bcc32 compiler warning when assigning null to pointer

When i compile this code:

Code:

---

#include <iostream.h>

int main()
{
unsigned short *pnt=new unsigned short;
 if (pnt==0)
 {
 cout<<"Not enough free memory available for pnt.";
 return 0;
 }
*pnt=7;
cout<<*pnt;
delete pnt;
pnt=0;                //HERE IS THE PROBLEM
return 0;
}

---

I get a warning from my borland bcc32 5.5 compiler
that says: 'pnt' is assigned a value that is never used in function main()

I was told to take warnings seriously but the program works, and you should always assign pointers to point to null after calling delete on them. So...

...CAN I IGNORE THIS WARNING???

short answer: yes.

borland's just letting you know that you might have forgotten a step because you never use that variable again. but in reality you're just being cautious.

if you should ever alter that piece of code and make a mistake, you would know immediately if your pointer was a null pointer because they automatically trigger error messages. i would use NULL instead of zero, though. it makes the code look better, imho.

Actually you dont need to assign the pointer pnt to null (in this case) because the program terminates anyway after it.

I would think that after you assign pnt to 0 it expect you to use is later in your program and because you don´t the compiler issues a warning.

>I was told to take warnings seriously
Indeed you should if you don't know whether or not the warning is simply blowing smoke.

>you should always assign pointers to point to null after calling delete on them
This is a good habit to get into, but you've found one of the flaws in securing your code. Some compilers will warn you unnecessarily about unreferenced values. So if you have one of these compilers, you have to choose between the security of NULLing out a pointer and clean compiles in case real problems exist. Of course, in some cases there's no need to NULL out the pointer if the program is going to terminate shortly and the pointer will not be used again.

>...CAN I IGNORE THIS WARNING???
Yes.

>i would use NULL instead of zero, though. it makes the code look better, imho.
Perhaps, but because of C++'s stricter type checking, you tend to have fewer problems by using the integer constant 0.

-Prelude

Thanks y'all for your help.