C programming good practice or not ? - c

Question

I have written a small piece of code to calculate quadratic equations, but if the discriminant is negative, i wanted it to write that there are no real numerical values for this quadratic equation.
To make this happen, I had to call a function with a fourth parameter of 0, which i think , i have no idea why, would be a bad programming practice ? Is it the case or am i just being too picky of my code ? Thank you. (The reason I'm asking this is because i dont want to pick up some bad habits early on in my programming 'career').
Here's the code.
#include <stdio.h>
#include <math.h>
#include <string.h>
double quadratic_equation(double a, double b, double c, double d);
int main(void)
{
char command[20];
int i;
printf("Enter your command: ");
fgets(command, 20, stdin);
for (i = 0; i < 20; i++) {
if (command[i] == '\n') {
command[i] = '\0';
break;
}
}
if (strcmp(command, "quadratic equation") == 0) {
double a, b, c, x;
printf("Enter A: ");
scanf("%lf", &a);
printf("Enter B: ");
scanf("%lf", &b);
printf("Enter C: ");
scanf("%lf", &c);
x = quadratic_equation(a, b, c, 0); // THIS PIECE HERE MIGHT BE BAD PRACITCE ?
if (x == 0) {
printf("There are no real numerical values to this quadratic equation.");
}
else {
printf("------------\n");
printf("x1 = %.2f\n", quadratic_equation(a, b, c, 1));
printf("x2 = %.2f", quadratic_equation(a, b, c, -1));
}
}
return 0;
}
double quadratic_equation(double a, double b, double c, double d) {
double discriminant, x, insideroot;
insideroot = ((b*b) - (4*a*c));
if (insideroot < 0) {
return 0;
}
discriminant = sqrt(insideroot);
x = (-b + (d * discriminant)) / (2 * a);
return x;
}
Thank you very much for your help :d !

Answer 1

This certainly is bad practice. Since the roots of a formula a, b, and c an be any double you do need some sort of passing.
I would suggest a parameter that is a pointer to an int. If the pointer is NULL it is ignored, otherwise it will be set to 1 or 0 depending whether a real root exists:
double quadratic_equation(double a, double b, double c, int *root_exists) {
double discriminant;
discriminant = ((b*b) - (4*a*c));
if (discriminant < 0) {
if (root_exists != NULL) *root_exists = 0;
return 0.0;
}
x = (-b + sqrt(discriminant)) / (2 * a);
if (root_exists != NULL) *root_exists = 1;
return x;
}
A more rigorous approach is this:
typedef struct {
int num_roots;
double roots[2];
} quadratic_roots_t;
quadratic_roots_t quadratic_equation(double a, double b, double c) {
quadratic_roots_t roots;
double d;
d = b*b - 4*a*c;
if (d < 0.0) {
roots.num_roots = 0;
} else if (d == 0.0) {
roots.num_roots = 1;
roots.roots[0] = -b / (2 * a);
} else {
roots.num_roots = 2;
roots.roots[0] = (-b - sqrt(d)) / (2 * a);
roots.roots[1] = (-b + sqrt(d)) / (2 * a);
}
return roots;
}

Answer 2

I'd say it's not great to do. What you could do is something like this:
int quadratic_equation(double a, double b, double c, double *root_a, double *root_b) {
double discriminant = ((b*b) - (4*a*c));
if (discriminant < 0) {
return -1;
}
if (root_a != NULL) {
*root_a = (-b + sqrt(discriminant)) / (2 * a);
}
if (root_b != NULL) {
*root_b = (-b - sqrt(discriminant)) / (2 * a);
}
return 0;
}
Then you could call that like so:
double root_a;
double root_b;
int ok = quadratic_equation(a, b, c, &root_a, &root_b);
if (ok < 0) {
// It wasn't OK. Print out an error.
} else {
// It was OK. Print out the results.
}
Note that you should also check other error cases in the function and return -1 for them as well. E.g. a being zero.

Answer 3

Consider using the return value to indicate whether everything worked, and passing an array to the function to receive the return values:
enum QE_Status { QE_OK = 0, QE_NON_QUADRATIC, QE_COMPLEX_ROOTS, QE_NULL_POINTER };
enum QE_Status quadratic_equation(double a, double b, double c, double *r)
{
double discriminant;
if (r == 0)
return QE_NULL_POINTER;
if (a == 0.0)
return QE_NON_QUADRATIC;
discriminant = (b * b) - (4 * a * c);
if (discriminant < 0)
return QE_COMPLEX_ROOTS;
discriminant = sqrt(discriminant);
r[0] = (-b + discriminant) / (2 * a);
r[1] = (-b - discriminant) / (2 * a);
return QE_OK;
}
You can extend the system to handle numerical instability (because b*b is almost equal to 4*a*c, or because a is very small, etc).
The calling code can then be:
double a, b, c, x[2];
if (quadratic_equation(a, b, c, x))
...oops, something went wrong...
Or:
switch (quadratic_equation(a, b, c, x))
{
case QE_OK:
...print or use results in x...
break;
case QE_NON_QUADRATIC:
case QE_COMPLEX_ROOTS:
...print appropriate error message about user's data...
break;
case QE_NULL_POINTER:
...Oops - programming error...
break;
}

Answer 4

I would certainly call it bad practice because the code is very unclear.
Firstly, you are calling the function three times, when once should be enough.
I'd consider returning/filling a list in your quadratic_equation() function instead of returning the roots one by one.
This would also allow you to determine if there are real roots -- if there aren't, just return an empty list.
As a whole, this would be much more elegant than your current solution, and it would eliminate the need for checking whether there are any solutions beforehand.