On 09/27/2012 01:42 AM, John Bytheway wrote:
On 25/09/12 22:38, Mario Mulansky wrote:
John Bytheway <jbytheway+boost@gmail.com> wrote:
The more general idea of automatically dispatching to the right algebra is more dangerous, because for example it is entirely possible to have a type which is both a Fusion sequence and a Range (but with different semantics in each interpretation). So, I am not suggesting that (though I wouldn't argue against it either).
This would be an edge case where explicit user definition is required, but i think in 99% an algebra dispatcher would be sufficient.
On further reflection I thought of a particular case where this happens. Any std::pair<Iterator, Iterator> is both a Fusion Sequence and a Range, with very different semantics. Just something to consider.
Having a range as state_type is difficult maybe even impossible, since ranges can not be resized. A range models something that already exists and inside the stepper a temporary must be created. (But a range can be passed to the do_step method.) Nevertheless, it is possible the create a dispatcher like template< class State , class Enabler = void > struct algebra_dispatcher; template< class T , class A > struct algebra_dispatcher< std::vector< T , A > , void > { typedef range_algebra type; }; // this will create errors for ranges of pairs of iterators template< class R > struct algebra_dispatcher< R , typename boost::fusion::is_sequence< R >::type > { typedef fusion_algebra type; }; The steppers simply get the value of the algebra dispatcher as a default parameter: template< class State , ... , class Algebra = typename algebra_dispatcher< State >::type , ... > struct stepper { ... }; Then, the user will mostly get the right algebra, and in case this is not possible he can easily specify the desired one. (The same should be done for the operations.)