Name: rmT116609			Date: 05/06/2004


A DESCRIPTION OF THE REQUEST :
NewDirectByteBuffer always has order ByteOrder.BIG_ENDIAN, even when the order is ByteOrder.LITTLE_ENDIAN.

The following might be a fix for this problem, except that isDirect() is abstract, so JVMs are not required to implement the isDirect() method:

ByteBuffer bb = nativeMethodReturningDirectByteBuffer();
if (bb.isDirect()) {
  bb.order(ByteOrder.nativeOrder());
}

Also note that passing a memory address from Java to C++ works similarly... C++ assumes the data is LITTLE_ENDIAN unless the ByteBuffer that refers to the address is explicitly given order ByteOrder.nativeOrder().

JUSTIFICATION :
isDirect() isn't the best solution. Maybe an isNative() method would be better?

But preferably the ByteBuffer returned by NewDirectByteBuffer would already be ByteOrder.nativeOrder() by default.

Someone on comp.lang.java.programmer suggested htonl or kind. I have no idea what these are, but possibly the would be a good  solution to the whole big-endian little-endian mess.

EXPECTED VERSUS ACTUAL BEHAVIOR :
EXPECTED -
Any ByteBuffer returned by NewDirectByteBuffer should have order ByteOrder.nativeOrder().

Obvious, no?
ACTUAL -
ByteBuffer returned by NewDirectByteBuffer has order ByteOrder.BIG_ENDIAN, which is false, mis-leading, incorrect, etc.

---------- BEGIN SOURCE ----------
ByteBuffer bb = nativeMethodReturningDirectByteBuffer();
System.out.println("bb.order() " + bb.order());

Complete C++ and Java source code is available here:
http://www.alpha-mu.com/alpha/files/



---------- END SOURCE ----------

CUSTOMER SUBMITTED WORKAROUND :
ByteBuffer bb = nativeMethodReturningDirectByteBuffer();
if (bb.isDirect()) {
  bb.order(ByteOrder.nativeOrder());
}
(Incident Review ID: 261264) 
======================================================================