Re: [c++0x] unique_ptr

["Chris Fairles" <chris dot fairles at gmail dot com>]

On Wed, Jun 11, 2008 at 2:57 AM, Paolo Carlini <paolo.carlini@oracle.com> wrote:
>  Hi,
>
>> It seems that there are some changes required in shared_ptr
>> to work with
>> unique_ptr, as in:
>>
>> std::unique_ptr<int> uptr(new int(5));
>>
>> std::shared_ptr<int> sptr = uptr;
>> assert((bool)uptr == false);
>>
>> This way unique_ptr can substitute auto_ptr nicely.
>
> Yes. Actually, we have open quite a few shared_ptr proper issues too, I don't think should delay in any sense the unique_ptr work.
>
>> Also, maybe Paolo can confirm this, but there's an active issue
>> #740 by
>> Herb Sutter regarding the static array partial specialization.
>> They're
>> planning not to provide this partial specialization.
>>
>> Reference:
>> http://home.twcny.rr.com/hinnant/cpp_extensions/issues_preview/lwg-active.html#740
>
> Good point. The status of that issue is Ready, thus we can just immediately scratch the specializations (and maybe add a comment about that in the code, our usual practice with DRs)

Removed specializations for array objects with compile time length.
Comments added (btw - whats up with the _GLIBCXX_RESOLVE_LIB_DEFECTS ?
I put it in because I see it with other DR notes, but I'm not sure
what its used for).

Also added the beginnings of a test suite. exlicit_instantiation.cc
and assign.cc (although, assign still tests the compile-time length
which i do believe is redundant now).

>
>> > I've been using your unique_ptr implementation for a while now
>> and it's
>> been working great so far.
>
> Excellent, thanks a lot for your feedback!
> Paolo.
>
>

Chris

Attachment: unique_ptr.patch
Description: Binary data