Double is rounding my last NSNumber decimal

Question

For example

                    print(message.timestamp!)

gives me: 1470938337.572778

                    print(Double(message.timestamp!))

gives me: 1470938337.57278

I need to use the double to convert to NSDate like this:

date: NSDate(timeIntervalSince1970: Double(message.timestamp!))

but I also need the timestamps to be accurate why is it rounding?

It seems to be working just fine http://swiftlang.ng.bluemix.net/#/repl/57acc49a603420ce5f482bb3 — Alexander, Aug 11 '16 at 18:32
No, it's an NSNumber, It's really confusing why it's doing that — slimboy, Aug 11 '16 at 18:39
@AlexanderMomchliov I tried it with 6 decimal places and it rounded to 5 — slimboy, Aug 11 '16 at 18:44
@slimboy What input did you try, what output did you get, and what did you expect? — Alexander, Aug 11 '16 at 18:46
@AlexanderMomchliov using the values you offered yourself on: [link] (swiftlang.ng.bluemix.net/#/repl/57acc49a603420ce5f482bb3) I get Input: let input: NSNumber = 1470938337.572778 Output: 1470938337.57278 The last 6th decimal is rounded. I need to use the timestamp as an ID so it must remain 6 when converting to double — slimboy, Aug 11 '16 at 18:50
@slimboy Take a careful look at [this.](http://swiftlang.ng.bluemix.net/#/repl/57acc9bf603420ce5f482bb7). The literal value being assigned to `input` is `1470938337.572778`. What's printed (even whilst still in `NSNumber` form) is `1,470,938,337.57278`. — Alexander, Aug 11 '16 at 18:55
This is a case of exceeding the precision representable by an IEEE754 Double width floating point data type. The issue isn't in the `NSNumber` -> `Double` conversion (which makes sense, because `NSNumber` is a wrapper for a `Double`). — Alexander, Aug 11 '16 at 18:56
Let us [continue this discussion in chat](http://chat.stackoverflow.com/rooms/120735/discussion-between-alexander-momchliov-and-slimboy). — Alexander, Aug 11 '16 at 18:58

score 1 · Accepted Answer · edited Aug 11 '16 at 21:03

Take a look at this sample code:

import Foundation

let input: NSNumber = 1470938337.572778

print("   NSNumber: \(input)")
print("     Double: \(Double(input))")
print("doubleValue: \(input.doubleValue)")

Output:

   NSNumber: 1,470,938,337.57278 //not even this is printing correctly.
     Double: 1470938337.57278
doubleValue: 1470938337.57278

This is a case of exceeding the precision representable by an IEEE 754 Double width floating point data type. The issue isn't in the NSNumber -> Double conversion (which makes sense, because NSNumber is a wrapper for a Double).

Double is rounding my last NSNumber decimal

1 Answers1