Access violations and lost PBD files

[Sp_ceNerd]

I have tried writing a code to print Pascals triangle of n size. In it I have used pointers and bidimensional arrays which I am not very skilled with.When I run it I get two error messages, but never at the same run. It complains about access violations (I put the error message next to the code that it was aimed at) which probably depends on that I've used either arrays or pointers in the wrong way. The error messages about lost PBD files, or files that can not be opened, I don't actually know what mean.
I'd be greatful for any help at all. You could just point me in the right direction. (I don't know if the code does what I want it to but I can't get it to run si I wouldn't know if it doesn't.

#include <iostream>
#include <new>
using namespace std;

int setUp(){
int input;

	cout << "This program will let you create a Pascals triangle of prefered size." << endl;
    cout << "Enter the amount of rows you wish the triangle to have: ";
    cin >> input;
    cout << endl << endl;

return input;
}

int triangle(int wid){
int i, j, k;
int **p, row = wid, col = wid;

	p = new int* [row]; //create rows of triangle
	for (int row = 0; row < wid; row++)
		p[row] = new int[col]; //create columns of triangle

	for (i = 0; i < wid; i++)
		p[0][i] = 1;

	for (i = 0; i < wid; i++)
		p[i][0] = 1; //fix
		cout << p[i][0] << endl; //fix
		/*
		Error message: Unhandled exception at 0x00da171b in pascalsT.exe: 0xC0000005: Access violation reading location 0xfdfdfdfd.
		It also complains about not being able to find or open a bunch of PBD files half the time as well. Have I saved the file in the wrong way somehow?
		*/

		/* 
		This could come in handy:
		https://msdn.microsoft.com/en-us/library/6decc55h.aspx on fixing an access violation
		and learn how to ****ing use and print ****ing arrays man. Both one and multidimensional
		http://www.cplusplus.com/doc/tutorial/pointers/ (also has a chapter on arrays)
		*/
	
	for (i = 1; i < wid; i++){
		for (j = 1, k = i; i < wid; i++){
			p[j][k] = p[j][k-1] + p[j][k-1];
			cout << p[j][k];
		}
			cout << endl;
	}

return 0;
}

int main(int){
int width = setUp();
	triangle(width);

return 0;
}

[GCDEF]

This is what your debugger is for. Use it to look at the value of i at your cout statement. Perhaps you meant to have cout included as part of your for loop.

[ahoodin]

Maybe your talking about pdb files? Program database files? Try rebuilding your project and debugging as just suggested.

[TubularX]

Raw pointers are the cause of many errors. Learn how to avoid them!

Here you can use std::vector instead. It can help you by showing assertions or throwing exceptions when something goes wrong, which makes it much easier to find errors.

You should also add some extra checks, for example to make sure that the input is a value greater than 0.

If you mean PDB files, then it's probably just a warning. Maybe some library you depend on was built without debug information?

[ahoodin]

There is no doubt that raw pointers cause problems, but they still like to see them in academia as well as at interviews apparently. The extra shared pointer part just adds confusion to the professor or the interviewer. Don't get me wrong, in practice, best use those shared pointers. I know I do.

[GCDEF]

The OP is stepping over the end of the array. No need to confuse things with discussions of the evils of pointers.

[TubularX]

The OP is stepping over the end of the array. No need to confuse things with discussions of the evils of pointers.

Regarding the question about access violation, then yes, that's all there's to it.

But there are lots of wrong things in that code, for example memory leaks. I'm just trying to point the topic starter in the right direction.

It's unfortunate that people still learn to write such error-prone code. Long functions with lots of loops and pointers, that can often be rewritten in just a few lines of code without any runtime overhead. Then bugs can't hide as easily!

I believe it's for the best to push people to write good code from the start. Well, maybe this isn't the right place to do so, but a few lines of recommendations can't hurt.

[ahoodin]

Additionally, pdb files contain debugging and state information,
so if these files have problems, that may lend to trouble debugging
his project. Hence the re-build.

[2kaud]

to print Pascals triangle of n size

If the requirement is to just print a triangle of a given number of rows, then all you really need are two arrays - one to hold the previous row and one to hold the row being computed. The row being computed can be calculated from the previous one. Unless it is a requirement of the assignment to use pointers and new(), then as TublarX said in post #4 it is better to use vectors. See http://www.cplusplus.com/reference/vector/vector/

[ahoodin]

I concurr on the vectors, because of the issues surrounding shared pointers and the use of arrays. You have to create a custom deleter to pass to the shared pointer constructor. This is not really a brand new beginners bag of code with array, but vector could fix that.

[Sp_ceNerd]

Thank you for the help. I will look into making the code tidier now that I can see what it prints.
Would it fix the memory leak issues if I put delete[] p; at the end of triangle()? If that isn't all that needs to be done I would really appreciate if you could give me a keyword that I could use when searching for ways to further improve it

[Sp_ceNerd]

There aren't any specific requirements on what to use, it's just that we haven't gone through vectors on the lessons so arrays felt more at home. This assignment isn't actually on the course, it's just something thrown in to keep the ones a bit further away busy and I guess looking into new things. So I can do pretty much all I want with it.
I hadn't even thought about what I need to keep with me when calculating the new values. It's actually a good idea. Wouldn't have me in over my head with pointers and multidimensional arrays

[2kaud]

we haven't gone through vectors on the lessons

At its simplest, vectors can be used like arrays where the size can be set at run-time rather than compile time. Once a vector has been defined with a size then it can be used like an array - but without the issues re new/delete. Consider

#include <vector>
using namespace std;

int main()
{
	constexpr int v_size = 10;

	vector<int> vec(v_size);

 	vec[0] = 34;
	vec[v_size - 1] = 56;
	vec[1] = vec[0] + 10;
}

vec is a vector of size v_size that will hold values of type int. It has v-size number of elements which start at 0 just like arrays. v_size can be set at run-time. Once defined with a size then vec can be used like an array. There's much more to vectors than just using them like an array but for your task this'll do fine. If you want more info have a look at the link I posted in my post #9. Happy coding!

[2kaud]

Would it fix the memory leak issues if I put delete[] p; at the end of triangle()?

You're on the right lines but no, not quite. For each row you have allocated memory for the columns with new[], so you need to iterate again the rows and use delete[] to free this column memory. Then you use delete[] again to free the memory allocated for the rows. Vectors are so much less hassle/worry to use!