[boost] Re: ptrdiff_t and smart pointers

From: Beman Dawes (beman_at_[hidden])
Date: 1999-10-11 09:05:32

Ed Brey wrote:

>Valentin Bonnard wrote:
>> 
>> Ed Brey wrote:
>> 
>> > I just noticed a minor oddity: for scoped_array and
shared_array,
>> > operator[] takes an
>> > argument of type int.  Any reason it isn't type std::ptrdiff_t?
>> 
>> You do mean size_t, don't you ?
>> (and yes int is incorrect)
>
>I meant ptrdiff_t, but you are right it should be size_t.  I was
>thinking ptrdiff_t because that is what's used in section 13.6/13,
>where operator[] is defined for built-in types.  But I forgot that
for
>built-ins, the index can be negative, whereas it can't for our smart
>arrays.
>
>Indeed, std::vector uses size_type, and so should we.  Only for our
>case, I don't think it is worth typedefing a size_type within
>scoped_array or shared_array, so we should just use std::size_t, as
>you indicated.

Originally, int rather than std::size_t was used as a workaround for
the BOOST_NO_STDC_NAMESPACE problem.  Now that we have
BOOST_NO_STDC_NAMESPACE, smart_ptr.hpp has been change to
std::size_t.

Thanks Ed and Valentin.

--Beman

PS: Ed's bio and picture are now up on the web site.