I sincerely hope I'm missing something about how java.util.map can be
used, but it seems that there is no way to update a single value
without traversing the map twice, at least not cleanly.  As an
example, consider this code taken from the Sun autoboxing example:

public static void main(String[] args) {
 Map<String, Integer> m = new TreeMap<String, Integer>();
 for (String word : args) {
   Integer freq = m.get(word);
   m.put(word, (freq == null ? 1 : freq + 1));
 }
}

Notice how both get() and put() are invoked.  Clearly this means that
the map is traversed twice - once to find the value, and once to
figure out where to put a new value (which happens to be the same
place!).  In C++ (where I come from), maps contain objects of type
pair<key, value>, and one can therefore retrieve the pair<> in
question and update the value with only one traversal of the map.  Is
there any way to avoid two traversals in Java short of writing a kludgy
wrapper object such as

class WrapsAnInt {
 private int wrapped;

 public WrapsAnInt( int w )        { wrapped = w; }
 public void setWrapped( int w )   { wrapped = w; }
 public int getWrapped()           { return wrapped; }
}

and put *those* in the map?  If not, why didn't Java simply introduce
the concept of a pair<> like C++ did and fix the problem?