I use the following construct to create a mpl vector of types.
struct Struct1{
typedef int type;
};
struct Struct2{
typedef float type;
};
template<typename T> struct TypeReader{
typedef typename T::type type;
};
int main()
{
typedef bm::vector<Struct1,Struct2> MPLVector;
typedef bm::transform<MPLVector, TypeReader<bm::_1>>::type TypeList;
static_assert(bm::equal<TypeList, bm::vector<int,float> >::value, "Is not same");
}
So far this works as expected. Now what I would like to do is the following
struct Struct3{
typedef bm::vector<char,double> type;
};
typedef bm::vector<Struct1,Struct2,Struct3> MPLVector;
typedef bm::transform<MPLVector, TypeReader<bm::_1>>::type TypeList;
static_assert(bm::equal<TypeList, bm::vector<int,float,char,double> >::value, "Is not same");
This does not work. So how do I need to change my MetaFunction struct that it will work with both, just a typedef and a mpl::vector?
Or if this is not possible would it be possible to do this if I change all type typedefs to mpl vectors?
I don't think this is possible with mpl::transform
, since it would need to generate several elements in the resulting sequence from a single one in the source sequence. However, this can be done with mpl::fold
, along with specialization with mpl::is_sequence
:
// Default case
template < typename Seq, typename TWrapper, typename Enable = void >
struct Flatten_imp
: mpl::push_back< Seq, typename TWrapper::type >
{
};
// Sequence case
template < typename Seq, typename TWrapper >
struct Flatten_imp<
Seq,
TWrapper, typename
boost::enable_if<
mpl::is_sequence< typename
TWrapper::type
>
>::type
>
{
typedef mpl::joint_view<
Seq, typename
TWrapper::type
> type;
};
template < typename Seq >
struct Flatten
: mpl::fold< Seq, mpl::vector<>, Flatten_imp< mpl::_, mpl::_ > >
{}
int main()
{
typedef mpl::vector< Struct1, Struct2, Struct3 > MPLVector;
typedef Flatten< MPLVector >::type TypeList;
static_assert(
mpl::equal<
TypeList,
mpl::vector< int, float, char, double >
>::value, "Is not same");
}
If you want the flattening to be recursive, you can call Flatten
in Flatten_impl
to flatten the sequence before appending it; however, be careful that this recursion will only work if your sequences contain wrappers and not direct types, e.g.
struct Struct3
{
typedef mpl::vector< mpl::identity< char >, mpl::identity< double > > type;
}