In the paragraph "Limitations", where the forwarding problem is discussed, the following appears: "...An oft-proposed "solution" to this problem is to add an overload: template<class T> void f(T & t); template<class T> void f(T const & t); Unfortunately, this (a) requires providing 512 overloads for nine arguments and (b) does not actually work for const arguments, both l- and r-values, since the two templates produce the exact same signature and cannot be partially ordered." It's asserion (b) that puzzles me. Doesn't the standard say that the second overload of f is more specialized thus yielding a well-defined partial order? This tricky businesses always elude me, so I might be totally wrong, of course :) Joaquín M López Muñoz Telefónica, Investigación y Desarrollo