[boost] Re: Interest in smart pointer with value semantics?

From: David B. Held (dheld_at_[hidden])
Date: 2004-01-05 04:10:36

"Dan W." <danw_at_[hidden]> wrote in message
news:btb4ff$gmk$1_at_sea.gmane.org...
> [...]
> class x
> {
>     y deep_const *a;
> public:
>     y* get_a();                         //boom!
>     y const * get_a() const;            //boom!
>     y deep_const * get_a() deep_const;  //ok
>     y const * get_get_a() const
>     {
>         get_a();  //boom!
>     }
>     y deep_const * get_get_a() deep_const
>     {
>         get_a();  //ok
>     }
> };
> [...]

Ok, so how would this apply to our smart_ptr<> example?  Maybe:

smart_ptr<T> deep_const p;

>From what I can tell, this would require us to define deep_const
accessors inside smart_ptr<> and/or do a deep_const_cast<>
of the raw pointer.  But then, that doesn't seem to be any
different from implementing deep const manually.  Are you
suggesting that the language *enforce* deep const, but not
actually *help* you implement it?  At least const propagates
to aggregated members, so it is both enforcing a property and
helping you out at the language level.  But from what I can see
above, you're just asking for enforcement, with a parallel
const concept.  What I was thinking is that if you declare an
object deep_const, it will make all pointers inside that object
const/deep_const as well, without you ever having to declare
deep const versions of anything; just like declaring an object
const makes the members effectively const.  So it's not y or
y* that we want to make deep_const.  It's an instance of x.
What we really should be asking ourselves is how

x deep_const X;

is implemented/behaves.

Dave

---
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.556 / Virus Database: 348 - Release Date: 12/26/2003