Re: [boost] FW: Suggestion for boost`s smart pointers

From: Howard Hinnant (howard.hinnant_at_[hidden])
Date: 2006-12-01 20:02:21

On Dec 1, 2006, at 7:37 PM, Sohail Somani wrote:

>> -----Original Message-----
>> From: boost-bounces_at_[hidden]
>> [mailto:boost-bounces_at_[hidden]] On Behalf Of Howard Hinnant
>> Sent: Friday, December 01, 2006 4:34 PM
>> To: boost_at_[hidden]
>> Subject: Re: [boost] FW: Suggestion for boost`s smart pointers
>>
>> On Dec 1, 2006, at 5:00 PM, Sohail Somani wrote:
>>
>>> Most C libraries have make_X/free_X pairs. So a typical C call could
>>> look like:
>>>
>>> ...
>>> struct T * c = make_T(args);
>>> do_stuff_with_a_T(c);
>>> free_T(c);
>>> ...
>>>
>>> Which I would like to replace with:
>>>
>>> ...
>>> typedef some_ptr<T,Deleter<free_T> > T_ptr;
>>> T_ptr c(make_T(args));
>>> do_stuff_with_a_T(c.get());
>>> ...
>>
>> Does this work for you?
>>
>>      typedef unique_ptr<T, void (*)(void*)> T_ptr;
>>      int args;
>>      T_ptr c(make_T(args), std::free);  // assumes make_T uses
>> std::malloc
>>      do_stuff_with_a_T(c.get());
>>
>> http://home.twcny.rr.com/hinnant/cpp_extensions/unique_ptr_03.html
>
> I wouldn't really need the extra indirection that seems to be implied
> there but that's a step in the right direction.

<nod> Agreed.  The indirection is in there really only because you  
said "C".  The deleter type can easily be a C++ class type which can  
then do anything you want.

-Howard