program problem...

[ashesh]

I have this small C program and I am unable to locate the mistake. Can someone help me locate it?

#include<stdio.h>
 int main()
      {
           float a=0.0,b=0.0;
           float a1=0.0;
           a1=1.05;
           b=0.05;
           printf("\na1=%f\tb=%f\n",a1,b);
           a=a1-1.0;
           printf("\na=%f\tb=%f\n",a,b);
           if(a<b)
              printf("\nHELLO:b is greater than a\n");
           else
              printf("\ncondition is not satisfied");
           printf("\n");
       }

The program is supposed to print "Condition is not satisfired", but instead the program gives "HELLO:b is greater than a"

Hope someone can help me locate the error.

Thanks in advance.

[quzah]

You mean other than the fact that you forgot how to indent, and that you've neglected to pay attention to your compiler's warnings?


Quzah.

[ashesh]

Oops.. sorry abt that...was in a hurry when I posted that. Will take care next time and indent.

There were no warning given by the compiler. I executed the program on Fedora 1 gcc compiler and there were no warnings or error messages.

[^xor]

Just a recommendation...

alias gcc="gcc -Wall -Wextra -ansi -pedantic"

[Salem]

> Hope someone can help me locate the error.
Yeah, your assumption that floating point numbers are totally accurate is the problem.

Mathematically, you expect a and b to be the same, but in floating point calculations, there is always a small degree of error.
So what you think might be a == b could result in any of these conditions being true
a < b
a == b
a > b

http://www.physics.ohio-state.edu/~d...point_math.pdf

[ashesh]

Thanks Salem. It turns out that the assumption was only the problem. When I printed up to 20 digits after the decimal point, I could see the difference in the two numbers.