compare double value to 0 - unreliable comparisons warning

Hey - Quick one for ya...

I get the warning message below at compile time because i'm trying to compare a double value to 0.

What should I use to compare it? I'm working on a matrix multiplier so thought a cast would be a slow, if not accuracy losing method of doing it.

Quote:

---

warning #1572: floating-point equality and inequality comparisons are unreliable

---

Code:

---

double alpha;
...
 if ((m == 0) || (n == 0) || (((alpha == 0) || (k == 0)) && (beta == 1)))

---

Elementary stuff no doubt but in my defence it's getting late....

Thanks,

Colly.

When you use floats and doubles, because of the way they work, the variable does not store the exact value. So 3 / 2 may not be exactly 1.5, it might be 1.50000001 or something. This is why inequality and equality comparisons are unreliable with them. I'm afraid an alternative doesn't come to mind. It's been a loong time since I worked with matrices.

To further the point, when using floating point numbers, don't compare to an integer. Compare to a float. Not that it's really going to help this, but for correctness sake, you want this instead:

Code:

---

if( myfloat == 0.0 )

---

The only way to really have accuracy is to skip floating point numbers all together.

Quzah.

http://www.eskimo.com/~scs/C-faq/q14.5.html