Open Radar

 
Summary:
Using math functions such as ceil() causes compilation errors when used within extensions of CGGeometry types (CGRect, CGSize etc) and built against iOS simulators or in fat binaries.

Example:

An extension to CGSize to convert a size to a specified aspect ratio and ensure it falls on an exact multiple value of that ratio would be very useful. The code below provides such a conversion, returning a new CGSize struct with the desired aspect ratio.

extension CGSize {

    func normalisedSizeWithAspectRatio(ratio:CGSize) -> CGSize {
        let base = ceil( self.width / ratio.width );
        return CGSizeMake(base * ratio.width, base * ratio.height)
    }

}

This code will always cause a compilation error in one way or another if used in an iOS project. When compiling for the 64-bit simulators (iPhone 5s, iPad Air) you need to use the ceil() function, as CGFloat is defined as a Double. When compiling for 32-bit simulators you need to use the ceilf() function as CGFloat is defined as a Float. Compiling for devices always fails one way or the other.

I remain convinced that if I throw enough casts at the problem I will be able to make it work eventually, but I have not found the solution and the code this will cause me to end up writing would be highly obfuscated.

Steps to Reproduce:
Add the attached file to an iOS project and compile for device, 32-bit simulator and 64-bit simulator

Expected Results:
The same code should be able to compile on all three target devices

Actual Results:
Code using ceil() will compile for 64-bit, code using ceilf() will compile on 32-bit. 

Version:
Xcode 6 beta 1