«

I am surprised to find out why the destructor function does not test
pointer before memory is deleted.

struct Object
{
Object( int n ) : ptr( new char[ n ] )
{
}

~Object()
{
// comment if condition
// if( ptr != NULL )
delete [] ptr;
}

char *ptr;
};

Any C++ books say you must always include if condition to test
pointer if it is not zero before memory is allowed to be deallocated.
I found out that delete function already has if condition. The
delete function returns zero without deallocating memory if pointer is
zero.

On Sep 8, 8:56*am, Nephi Immortal <immortalne...@gmail.com> wrote:
> * * * * I am surprised to find out why the destructor function does not test
> pointer before memory is deleted.
>
> struct Object
> {
> * * * * Object( int n ) : ptr( new char[ n ] )
> * * * * {
> * * * * }
>
> * * * * ~Object()
> * * * * {
> * * * * * * * * // comment if condition
> * * * * * * * * // if( ptr != NULL )
> * * * * * * * * * * * * delete [] ptr;
> * * * * }
>
> * * * * char *ptr;
>
> };
>
> * * * * Any C++ books say you must always include if condition totest
> pointer if it is not zero before memory is allowed to be deallocated.

Throw all your C++ books away then.

> * * * * I found out that delete function already has if condition.. *The
> delete function returns zero without deallocating memory if pointer is
> zero.

Yep.

On Sep 8, 5:05*pm, Noah Roberts <roberts.n...@gmail.com> wrote:
> On Sep 8, 8:56*am, Nephi Immortal <immortalne...@gmail.com> wrote:
>
>
>
>
>
> > * * * * I am surprised to find out why the destructor function does not test
> > pointer before memory is deleted.
>
> > struct Object
> > {
> > * * * * Object( int n ) : ptr( new char[ n ] )
> > * * * * {
> > * * * * }
>
> > * * * * ~Object()
> > * * * * {
> > * * * * * * * * // comment if condition
> > * * * * * * * * // if( ptr != NULL )
> > * * * * * * * * * * * * delete [] ptr;
> > * * * * }
>
> > * * * * char *ptr;
>
> > };
>
> > * * * * Any C++ books say you must always include if condition to test
> > pointer if it is not zero before memory is allowed to be deallocated.
>
> Throw all your C++ books away then.
>
> > * * * * I found out that delete function already has if condition. *The
> > delete function returns zero without deallocating memory if pointer is
> > zero.
>
> Yep.

Besides, I thought new couldn't return NULL anyway?

On 9/8/2011 2:52 PM, Paul N wrote:
> On Sep 8, 5:05 pm, Noah Roberts<roberts.n...@gmail.com> wrote:
>> On Sep 8, 8:56 am, Nephi Immortal<immortalne...@gmail.com> wrote:
>>
>>
>>
>>
>>
>>> I am surprised to find out why the destructor function does not test
>>> pointer before memory is deleted.
>>
>>> struct Object
>>> {
>>> Object( int n ) : ptr( new char[ n ] )
>>> {
>>> }
>>
>>> ~Object()
>>> {
>>> // comment if condition
>>> // if( ptr != NULL )
>>> delete [] ptr;
>>> }
>>
>>> char *ptr;
>>
>>> };
>>
>>> Any C++ books say you must always include if condition to test
>>> pointer if it is not zero before memory is allowed to be deallocated.
>>
>> Throw all your C++ books away then.
>>
>>> I found out that delete function already has if condition. The
>>> delete function returns zero without deallocating memory if pointer is
>>> zero.
>>
>> Yep.
>
> Besides, I thought new couldn't return NULL anyway?

A placement form with 'std::nothrow' can.

V
--
I do not respond to top-posted replies, please don't ask

On Thu, 2011-09-08, Paul N wrote:
> On Sep 8, 5:05*pm, Noah Roberts <roberts.n...@gmail.com> wrote:
>> On Sep 8, 8:56*am, Nephi Immortal <immortalne...@gmail.com> wrote:

>> > * * * * I am surprised to find out why the destructor function does not test
>> > pointer before memory is deleted.
>>
>> > struct Object
>> > {
>> > * * * * Object( int n ) : ptr( new char[ n ] )
>> > * * * * {
>> > * * * * }
>>
>> > * * * * ~Object()
>> > * * * * {
>> > * * * * * * * * // comment if condition
>> > * * * * * * * * // if( ptr != NULL )
>> > * * * * * * * * * * * * delete [] ptr;
>> > * * * * }
>>
>> > * * * * char *ptr;
>>
>> > };
>>
>> > * * * * Any C++ books say you must always include if condition to test
>> > pointer if it is not zero before memory is allowed to be deallocated.
>>
>> Throw all your C++ books away then.
>>
>> > * * * * I found out that delete function already has if condition. *The
>> > delete function returns zero without deallocating memory if pointer is
>> > zero.
>>
>> Yep.
>
> Besides, I thought new couldn't return NULL anyway?

A standard new cannot. But there are often other valid reasons 'ptr'
may be null at the time the destructor is called. In the class above,
a single 'obj.ptr = 0' will do.

/Jorgen

--
// Jorgen Grahn <grahn@ Oo o. . .
\X/ snipabacken.se> O o .

On 9/8/2011 8:56 AM, Nephi Immortal wrote:
>
> Any C++ books say you must always include if condition to test
> pointer if it is not zero before memory is allowed to be deallocated.

I'm not aware of any of such books. Must be some garbage book. All
standard memory deallocation facilities perform a this test internally,
which is why you shouldn't normally do it yourself.

> I found out that delete function already has if condition.

Delete function? What "delete function" are you talking about?

> The
> delete function returns zero without deallocating memory if pointer is
> zero.

Er... "Returns zero"? I don't know what "delete function" you are
referring to, but any potential candidates I know of don't return
anything at all.

--
Best regards,
Andrey Tarasevich