目标 C 问题与舍入浮点数

Question

I have an issue with rounding the result of a calculation to two decimal places.

It is a financial calculation and when the result involves half a penny I would expect the number to be rounded up but it is in fact being rounded down.

To replicate the issue:

float raw = 16.695;
NSLog(@"All DP: %f",raw);
NSLog(@"2 DP: %.2f",raw);

Returns:

All DP: 16.695000
2 DP: 16.69

Whereas I would expect to see:

All DP: 16.695000
2 DP: 16.70

Can anyone advise if this is by design or if (most likely) I am missing something and if there is anything I can do to get around it. It is vital that it rounds up in this scenario.

Thanks in advance, Oli

Answer 1

Don't use float s for financial calculations!

There are values that floats cannot represent exactly . 16.695 is one of them. When you try to store that value in a float, you actually get the closest representable value. When you perform a series of operations on a float, you can lose precision , and then you lose accuracy. This leads to losing money.

Use an actual currency type, use NSDecimalNumber , or do all your calculations with int s that count the smallest unit you care about (ie, 1050 is $10.50 in whole cents, or $1.050 if you want fractions of pennies).

Answer 2

As far as I am aware the NSLog() function only takes formatting arguments and makes no attempt to round.

You may be use to using a printf() style function that does support rounding.

I suggest using one of the many functions in math.h to round your value before output and only rely on NSLog() for formatting.

Answer 3

After seeing the comments

Use the C standard function family round(). roundf() for float, round() for double, and roundl() for long double . You can then cast the result to the integer type of your choice

Answer 4

you could try this:

 - (void)testNSlogMethod {

    float value = 16.695; //--16.70
    value = value *100;

    int mulitpler = round(value);

    NSLog(@"%.2f",mulitpler/100.0f);
}