Re: [boost] Question about is_enum

From: David Abrahams (dave_at_[hidden])
Date: 2002-10-14 09:53:53

"Kevin S. Van Horn" <Kevin.VanHorn_at_[hidden]> writes:

> I've been looking over some of the type traits stuff, and I had a question
> about the implementation of is_enum.  Here is the code in question:
> 
> template <typename T> struct is_enum
> {
> private:
>    typedef ::boost::add_reference<T> ar_t;
>    typedef typename ar_t::type r_type;
>        
>    BOOST_STATIC_CONSTANT(bool, selector =
>       (::boost::type_traits::ice_or<
>            ::boost::is_arithmetic<T>::value
>          , ::boost::is_reference<T>::value
>        // We MUST do this on conforming compilers in order to
>        // correctly deduce that noncopyable types are not enums (dwa
>        // 2002/04/15)...
>          , ::boost::is_class<T>::value
>       >::value));
>     typedef ::boost::detail::is_enum_helper<selector> se_t;
>     typedef typename se_t::template type<r_type> helper;
> public:
>     BOOST_STATIC_CONSTANT(bool, value = helper::value);
> };
> 
> I'm quite mystified by the comment about ORing is_class<T>::value into 
> selector.  How is the fact that a class C is noncopyable going to make
> is_enum_helper<false>::type<r_type>::value true?  

It doesn't help make anything true. It makes sure that the return
value is taken from

   is_enum_helper<true>::type<r_type>::value
                  ^^^^

In the case where T is a class type. Otherwise, 

   is_convertible<T,::boost::detail::int_convertible>

gets instantiated, which is only legal if T is copyable.

> I tried to look up the mentioned message

Message? Mentioned where?

> in the mailing list archives, and it didn't shed any 
> light.

-- 
           David Abrahams * Boost Consulting
dave_at_[hidden] * http://www.boost-consulting.com