[boost] Re: boos::mpl::for_each: call function with return value

From: David Abrahams (dave_at_[hidden])
Date: 2003-11-03 21:52:32

"Sebastian Faust" <sfaust_at_[hidden]> writes:

> Hi,
>
>>The simplest way would be something like this:
>>
>>    namespace mpl = boost::mpl;
>>    using namespace mpl::placeholders;
>>
>>    struct pass_through_make
>>    {
>>        static vehicle* make( vehicle* ptr )
>>        {
>>            return ptr;
>>        }
>>    };
>>
>>    template< typename T, typename Base > struct make_vehicle
>>    {
>>        static vehicle* make( vehicle* ptr )
>>        {
>>            return new T( Base::make( ptr ) );
>>        }
>>    };
>>
>>    typedef mpl::fold<
>>          mpl::vector<car,bicycle,train>
>>        , pass_through_make
>>        , make_vehicle<_2,_1>
>>        >::type factory;
> I tested the code above but my compiler gives me the following errors:
>
> c:\projekte\boost-1.30.2\boost-1.30.2\boost\mpl\aux_\preprocessed\msvc70
> \apply.hpp(97): error C2039: 'boost::mpl::apply<F,T1,T2,T3,T4,T5>' : is
> not a member of 'make_vehicle<T,Base>'
>  
> c:\projekte\boost-1.30.2\boost-1.30.2\boost\mpl\aux_\preprocessed\msvc70
> \apply.hpp(54): error C2039: 'boost::mpl::apply<F,T1,T2,T3,T4,T5>' : is
> not a member of 'pass_through_make'
>  
> Sadly I am not that familiar with mpl-programming and I can't find in
> the documentation information about the _1, _2 templates. So it would be
> nice if you could give me any further hints.

When you're using a broken compiler you need to resort to slightly
uglier code.  Try this definition of make_vehicle:

   struct make_vehicle
   {
    template< typename Base, typename T > 
    struct apply
    {
        static vehicle* make( vehicle* ptr )
        {
            return new T( Base::make( ptr ) );
        }
    };
   };

Now use make_vehicle instead of make_vehicle<_2,_1>

> Furthermore, where can I learn more about the mpl? Is there any
> literature? Cause I am really getting fascinated by this
> compile-time programming, which I sadly never recognized before.

We're writing a book; in the meantime you could take a look at our
paper at http://www.boost.org/libs/mpl/doc/index.html, but it's a bit
rough and not terribly complete.  There's also the reference docs, of
course...

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com