all heap-allocated storage be zeroed out before exposing a pointer to it.

The consequence of this rule is that everything heap-allocated has a default value of zero (false, null, whatever the binary zero amounts to for the type).