Re: [boost] [scope_guard] New revision available + naming consideration

From: Daryle Walker (darylew_at_[hidden])
Date: 2005-09-23 06:41:28

On 9/21/05 9:38 PM, "David Abrahams" <dave_at_[hidden]> wrote:

> Ariel Badichi <abadichi_at_[hidden]> writes:
> 
>> David Abrahams wrote:
>>>>> it's easy to arrange that only a literal zero works there.
>>>> 
>>>> How?
>>> 
>>> Allow assignment only from a private member pointer type.  It's
>>> similar to the operator safe_bool idiom.
>> 
>> I thought you might mean that. That allows more than a literal zero,
>> though (e.g., 5 * 5 - 25).
> 
> Yikes!  I  didn't know that.  Too bad!

Maybe it's not as "too bad" as you two think.  You can use more than a
literal zero for the conventional null-pointer too!  Any integral expression
that can be resolved at compile time that evaluates to zero can be used:

    void *  x = 20 - 22 + 2;  // works as NULL pointer!

See section 4.10, paragraph 1 of the standard.  (Also see s4.11-p1 and
s5.19-p1.)  So allowing this case is not any more surprising than the rules
for built-in pointers.

>> By the way, that might work well for smart
>> pointers; why don't shared_ptr and friends use this idiom?
> 
> Maybe because it allows more than a literal zero?

If that's the reason, maybe it could be re-evaluated.

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