On 13.10.2010 16:15, Andrew Sutton wrote:
On the topic of abstraction, I would say that Matt's solution seems to be the right approach -- from an iterator perspective. He claims that segmentation concepts is an orthogonal to the standard iterator concepts. I don't especially like the implementation, requiring an explicit specialization of segmented_iterator_traits. I see nothing preventing the segmented iterator class from defining those associated types and operations and allowing the traits class to use them by default.
Easy enough. Your class just has to contain a type called is_segmented, definition irrelevant. But there is no way around having such a marker. You don't want to trigger off the classical iterator tag, since that would mean twice the number of tags. BOOST_MPL_DEFINE_HAS_MEMBER(is_segmented) template <typename T, bool Enable = has_is_segmented<T>::value> struct segmented_iterator_traits { typedef false_type is_segmented_iterator; }; template <typename T> struct segmented_iterator_traits<T, true> { typedef true_type is_segmented_iterator; typedef T iterator; typedef typename T::segment_iterator segment_iterator; // ... }; Sebastian