Fixed a long-standing but well hidden bug in message

protocol.  In the wild, I've only seen this crop up in
really really odd circumstances.  As it turns out, the 
throughput tests eventually trigger this when testing
TCP.  I've lost sleep over wondering when this was going
to bite for real so I'm glad to have what is essentially
the last known bug in message transfer fixed.
This change handles the case where so far only one
byte of the two byte message size has been read from
the network.  Rare, but it can happen.


git-svn-id: https://jmonkeyengine.googlecode.com/svn/trunk@7976 75d07b2b-3a1a-0410-a2c5-0572b91ccdca

engine/src/networking/com/jme3/network/base/MessageProtocol.java

@@ -58,6 +58,7 @@ public class MessageProtocol
     private LinkedList<Message> messages = new LinkedList<Message>();
     private ByteBuffer current;
     private int size;
+    private Byte carry;
  
     /**
      *  Converts a message to a ByteBuffer using the Serializer
@@ -110,11 +111,35 @@ public class MessageProtocol
         while( buffer.remaining() > 0 ) {
 
             if( current == null ) {
+
+                // If we have a left over carry then we need to
+                // do manual processing to get the short value
+                if( carry != null ) {
+                    byte high = carry;
+                    byte low = buffer.get();
+                    
+                    size = (high & 0xff) << 8 | (low & 0xff);
+                    carry = null;
+                }
+                else if( buffer.remaining() < 2 ) {
+                    // It's possible that the supplied buffer only has one
+                    // byte in it... and in that case we will get an underflow
+                    // when attempting to read the short below.
+                    
+                    // It has to be 1 or we'd never get here... but one
+                    // isn't enough so we stash it away.
+                    carry = buffer.get();
+                    break;
+                } else {
                     // We are not currently reading an object so
                     // grab the size.
                     // Note: this is somewhat limiting... int would
                     // be better.
                     size = buffer.getShort();
+                }               
+ 
+                // Allocate the buffer into which we'll feed the
+                // data as we get it               
                 current = ByteBuffer.allocate(size);
             }