No errors, but wont run...

Hi everyone :)

Im a real newbie to this, so i guess ive prob done soemthing stupid...

My program has be working fine, and i was testing that various parts of it worked, and it was goingn ok. I had to restart the compiler/editor (borland c++ 3) and now when i try to tun the program it starts then quits instantly! :mad:

If i switch to the output screen it says "Abnormal Program Termination".

Its been working fine for ages, any ideas what could be wrong?

Post your code

its 1000 lines long....

isnt that a bit much to post :confused:

Yes, that's a bit much
But what's a newbie doing with a 1000 line program?

First thing, does it run outside the IDE? It may simply be running out of memory.

Are you for example assuming that malloc() always succeeds?

Quote:

---

Originally posted by Salem
Yes, that's a bit much
But what's a newbie doing with a 1000 line program?

First thing, does it run outside the IDE? It may simply be running out of memory.

Are you for example assuming that malloc() always succeeds?

---

Ive been programming for 4 months now but i still think im a newbie... Its my major coursework piece for college so thats why tis longish.

Nope, it wont work outside the IDE.

whatis malloc()? ive never used it before.

malloc() allocates memory based on the size given.

Eg --

char *t = malloc(sizeof(char)*5);

t would have enough space for 5 chars...I think :p :D

Have you got a debugger? Perhaps you should get one if you don't!

I use Borland Turbo C++ 3.0 (Old DOS based program).

It has everything in one, a debugger, compiler and editor all in one.

no errors are reported at all....

Well if you're going to post that much code, then I suggest

1. Make sure you've indented the code properly. See the samples in the FAQ for some typical examples of well indented code.

2. Use the "Attach file" at the bottom of the reply page to attach a copy of your source.

3. Don't expect quick miracles for such large amounts of code.

4. Get a better compiler. The standard of C++ which Turbo C++ 3.0 uses is quite far removed from what is current practice.
Try www.compilers.net and find Dev-C++

Are you sure you're not doing a return 1 somewhere in main(), which would make the debugger think it's not ended right?
Also, if the compiler gives you warnings, try to fix them, because they may have become fatal.
Just some suggestions, not sure what could cause a program to act that weird.

ive been looking at an older version of my code that i had saved an comparing the differences, i seem to have managed to fix it. :D

Thank you, for trying to help me, sorry if i have wasted your time.


BTW i have to use Turbo C++ 3.0, my teacher says i have to... :(

ok, sorry about this, ive jsut stumbled on the exact casue of the problem.

It appears when i insert this line:

Code:

---

float price;

---

into this struct:

Code:

---

struct car 
{
char company_name[30];
char regAB[3];
char reg01[3];
char regXYZ[4];
long int mileage;
char type_of_tyre[10];
int no_of_tyres;
int dd;
int mm;
int yyyy;
float price;//this is the one that makes it go wrong, it runs without it
char comment[50];

};

---

:confused:

any ideas why this could be?

Old DOS compilers have always been a bit funny about detecting when floating point support is required or not.

If you just have a declaration and something like scanf(), then it usually gets it wrong, and reports some "floating point not linked" message.

The hacky fix seems to be to declare a dumm global variable like

Code:

---

double fix_my_broken_compiler = 1.0;

---

> BTW i have to use Turbo C++ 3.0, my teacher says i have to
Yeah, and what does that tell you about the quality of the teacher?
Don't tell me, they use void main() as well.

ok, ill try that thanks...

btw i use cout<<, not scanf()

and yeah we used void main() (whats wrong with that) :confused:

nope that variable didnt fix it...

> and yeah we used void main() (whats wrong with that)
See the FAQ

> nope that variable didnt fix it...
I think you're going to have to attach the whole file then.

ok, thanks for your help, ive fixed it now :D

I had declared it like this:

Code:

---

car car_details[300];

---

but when i reduced it to:

Code:

---

car car_details[285];

---

it worked.:)

Thats one of those stange bugs that indicates a buffer overflow or uninitialized variables somewhere. I've once achieved something similar to this by using

Code:

---

for (i; i<10; i++) {
  get some memory here
}

---

without ever initalizing i. The bugs you get afterwards are strange and change when changing the program, sometimes they seem fixed but then they pop up again. Until you find the real cause.

Salem might remember this.

> car car_details[285];
Each car is just over 100 bytes
printf( "sizeof car = %lu\n", (unsigned long)sizeof(car) );

All cars are going to be 100*285, which is about 29K bytes.
printf( "sizeof all cars = %lu\n", (unsigned long)sizeof(car_details) );

My guess is, that adding all your other global variables takes this up to the max allowed by 16 bit compilers of 64K of data.

The way you get round this is by
1. choosing the large model (not small)
2. putting global variables in different source files (then you get a max of 64K per variable), not 64K in total.

I repeat my earlier comment - GET A DECENT COMPILER!.
Modern 32 bit compilers have the same limit, but its like 2GB, not 64KB
There is no possible justification for continuing to use that old fossil.

> float price;//this is the one that makes it go wrong, it runs without it
Only because it saved you a K or so of data space.