Re: [boost] [smart_ptr] const-correctness as function argument

From: Larry Evans (jcampbell3_at_[hidden])
Date: 2002-10-02 12:47:24

scleary_at_[hidden] wrote:

>>-----Original Message-----
>>From: Larry Evans [mailto:jcampbell3_at_[hidden]]
>>
>>I haven't looked real close, but doesn't the subsection titled
>>"Smart Pointers and const" in Item 28 of Scott Meyers
>>_More Effective C++_ contain a solution to this problem?
>>Basically, it describes SmartPtr<T>  derived from
>>SmartPtrToConst<T> and SmartPtrToConst<T> containing
>>a union of const and non-const pointers to T.  Unfortunately,
>>this would require changing shared_ptr.
>>    
>>
>
>That item is concerned with allowing a conversion of smart_ptr<T> to
>
But this is what D.Held wants, isn't it.  To be more explicit, let 
shared_ptr_const
be the analog to SmartPtrToConst.  Then define:

    void g (shared_ptr_const<foo>&p);

then:

    shared_ptr<foo> a_foo_ptr;
     g(a_foo_ptr);

would work as I think D.Held wants.

>  The Standard solves the same problem in auto_ptr by the
>use of templated constructors/assignment.  Boost solves that problem in
>shared_ptr in essentially the same way as the Standard.  
>
I think Held objects to this because Held wants to avoid "any needless 
copies".
With the current shared_ptr, wouldn't the shared_ptr<foo> have to be
converted to shared_ptr<foo const> via a copy constructor?