David Abrahams wrote:
But I don't really understand what you're talking about. Too much missing detail :(
You're right, of course. I was under impression yesterday and posted it without any explanation. The cleanest way to explain it is an analogy with mpl unique algorithm. Let say we have mpl::vector700 of free unary function signatures with only few of them being unique. The algorithm I've tested is similar to this: typedef mpl::vector700< /* 700 signatures of free unary functions */ > signatures; typedef mpl::unique<signatures, is_same<_1,_2> >::type unique_signatures; My algorithm takes positions as mpl::range_c<int,1,701> and a class with 700 call-operators. It returns a positions of unique signatures: struct X { void operator()(overloads::overload_id<1>, int) const {} void operator()(overloads::overload_id<2>, char) const {} // ... void operator()(overloads::overload_id<10>, int***) const {} // ... void operator()(overloads::overload_id<200>, int*) const {} // ... void operator()(overloads::overload_id<700>, int*) const {} }; int main() { typedef unique<X, mpl::range_c<int,1,701>, mpl::identity<_1>, uniform_row_search>::type result; BOOST_MPL_ASSERT(( mpl::equal<result, mpl::vector_c<int,1,2,10,39,200> > )); } -- Alexander Nasonov