Question

I’m trying to extract the integer and decimal parts of a floating point value, and I seem to be running into some strange rounding problems, due probably to the imprecise way floats are stored.

I have some code like this to extract the fractional part:

double number = 2.01; int frac = int(floor(number * 100)) % 100; 

However the result here instead of 1 comes out as 0. This seems to be because the original double actually gets stored as:

2.0099999... 

However running sprintf seems to get such a conversion correct:

char num_string[99]; sprintf(num_string,'%f',number); 

How is sprintf getting the correct answer while the above method does not?

Answer 1

> However the result here instead of 1 comes out as one.

What do you mean?

2.099999…

Or, more like 2.00999...

As you’ve noted:

int frac = int(floor(number * 100)) % 100; 

will be:

int frac = int(floor(2.00999... * 100)) % 100;          = int(floor(200.999...)) % 100;          = int(floor(200.999...)) % 100;          = int(200) % 100;          = 200 % 100;          = 0; 

You may be interested in this.

Also, see modf from math.h:

double modf(double x, double *intptr) /* Breaks x into fractional and integer parts. */  

modf() is a better alternative than doing the juggling yourself.