Quick Intro question to smart pointers.

[Doug Mika]

I have the following code snippet:

std::shared_ptr<my_data> p;
void process_global_data()
{
std::shared_ptr<my_data> local=std::atomic_load(&p);
process_data(local);
}

what confuses me is why we don't write

std::shared_ptr<my_data> local=std::atomic_load(p);

after all, p is already a pointer, won't getting an address of p (&p) lead to me having a pointer to a pointer to my_data?

Thanks

[Alf P. Steinbach]

It's just the design of `atomic_load` is imperfect, that it expects a
pointer rather than a reference as argument.

You can look up standard library functions in any decent reference, e.g.
<url: http://en.cppreference.com/w/cpp/atomic/atomic_load>.

That said, the code shown appears to be very meaningless, and exhibiting
several Unholy Worst Practices™. So whatever source you got it from is
not a good source of examples to learn from.


Cheers & hth.,

- Alf

--
Using Thunderbird as Usenet client, Eternal September as NNTP server.

[Bo Persson]

On 2015-08-13 09:16, Alf P. Steinbach wrote:
> On 13-Aug-15 4:18 AM, Doug Mika wrote:
>> I have the following code snippet:
>>
>> std::shared_ptr<my_data> p;
>> void process_global_data()
>> {
>> std::shared_ptr<my_data> local=std::atomic_load(&p);
>> process_data(local);
>> }
>>
>> what confuses me is why we don't write
>>
>> std::shared_ptr<my_data> local=std::atomic_load(p);
>>
>> after all, p is already a pointer, won't getting an address of p (&p)
>> lead to me having a pointer to a pointer to my_data?
>
> It's just the design of `atomic_load` is imperfect, that it expects a
> pointer rather than a reference as argument.

The interface was supposed to be compatible with C, and a pointer is all
they have.


Bo Persson

[Alf P. Steinbach]

* 13-Aug-15 5:33 PM, Bo Persson:
> * 2015-08-13 09:16, Alf P. Steinbach:

> >
> > It's just the design of `atomic_load` is imperfect, that it expects a
> > pointer rather than a reference as argument.
>
> The interface was supposed to be compatible with C, and a pointer is all
> they have.

Well, 2 things wrong with that idea:

* The function is in namespace `std`, no such in C.

* There's no problem with the same function having binding as foo(T&) in
C++ and as foo(T*) in C; that's what language bindings are all about,
offering an interface that's suitable for the language.