Ok - I just fixed this- I don't understand WHY it fixed it but it did. I moved the if_<> into a metafunction class and passed the metafunction class to transform. I would think that either using it inline or as a mfc would be equivalent. (out of curiosity)- Anyone know the answer to this why this is the case?

-----Original Message----- From: Brian Braatz Sent: Wednesday, July 27, 2005 3:41 PM To: 'boost@lists.boost.org' Subject: MPL:Possible Placeholder problem with transform & if_ (?)

Either something strange is happening with arg<> and transform or I did not write this code correctly. ( :) )

Any help pointing me to either conclusion appreciated. The next step is to try the same algorithm with transform_view. I really don't want a lazy sequence (how could mother be proud?), but sometimes life is disappointing.

Thanks Guys.

CODE:

struct print_type { std::ostream & m_os; print_type(std::ostream & os = std::cout) : m_os(os) {}

template <typename T> void operator()(T const& v) const { m_os << "[ " << typeid(v).name() << " ] "; } };

template <class seq_t> void _print_types(const char * pHeading) { std::cout << "** " << pHeading << std::endl; mpl::for_each<seq_t>(print_type()); std::cout << std::endl; } #define print_types(seq) _print_types<seq>(#sew);

template <class T> struct YES { typedef YES type; }; template <class T> struct NO { typedef NO type; };

struct A{};

typedef mpl::vector<A> vec_t;

typedef mpl::transform < vec_t, mpl::if_ < boost::is_same<mpl::_1, A >, YES< mpl::_1 >::type, NO< mpl::_1 >::type

...

::type ::type result;

void test() { print_types (result);

/* //////////////////////////

PRINTS:

** result [ struct NO<struct A> ]

SHOULD PRINT:

** result [ struct YES<struct A> ]

//////////////////////////// */ }