[C++-sig] Re: Injected constructors

[Nicodemus]

David Abrahams wrote:

>"Ralf W. Grosse-Kunstleve" <rwgk at yahoo.com> writes:
>
>  
>
>>--- David Abrahams <dave at boost-consulting.com> wrote:
>>    
>>
>>>>Another idea: could we expand the init<> interface so that it would
>>>>accept an optional factory function, allowing the user to write:
>>>>
>>>>.def(init<>(x_factory))
>>>>        
>>>>
>>>I thought of that; do you think it would be confusing because of the
>>><>s?
>>>      
>>>
>>I like this best if it is a possibility.
>>    
>>
>
>I resisted it for a long time I guess it is a real possibility.  I was
>going to say that it imposes a small overhead for:
>
>  .def(init<>())
>
>because that init object needs to include refcount management for a
>possible factory.  If you don't like that overhead, though, you can
>just let class_<X>("X") give you a default init function without the
>overhead.
>
>Sweet.
>
>Nobody thinks it's a case of trying to pack too much functionality
>into one object?  If not, so much the better.
>

Yeah, I agree with Ralf, since it is similar with the current way to 
export constructors. And I don't think that it's the case of trying to 
pack too much functionality, since the purpose of init<> is to export 
constructors... the factory overload would be just another method to 
accomplish the same thing.