CODEWITHSUNDEEP
c++
daleotar
daleotar

Reputation: 325

C++ (int) static cast weird error

I have a very weird error with c++. I have two values, max and singleStep. The number of steps is stepsInt and/or stepsDbl:

max = 100;
singleStep = 0.1;
// This means that I have 100/0.1 = 1000 numbers betwwen 0 and 100
double stepsDbl = max/singleStep;
int stepsInt = (int)(stepsDbl);
cout << stepsDbl << stepsInt;

You can expect an output like this:

1000 1000

However I'm getting this output:

1000 999

What is more, if I try a different value for singleStep, for instance 0.2, I get again wrong values

500 499

I don't know what is happening but is pretty weird... If anyone has a solution for this problem I would appreciate that solution.

Thanks

Upvotes: 0

Views: 126

Answers (3)

Blindy
Blindy

Reputation: 67380

Try this instead:

int stepsInt = (int)floor(stepsDbl+.5);

The "problem" you're seeing is because of the way floating point numbers are stored internally.

Upvotes: 0

Kerrek SB
Kerrek SB

Reputation: 477010

Floating point numbers on your platform cannot exactly represent numbers which do not have a finite binary expansion, so you will naturally have uncertainty when using those numbers. Use rationals instead and keep "0.1" as "1/10" when possible.

("Finite binary expansion" means "finite sum of powers of two". 1/5 is not a finite sum of powers of 2, and neither is 1/3 nor 1/10.)

Upvotes: 0

geekosaur
geekosaur

Reputation: 61369

(int) casts round toward zero, and (double) arithmetic is inherently unreliable (look up floating point math). operator<<(double) rounds to nearest, so a value like 999.999999999999 resulting from accumulated FP inaccuracy will print as 1000 directly and 999 when naïvely cast to (int).

Upvotes: 1

Related Questions