Re: [boost] Re: SmartPtr (Loki) - auto_ptr/move c'tor issue

From: Greg Colvin (Gregory.Colvin_at_[hidden])
Date: 2003-01-28 12:48:25

At 10:08 AM 1/28/2003, Andrei Alexandrescu wrote:
>"David B. Held" <dheld_at_[hidden]> wrote in message
>news:b16a0k$7sv$1_at_main.gmane.org...
>> "Peter Dimov" <pdimov_at_[hidden]> wrote in message
>> news:009f01c2c6d7$91024ab0$1d00a8c0_at_pdimov2...
>> > [...]
>> > The first question, of course, is: do you really need SmartPtr<...> to
>> > support move semantics (in current C++)?
>>
>> Why wouldn't you want that?  At the very least, it seems like a glaring
>> omission to create a smart pointer framework that can't even emulate
>> auto_ptr<>.  Beyond that, it seems that there are resources that would
>> benefit from or outright require move semantics to work properly, and
>> why wouldn't you want to let SmartPtr<> manage those?
>
>I think what Peter refers to is that C++ might change to make move semantics
>easier to implement. That would render the effort unnecessary. If smart_ptr
>is to be proposed for standardization, the committee can just as well
>package the new smart_ptr together with new language features, notably move
>semantics and template typedefs which would fit smart_ptr like a glove.

Also, auto_ptr is an ugly hack that needn't be replicated.