Re: [boost] numeric_cast

22 Jun 2011

      On 6/21/2011 4:33 PM, Antony Polukhin wrote:
...
First of all, it looks like the numeric_cast library is not maintained
any more, so you won't get a fast answer.
I was hoping the community might step up on this.
...
There is a chapter called "Type Requirements and User-defined-types
support" in numeric_cast library, that describes how to make your new
numeric type usable with numeric_cast.
I've read it.
...
Chapter "converter<>  function object" describes how to use
boost::numeric::converter with different casting polices.
Did I answered your questions or have i missed something?
I think you've missed something.

Here is the code for numeric_cast that is in boost:

   template<typename Target, typename Source>
   inline
   Target numeric_cast ( Source arg )
   {
     typedef boost::numeric::converter<Target,Source> Converter ;
     return Converter::convert(arg);
   }

Note the way Converter is defined. Here is the definition of converter:

template<class T,
          class S,
          class Traits           = conversion_traits<T,S>,
          class OverflowHandler  = def_overflow_handler,
          class Float2IntRounder = Trunc< BOOST_DEDUCED_TYPENAME 
Traits::source_type>  ,
          class RawConverter     = raw_converter<Traits>,
          class UserRangeChecker = UseInternalRangeChecker
         >
struct converter : convdetail::get_converter_impl<Traits,
                                                   OverflowHandler,
                                                   Float2IntRounder,
                                                   RawConverter,
                                                   UserRangeChecker
                                                  >::type
{

As written there is no way for a user to specify a different overflow 
handler, rounding policy, or range checker. The defaults are always used 
when numeric_cast is called.

Regards,

Brandon Kohn