97
votes

I need to store a value as a Float, but the source data is a CGFloat:

let myFloat : Float = myRect.origin.x

but this results in the compiler error: 'NSNumber' is not of subtype 'Float'

So if I explicitly cast it like this:

let myFloat : Float = myRect.origin.x as Float

but this in turn results in the compiler error: 'Cannot convert the expression's type 'Float' to 'Float''

What's the correct way to do this and satisfy the compiler please?

3
Be aware that on 64-bit systems, casting CGFloat to Float loses precision – CGFloat is 64 bits on 64-bit systems and 32 bits on 32-bit system, Float is always 32 bits. To avoid this, you could use Double instead of Float.Lukas
if you drill down on a CGFloat in Xcode and see where it is defined you will see that it is typedef'd as a float on 32 bit architecture and a double on 64 bit architecture.jcpennypincher

3 Answers

185
votes

You can use the Float() initializer:

let cgFloat: CGFloat = 3.14159
let someFloat = Float(cgFloat)
31
votes

If you are as lazy as I am, in an Extensions.swift define the following:

extension Int {
  var f: CGFloat { return CGFloat(self) }
}

extension Float {
  var f: CGFloat { return CGFloat(self) }
}

extension Double {
  var f: CGFloat { return CGFloat(self) }
}

extension CGFloat {
  var swf: Float { return Float(self) }
}

Then you can do:

var someCGFloatFromFloat = 1.3.f
var someCGFloatFromInt = 2.f
var someFloatFromCGFloat = someCGFloatFromFloat.swf
13
votes

Usually, the best solution is to keep the type and use CGFloat, even in Swift. That's because CGFloat has different size on 32bit and 64bit machines.

Keyword as can be used only for dynamic casting (for subclasses), e.g.

class A {
}

class B : A {
}

var a: A = B()
var b: B = a as B

However, Double, Int, Float etc are not subclasses of each other, therefore to "cast" you have to create a new instance, e.g.

var d: Double = 2.0
var f: Float = Float(d) //this is an initialiser call, not a cast
var i: Int = Int(d) //this is an initialiser call, not a cast