boost::bind vs boost::mem_fun (was RE: [Boost-users] Re: shared_ptrs with std::mem_fn)

From: Mark Storer (MStorer_at_[hidden])
Date: 2004-05-10 12:04:12

> > template<class _R, class _Ty>
> > 	class mem_fun_t : public unary_function<_Ty *, _R> {
> > public:
> > 	explicit mem_fun_t(_R (_Ty::*_Pm)())
> > 		: _Ptr(_Pm) {}
> > 	_R operator()(_Ty *_P) const
>                   ^^^^^^^
> > 		{return ((_P->*_Ptr)()); }
> > private:
> > 	_R (_Ty::*_Ptr)();
> > 	};
> > ---
> >
<...>

> No, the relevant part is the signature I highlighted above.

<...>

> 
> All I have to say is "use boost::mem_fn, which works, and never
> touch the one in std::".

Thanks for clearing that up, it was really bugging me.  

So to speak. ;)

I've recently started using boost::bind, and I think I'm in love.  But boost::mem_fun raises another question in my mind:  Are there any cases where "boost::bind( &foo::bar, _1 )" is superior to "boost::mem_fun( &foo::bar )" (other than a few less keys to press)?  I'd say that boost::bind is clearly supperior when "bar" takes a parameter, but when it does not, the water gets a little muddy.

I'd imagine that you wouldn't get the long stream of error messages (expected N params not Y, expected N + 1 params, not Y, expected N + 2 ...) from bind when all the different variants miss, but is there anything else?

--Mark Storer
  Senior Software Engineer
  Verity, Inc.