[boost] Re: [rational] Announcing new maintainer -- call for featurerequests

From: Andy Little (andy_at_[hidden])
Date: 2005-04-09 08:10:08

"Jonathan Turkanis" <technews_at_[hidden]> wrote 

> 2. What changes or additions would people like to see?

One could make some modifications to rational  to make them it user friendly.  

For example the following results in a compile time failure n VC7.1:

    boost::rational<int> r1;
    boost::rational<long> r2;
    r1 + r2;

Naturally if  a bigint rational was to be used, I would expect 

    rational<bigint>() + rational<int>()  to work.

This is fairly trivial to accomplish using  BOOST_TYPEOF:
    eg something like  
   template <typename Lhs,  typename Rhs>
    typename boost::enable_if<  // only for compat with current header
        boost::mpl::not_<
            boost::is_same< Lhs, Rhs>
        >,
        rational<typename BOOST_TYPEOF(Lhs() + Rhs())>
    >::type 
    operator + (rational<Lhs> const & lhs, rational<Rhs> const& rhs);

FWIW this applies as well to other UDTs including interval and complex. 
which are currently inconsistent 

eg   std:::complex<double> d  ; 

d * 2; // fails to compile
 d *= 2;  // compiles ok   

regards
Andy Little