Re: ANN: Optimized numeric_cast<> implementation

From: Fernando Luis Cacciola Carballal (fcacciola_at_[hidden])
Date: 2001-11-05 11:07:37

--- In boost_at_y..., helmut.zeisel_at_a... wrote:
> --- In boost_at_y..., "Fernando Cacciola" <fcacciola_at_g...> wrote:
> > Hi,
> > 
> > I uploaded a new version of the improved numeric_cast<>.
> > 
> > http://groups.yahoo.com/group/boost/files/improved_numeric_cast
> > 
> > It now works with bc5.5.1 and gcc2.95.2+
> > 
> > I also uploaded the individual files along with the .zip archive.
> > 
> > I'll appreciate any comments...
> > 
> 
> Here is one more comment from me:
> 
> I get the impression that you changed the default
> behavior to rounding instead of truncating;
> i.e. numeric_cast<int>(5.7) == 6.
> 
Yes, I did.

> I am not sure what the best choice for the
> default behavior should be,
> but I definitely need truncating.
> 
I think that I was biased by my needs. But, on second thought, the 
default behaviour of the built-in cast is truncation, so I guess you 
are right and it should truncate by default.

> Is there a simple way to obtain truncating?
> 

Yes.
In "numeric_cast_traits_detail.hpp", 

In: 

  template<class T,class S,class range_checker>
  struct rounding_converter
  {
    ...

    static S round ( arg_type s )
    {
      return s+half();
    }

Change:

      return s+half();

to:

      return s ;

Best Regards,

Fernando Cacciola
Sierra s.r.l.
fcacciola_at_[hidden]
www.gosierra.com.