Re: [Boost-users] [parameter] Interface questions

From: Cromwell Enage (sponage_at_[hidden])
Date: 2006-07-05 10:35:27

--- Daniel Wallin wrote:
> Cromwell Enage wrote:
> > 1.  Will BOOST_PARAMETER_IMPL(name) expand to
> > the name of the function that actually "owns"
> > the main body of code?
> 
> No, or at least is doesn't at this point.

Which macro should I use then?  For better or for
worse, some of my code needs to pass ArgumentPacks
directly to other Boost.Parameterized functions.

> > 2.  Will there be a way to define a
> > Boost.Parameterized member function outside
> > the class definition?  Or should we let the
> > compiler decide whether or not to inline such
> > a function?
> 
> It is possible, but it's complicated by the
> return-type calculation. For example, the
> expansion-tail of:
> 
>   BOOST_PARAMETER_FUNCTION(
>     (void), f, tag,
>       (required (x, (float))
>       (optional (y, (float), 0))
>   )
> 
> is something like:
> 
>   template <
>       class ResultType
>     , class Args
>     , class x_type
>     , class y_type
>   >
>   ResultType boost_param_default_21f(
>       ResultType(*)()
>     , Args const& args
>     , x_type& x
>     , y_type& y
>   )
> 
> right now. We could improve the name and get
> rid of the __LINE__ from it, but it's hard (or
> impossible) to get rid of ResultType argument.

That looks painful.  It's okay, I don't need this
particular feature right now.

                              Cromwell D. Enage

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com