Re: [boost] Re: scoped_ptr: trivial dtor?

From: David Abrahams (abrahams_at_[hidden])
Date: 2001-04-02 20:20:06

----- Original Message -----
From: "Beman Dawes" <bdawes_at_[hidden]>
To: <boost_at_[hidden]>; <boost_at_[hidden]>
Sent: Monday, April 02, 2001 7:04 PM
Subject: Re: [boost] Re: scoped_ptr: trivial dtor?

> At 12:08 PM 3/30/2001, Greg Colvin wrote:
>
>  >> > > Ah, wonderful. We should put this riff in boost/utility.hpp
>  >> > >
>  >> > > template <class T>
>  >> > > void delete_(T const volatile* x)
>  >> > > {
>  >> > >     sizeof(T);
>  >> > >     delete x;
>  >> > > }
>  >> > >
>  >> > > Then "delete x" becomes boost::delete_(x)
>  >> >
>  >> > Yes.  Of course some compilers might complain about
>  >> > the apparently useless line, so perhaps
>  >> >
>  >> >     if (sizeof(T))
>  >> >        delete x;
>  >> >
>  >> > Regardless, a comment is in order.
>  >>
>  >> Good idea.
>
> Even though I don't think it applies to scoped_ptr, I think it would be a
> nice little addition to boost/utility.hpp.
>
> But the name ought to be a bit more explicit, so that readers will know by
> sight why it is used.  delete_complete_type() or something.

Why? Doesn't everybody assume they are deleting the complete type when they
use built-in delete?

> Hum...  Exactly when would you use it?  Presumably mostly in
> templates.  How does a template know if the delete requires a complete
> type?

I would use it /always/ in place of normal delete. I know somebody out there
thinks it's valuable to be able to delete pointers to things whose
definition isn't available, but I always forget that compilers are required
to let me get away with this extremely dangerous practice.

-Dave