java version "1.3.0"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.3.0-C)
Java HotSpot(TM) Client VM (build 1.3.0-C, mixed mode)


According to JLS 15.10.1, and to JVMS 6, multianewarray, if ANY dimension
expression is negative in an array creation expression, then a
NegativeArraySizeException MUST be thrown.  However, this program executes
without problems:

class Foo {
  public static void main(String[] args) {
    try {
      Object o = new int[0][-1];
      System.out.println("Oops - shouldn't get here");
    } catch(NegativeArraySizeException nase) {
      System.out.println("Correct behavior");
    }
  }
}
(Review ID: 119887) 
======================================================================

Manually check for a negative dimension when a previous one was 0.
======================================================================

The first array element is zero size therfor we silently discontunue evaluating next array element size. This will not be fixed. This is not a bug.  If user puts [1][-1] an exception will be thrown as desired and spec'd.
Closing as a will not fix.
  xxxxx@xxxxx   2002-04-08

FWIW, this contradicts the JVMS and the JLS. So it is a bug. 
I doubt if it makes a big difference to anyone either way, but the KVM
behaves differently (and for once, they're right).

  xxxxx@xxxxx   2002-07-31

This is a VM bug, and you should not have closed it.  The
specs of JVMS 6 specifically say that all dimensions must be
non-negative; and do not allow you to short-circuit
evaluation just because a 0-size dimension was encountered. 
Besides, when using reflection, the similar construct WILL
throw the appropriate NegativeArraySizeException:

java.lang.reflect.Array.newInstance(int.class, new int[]
{0,-1});