Re: has_foo revisited...

From:

Carl Barron <cbarron413@adelphia.net>

Newsgroups:

comp.lang.c++.moderated

Date:

2 May 2006 06:43:44 -0400

Message-ID:

<010520061936389661%cbarron413@adelphia.net>

In article <1146493260.632850.19690@g10g2000cwb.googlegroups.com>,
<rune.sune@yahoo.com> wrote:

struct has_foo
{
    typedef char One;
    typedef struct { char tmp[2]; } Two;

    typedef int (T::*PMF)() const;
    template <PMF> struct wrapper;

    template <typename U>
    static One test(U*, wrapper<&U::foo>* = 0); // why these two params?
    static Two test(...);

    enum { value = (1 == sizeof(has_foo<T>::test((T*)0)))};
};

    so that
    struct is_false {int a;};
    struct is_true {void foo() {}};

    has_foo<is_false> which does not have a member foo selects the
ellipsised function since is_false::foo does not exist and
&is_false::foo is a syntax failure so the only candidate is test(...)
where &is_true::foo is valid syntax and thus test(is_true *,
&is_true::foo) is a better match than test(...) and test(is_true *,
&is_true::foo) is chosen.

      [ See http://www.gotw.ca/resources/clcm.htm for info about ]
      [ comp.lang.c++.moderated. First time posters: Do this! ]