strange exception

[Christian Kirsch]

Hello everybody,

first of all, i want to tell you that i'm a newbie in c++ programming.
I former just used delphi and now i have to learn c++. It's a little
less save than delphi, so i've some difficulties.

But now i'm standing infront of a problem i can't solve on my one. I
tried to build a matrix-class:

class matrix
{
int x,y;
double **field;
double *fieldcol;

public:

matrix(int cols, int rows) //constructor
{
int i;
int j;
x=cols;
y=rows;
field = new double*[x-1];
for (i=0;i<x;i++)
{
field[i]=new double[y-1];
}
for (i=0;i<x;i++)
{
for (j=0;j<x;j++)
{
field[i][j]=0;
}
}
}
}
In the main program i initialize 2 variables:

matrix a(2,2);
matrix b(2,2);

When i now try to run the program i get a windows exeption. But when i
start the program in debugging mode, where i can jump through the
program, it works without any problem. Also when i only initialize a
without b, the program works fine. Or when i initialize both variables
with the parameters (1,1), it also works without any trouble.

Has anybody an idea? I use the Editor Dev-C++ coming along with an
gnu-compiler, both for windows xp.

Thanks for your help!

Chris

[Rolf Magnus]

Christian Kirsch wrote:

Hello everybody,

first of all, i want to tell you that i'm a newbie in c++ programming.
I former just used delphi and now i have to learn c++. It's a little
less save than delphi, so i've some difficulties.

But now i'm standing infront of a problem i can't solve on my one. I
tried to build a matrix-class:

class matrix
{
int x,y;
double **field;
double *fieldcol;

public:

matrix(int cols, int rows) //constructor
{
int i;
int j;
You should always give variables the smallest scope possible. So remove the
definition here, and below, in your for loop, just do:

for (int i = 0; i < x; ++i)
x=cols;
y=rows;
field = new double*[x-1];
Why one less?
for (i=0;i<x;i++)
{
field[i]=new double[y-1];
}
for (i=0;i<x;i++)
Here, you go past the end of the array, since you are accessing x elements,
but only have space for x-1 elements.
{
for (j=0;j<x;j++)
Same here.
{
field[i][j]=0;
}
}
}
}

[Phlip]

Christian Kirsch wrote:

double **field;
double *fieldcol;
Grab a book called /Accelerated C++/, and then don't use * there. Use
std::vector<double>. Then the Standard Library will manage the low-level
memory concerns for you.
matrix(int cols, int rows) //constructor
{
x=cols;
y=rows;
field = new double*[x-1];
Now field contains cols-1 elements, not cols elements.
for (i=0;i<x;i++)
{
field[i]=new double[y-1];

And now you use the element at field[cols-1], which exists (for a secret
reason) but which cannot be assigned to.

When you corrupt memory, the program will exhibit "undefined behavior". That
term will now play the leading role in your nightmares as you learn C++.

"Undefined behavior" means the program could crash, could limp along and
then crash, could appear to work correctly, or could make the nearest toilet
explode. So you when you run the program two different ways, you see two
different effects.

Use new..[cols], and switch to vector as soon as possible.

--
Phlip
http://c2.com/cgi/wiki?ZeekLand <-- NOT a blog!!!

[Andre Kostur]

On Fri, 12 May 2006 07:04:41 -0700, Christian Kirsch wrote:

Hello everybody,

first of all, i want to tell you that i'm a newbie in c++ programming. I
former just used delphi and now i have to learn c++. It's a little less
save than delphi, so i've some difficulties.

But now i'm standing infront of a problem i can't solve on my one. I tried
to build a matrix-class:

class matrix
{
int x,y;
double **field;
Why not a std::vector<std::vector<double> > ?
double *fieldcol;
And this isn't ever used.

public:

matrix(int cols, int rows) //constructor {
int i;
int j;
x=cols;
y=rows;
field = new double*[x-1];
And if you call create a matrix of (1, 1), you new 0 doubles? I think
your "x - 1" is wrong.
for (i=0;i<x;i++)
{
field[i]=new double[y-1];
Now I know your x-1 is wrong. You've already indexed off the end of your
dynamic memory. Undefined Behaviour. Anything can happen.
}
for (i=0;i<x;i++)
{
for (j=0;j<x;j++)
Uh, shouldn't that by "j < y" ?
{
field[i][j]=0;
}
}
}
}

And the times that it did "work"... were just flukes.

[Christian Kirsch]

>Now I know your x-1 is wrong. You've already indexed off the end of your

dynamic memory. Undefined Behaviour. Anything can happen.

Well my thought was this: I want x elements and the index starts with 0
so i have to decrease x one time. Or does new double [5] create an
array with 5 elements 0 to 4? Then of course x-1 should be replaced...
I'll try...

And thats my largest problem, in delphi the debugger tells me that i'm
out of range so I'm a little spoiled in that case ;-)

Thanks!
Chris

[Christian Kirsch]

>Why not a std::vector<std::vector<double> > ?

;-), because it was the exercise to do it that way, don't ask why...

Thanks Chris

[Phlip]

Christian Kirsch wrote:

Why not a std::vector<std::vector<double> > ?

;-), because it was the exercise to do it that way, don't ask why...

Regardless what your professor requests, also start learning the Standard
Library.

I would finish each assignment, once exactly as requested, and again using
the best style possible.

--
Phlip
http://www.greencheese.us/ZeekLand <-- NOT a blog!!!

[doina]

field = new double*[x-1];
--> this means that you want to allocate x-1 elements double*.

for (i=0;i<x;i++)
{
field[i]=new double[y-1];
your "i" is iterating from 0 to x-1 (including) --> x elements.
But you only allocated x-1 so you probably get an index out of bounds
exception because you're trying to access the element field[x-1] which
doesn't exist.

So to summarize your problem is that in c/c++ the index in an array
starts at 0 (not 1) and when you say field = new double* [x-1] this
means allocate x-1 elements starting from 0.

[Andre Kostur]

On Fri, 12 May 2006 07:41:33 -0700, Christian Kirsch wrote:

Now I know your x-1 is wrong. You've already indexed off the end of your
dynamic memory. Undefined Behaviour. Anything can happen.
Well my thought was this: I want x elements and the index starts with 0 so
i have to decrease x one time. Or does new double [5] create an array with
5 elements 0 to 4? Then of course x-1 should be replaced... I'll try...

Yep counting in C++ (learned from C) starts at 0. So double x[5] declares
an array containing 5 items... numbered from 0 to 4.

And thats my largest problem, in delphi the debugger tells me that i'm
out of range so I'm a little spoiled in that case ;-)

But... there's a run-time cost to that bounds checking. General C++
philosophy... don't pay for what you don't use.

[Christian Kirsch]

Yes, that was the problem. I knew, that the index starts with 0, the
thing I didn't know was that you actually have the length of the array
as new - parameter. I thought it would be the Index of the very last
item. So that was the problem.

Thanks to all of you!!!

Chris