Heap corruption with dynamic memory

I am using visual studio running a win32 console application.
I'm new to dynamic memory and i am trying to think of some extra things for an easy assingment. Basically I am trying to allow an input for the array size. I know this requires dynamic memory and i have it set up for that. I get an error after the program finishes with
" Windows has triggered a breakpoint in (my program).

This may be due to a corruption of the heap, which indicates a bug in (my program) or any of the DLLs it has loaded."
It then brings me to a breakpoint in the ostream, if i run it with 2 as my input, it gives me a different error.

Here's my code.

Code:

#include <iostream>
using namespace std;
// ------------------Function Prototypes -----------------------------
void EnterData (int *grades, int,int&);
void OutputData (int *grades, int,int);
//--------------------------------------------------------------------
void main()
{	
	int x;
	int *grades;
	cout<<"How many students would you like to calculate? ";

	cin>>x;;

	grades=new int(x);
	int total=0;
	int	NUMSTUDENTS=x;
	

	// enter the grades information
	EnterData (grades,NUMSTUDENTS, total);

	// display the grades information
	OutputData (grades,NUMSTUDENTS,total);
	delete grades;
	system ("pause");

}// end main
//--------------- Function definitions to follow --------------------------
void EnterData(int *grades, int NUMSTUDENTS, int &total)
/*	This function allows the user to enter the grades information into the data array.
	Pre: The grades array has been allocated sufficient memory
	Post: The grades entered by the user are returned within the array	*/
{
	int studentNumber;
	
	
	cout<<"Please enter the grades for the "<<NUMSTUDENTS<<" students...\n\n";
	for (studentNumber = 0; studentNumber<NUMSTUDENTS; studentNumber++)
	{
		cout<<"Enter grade #"<<studentNumber+1<<": ";//counter, inputs the data
		cin>>grades[studentNumber];//into each array component starting at zero
		total+=grades[studentNumber];//total for finding average
	}
}// end EnterData
//--------------------------------------------------------------------------
void OutputData(int *grades, int NUMSTUDENTS, int total)
/*	This function displays the grades to the user
	Pre: The grades array has already been filled with data
	Post: The grades are displayed to the user				*/
{
	int studentNumber;

	cout<<"The grades for the "<<NUMSTUDENTS<<" students are:\n\n";
	for (studentNumber = 0; studentNumber<NUMSTUDENTS; studentNumber++)
	{
		cout<<"Student #"<<studentNumber+1<<": ";
		cout<<grades[studentNumber]<<endl;
	}
	cout<<"The average is: "<<double((total/NUMSTUDENTS))<<endl;
}// end OutputData

Thanx for any help. Hope its not to easy.

Code:

grades=new int(x);

This allocates one int with the value of x.

totally just realized this as you replied. Why does it run anyways though? I mean, shouldn't it give me a compile error if i try to assign values to an array segment that doesn't exist?

Why shouldn't it? If you have a pointer, you need to know yourself how many items it points at. C++ is not supposed to check your pointer accesses at runtime.

Because the compiler does not know what you do at runtime; it only knows what you do at compile-time.
Also consider using std::vector and reading http://www.research.att.com/~bs/bs_faq2.html#whitespace and http://cpwiki.sourceforge.net/Void_main.

Originally Posted by anon

Why shouldn't it? If you have a pointer, you need to know yourself how many items it points at. C++ is not supposed to check your pointer accesses at runtime.

He's suggesting that it shold be able to detect array access, for something not allocated as an array, at compile time. This could perhaps have been possible if C++ had different types of pointers, one for a pointer to a single object, and one for a pointer to an array, but the language doesn't do that.

But boost::array triggers an assert if you try to write or read a non-existing element in the array, though. I find that of help.

Originally Posted by Elysia

But boost::array triggers an assert if you try to write or read a non-existing element in the array, though. I find that of help.

Yes that's all well and good, but again that's not at compile time. He's wondering why it doesn't fail to compile due to using (x) instead of [x], and the answer is that the language just doesn't make it possible to detect that as any kind of error, because both are perfectly valid and useful in their own way.

Sure, using something like a vector makes it impossible to make this exact mistake though.

In theory, the compiler could attempt to remember what is going on with allocated memory and how large it is. But this would ONLY work for situations where the number of objects allocated with new is a constant - in the above case that is true, but the absolutely most common case of using new to allocate memory and then using the pointer as an array is when the number of elements is variable. And that case is much harder for the compiler to understand.

Code-checkers, such as lint and coverity are sometimes able to follow these things better, but certainly do not catch EVERYTHING.

--
Mats