Re: [boost] Is there any interest in Boost.Conversion?

Subject: Re: [boost] Is there any interest in Boost.Conversion?
From: vicente.botet (vicente.botet_at_[hidden])
Date: 2010-05-29 08:19:14

----- Original Message ----- 
From: "Emil Dotchevski" <emil_at_[hidden]>
To: <boost_at_[hidden]>
Sent: Saturday, May 29, 2010 6:16 AM
Subject: Re: [boost] Is there any interest in Boost.Conversion?

On Fri, May 28, 2010 at 2:55 PM, vicente.botet <vicente.botet_at_[hidden]> wrote:
> ----- Original Message -----
> From: "Emil Dotchevski" <emil_at_[hidden]>

> > > I think that conversions and assignation are a special case of transformations that merit special attention. Don't you?
> > Or why we have conversion operators in C++? The question is why these operations need to be defined inside a class in a intrinsic way and not outside in a extrinsic way.

> Implicit conversions are something else entirely. The reason why they
> need to be members is not entirely clear to me, but I'd speculate that
> it has to do with syntax, not semantics.
_______________________________________________

I don't think the syntax could be the problem. We have different syntax for operator+ depending on whether it is used as a member inside a class or as non-member function, but the user uses it in the same way.

BTW, there is an old C++ proposal N2200 "OperatorOverloading" http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2007/n2200.pdf that propose "Allowing overloading of the cast operators + Allowing global overloading of all operators"

If these operator overloads where introduced into the language, we will not need convert_to. The function convert_to tries to replace these explicit and extrinsic conversion operator that are not available now in C++.

Best,
Vicente