Potential Final initializer bug.

Question

I noticed that the jdk 1.3 compiler, as well a the Visual Age 3.5.3 compiler, will allow this code:
public class FinalTest
     private final int _i;
     FinalTest()
          i();
          _i = 9;
          System.out.println(_i);
     private void i()
          System.out.println(_i);
     public static void main (String args[])
          new FinalTest();
}When I run it, the output is:
0
9This seems like a bug, unless there is some obscure rule about blank finals and calling instance methods which use them BEFORE they are initialized in the constructor. At any rate, this does not seem like a positive feature to me.

Answer 1

On a similar note, does anybody else find it worrying
that you can call non-private, non-final instance
methods from a constructor?
The simple answer is "don't do it", but people do do
it...It would be nice if the compiler could warn about public methods being called from the constructor, but could it detect the following? This particular example would (probably) be picked up rather quickly, but does (should) an external constructor count as a "dangerous" method to call from a constructor? Or only when static state is involved... the rules are starting to get hard :-)
public class A {
    private static final A INSTANCE = new A();
    public static final A getInstance() {
        return INSTANCE;
    private B b;
    private A() {
        b = new B();
    public String getProperty(String key) {
        return "something good";
class B {
    private String prop;
    public B() {
        prop = A.getInstance().getProperty("gimme");
}