How do you truncate unneeded floating point precision in Swift?

Question

I'm building a calculator app. When I press the % button on the calculator, it multiplies the current value by 0.01.

If I do 52 * 0.01, I'll get 0.52. But if I press it again, it will calculate 0.52 * 0.01, and I get 0.005200000000000001.

This value gets displayed to the user of the app. How do I truncate the extra numbers to end up with 0.0052 so that it makes more sense?

Use `Decimal` instead of `Double` – New Dev Aug 01 '20 at 02:06 — New Dev, Aug 01 '20 at 02:06
That's the solution, thanks. – Makoren Aug 01 '20 at 03:03 — Makoren, Aug 01 '20 at 03:03

score 0 · Answer 1 · answered Aug 01 '20 at 03:02

0

The answer is to use Decimal instead of Double or Float.

Make sure the calculation is done between Decimals too, meaning you shouldn't try to do a calculation between two Doubles, and cast the result to a Decimal.

answered Aug 01 '20 at 03:02

Makoren

321
1
3
12

Besides using Decimal type you need to make sure to only use its string initializer. – Leo Dabus Aug 01 '20 at 06:07

Patel Saurabh · Answer 2 · 2020-08-01T16:21:44.603

0

let formatterUpperBreakPoint = 999999999.9
let formatterLowerBreakPoint = 0.000000001
let maxInput = 9

let decimalFormatter = NumberFormatter()
decimalFormatter.numberStyle = .decimal
decimalFormatter.groupingSeparator = ","
decimalFormatter.maximumIntegerDigits = maxInput
decimalFormatter.maximumSignificantDigits = maxInput + 1

let number = 0.005200000000000001

if abs(number) < formatterUpperBreakPoint && abs(number) > formatterLowerBreakPoint {
    print(decimalFormatter.string(from: NSNumber(value: number)) ?? "0") // result 0.0052
}

edited Aug 01 '20 at 16:21

answered Aug 01 '20 at 03:12

Patel Saurabh

56
2

OP has no idea how many fraction digits the result might be – Leo Dabus Aug 01 '20 at 06:09
I wouldn’t recommend using double for a calculator app – Leo Dabus Aug 01 '20 at 14:33
@LeoDabus, I edited my previous answer, Thanks for guide. – Patel Saurabh Aug 01 '20 at 16:36
No need to instantiate a `NSNumber` obkject. You can use Formatter's method `string(for: Any)` – Leo Dabus Aug 01 '20 at 16:38
Btw `Decimal` would support a much broader precision. It uses 20 bytes instead of 8 and it doesn't suffer from the same issues as the FloatingPoint types. Just make sure to always use its string initializer. Check this post https://stackoverflow.com/a/55010456/2303865 – Leo Dabus Aug 01 '20 at 16:40

How do you truncate unneeded floating point precision in Swift?

2 Answers2