Re: [boost] move_ptr

From: Daryle Walker (darylew_at_[hidden])
Date: 2004-01-26 15:12:04

On 1/24/04 8:19 PM, "Jonathan Turkanis" <technews_at_[hidden]> wrote:

> "Daryle Walker" <darylew_at_[hidden]> wrote in message
> news:BC3875FB.576B%darylew_at_hotmail.com...
>> On 1/21/04 5:25 PM, "Howard Hinnant" <hinnant_at_[hidden]> wrote:
>> 
>> [SNIP]
>>> With move_ptr<T[]> you (or at least I) immediately think:  Ok, a smart
>>> pointer to an array.
>> [TRUNCATE]
>> 
>> I forgot; will "T[]" resolve to an array type?  In function parameters, it
>> reduces to "T*", which is something unsuitable here if the same thing
>> happens.

> Are you thinking of something like this:
> 
>   template<typename T>
>   void f(T t)
>   {
>       BOOST_STATIC_ASSERT((boost::is_same<T, int* >::value));
>   }
> 
>   int main()
>   {
>       int a[100];
>       f(a);
>   }
> 
> ?

No, I'm not.

> Here 'a' is converted to int* before template argument deduction, so T
> is deduced to be int*. This applies only to function templates, and
> doesn't happen at all if the template argument is explicitly
> specified.

I meant the template and function resolutions of "T[]" to be separate and
parallel analogies, not merged together.  For functions:

whatever function1( T[] t );
whatever function2( T* t );

these are considered the same.  What about:

template < typename U >  class my_template;
//...
template < typename T >  class my_template<T[]> { /*...*/ };
template < typename T >  class my_template<T*> { /*...*/ };

Are those last two entries considered the same?  If so, then we can't use
this technique.

-- 
Daryle Walker
Mac, Internet, and Video Game Junkie
darylew AT hotmail DOT com