From: Thorsten Ottosen (thorsten.ottosen_at_[hidden])
Date: 2007-11-01 14:08:21
Peter Dimov skrev:
> Thorsten Ottosen:
>> Doug Gregor skrev:
>>> On Oct 30, 2007, at 6:22 PM, Thorsten Ottosen wrote:
>>>> Is there any good reason that boost::reference_wrapper() does not
>>>> provide a forwarding operator()(), so that we may use it to pass
>>>> function objects by reference to algorithms?
>>> The only reason is laziness. I hacked up a partial implementation in
>>> the sandbox, but I never finished it.
>> Are you using the preprocessor to generate the overloads?
>> One could probably manually provide up to four arguments ... I assume
>> this is more than enough, since the main use-case would be when passing
>> a function object to an algorithm, in which case the mojaority of cases
>> would be 1 or 2 arguments.
> reference_wrapper::operator() needs result_of, and result_of would likely
> introduce a dependency into Boost.Bind on the PP library, the MPL, type
> traits, and everything underneath. The change may be worth it even at this
> cost, but it's not a no-brainer. It might be nicer if we could devise a way
> to deliver this functionality in a cost-free manner though, perhaps only in
Oh, totally forgot about this. That sucks.
I would be satisfied with a function
template< class Return, class T >
boost::function_obj_ref( const T& );
algo( first, last, boost::function_obj_ref<bool>( my_heavy_object ) );
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk