Diff compiler == diff program????

This is REALLY starting to bug me. Really.

The folling code, under:

gpp, compiles and removes the tenth integer form a space delimited list of integers in file.txt

Dev-C++, does nothing at all

Turbo, deletes the contents of the file and replaces it with meaningless data.

WHYYYY?????

Is this not ANSI compilant? Anyone????

Code:

/*
   Kth Integer remover

This program reads all integers up to and including the Kth, but no more, from a text file called file.txt. It then eliminates the Kth integer. No other numbers are read or otherwise modified.

Prerequisite: A space delimited integer file called file.txt exists in the cwd.

Note: It may be helpful to, when making the file, make one line of numbers, then another identical line below it. This allows you to better see the removal of the number.
*/


#include <iostream.h>
#include <fstream.h>
#include <math.h>


//This integer is to be removed from the file, bye bye!
#define Kth_integer 10

   int lenof(int);

   int main () {
      bool stop = false;
      int b = 0;
      int inums[Kth_integer];
      ifstream ifile("file.txt", ios::in);
      for (int b = 1; b <= Kth_integer; ++b)  {
         ifile >> inums[b];
      }
      ifile.close();
      ofstream ofile("file.txt", ios::in);
      for (int b = 1; b < Kth_integer; ++b)  {
         ofile << inums[b];
         ofile << ' ';
      }
      for (int x = 0; x < lenof(inums[Kth_integer]); ++x) {
         ofile << ' ';
      }
      ofile.close();
      return 0;
   }

   int lenof (int a) {
      int b = 1;
      while (a/pow(10, b) >= 1) {
         ++b;
      }
      return b;
   }

for starters, bool isn't a standard data type. it's compiler dependant, i think. can anyone confirm this? i remember djgpp didn't accept bools. (or hex notation, for that matter)

Oh, well that ought not to even matter, the boolean sentry value was not used in the program. I'll see if removing it has any effect.

bool is standard C++.

This is bad, your off by one..

int inums[Kth_integer];

for (int b = 1; b <= Kth_integer; ++b) {
ifile >> inums[b];
}

Should start at 0 and go < Kth_integer.

I don't know if this has any effect, but why do you pass in the ios::in flag to an ostream?

Passing an ios::in to an ostream leaves the ostream's put iterator at the very beginning of the file, allowing to to be overwritten, rather than appended to or truncated.

As for skipping the array's zero, I just did that for notational convenience, inums[1] is the first number, and inums[Kth_integer] is the Kth number, rather than being one lower.

You can't change the fundamentals of the langauge..

const int k = 10;
int a[k];
a[k]; // invalid, greatest valid index is k-1

ACH!!!

Thats probably it!!!

Thanks, I'll try that with k+1.

Nope. That as written compiles fine on gpp, and works. If I make is int inums[Kth_integer + 1]; it still compiles and runs fine on gpp, but neither does anything on Dev-C++.

Any borland/msvc users care to take it for a spin? Maybe its just Dev-C++ and Turbo.

im not sure if this will work but try adding ios::app to the open mode, this shouldn't be needed ut its worth a try.

im not up on C++ streams so this is just a guess on a buggy implementation.

::edit:: try also ios::binary the mode can be done differently than text mode on some compilers!

no-one, opening the ofstream with the append flag would place the put pointer at the end of the file, so instead of overwriting Kth digit with blanks, it would append all the numbers up to k to the end. This, unfortunately, would not be effective.

Thanks for the time, though!

What I just can't get is how DevC++ and gpp can treat what I thin ought to be somehwta low level implementation-independant code differently.

Can anyone pick out something from that prog that is implementation-dependant? Prelude?

um seek your way back...is what im saying, i thought you said that turbo fraps the file contents with jibberish, appending hopefully prevents this.

Seeking, perhaps, might be a solution. but then, why does it work under gpp????????

stick with what works i guess...

Problem being that I am trying to prove to someone that this can, in fact, be done. She uses turbo. My little proggie was not that impressive due to the different output.