# 4 Weeks Developing for iOS

Date: 06 Jun 2018 09:53

A brief introduction: it is currently my summer break after I finished my freshman year in NUS – and I was able to secure an internship at SEA Group (Garena Online) in the iOS team.

I had no experience in owning an Apple device, let alone developing for iOS, so I had to learn a lot of things.

For the first 2-3 weeks I was doing my entry tasks. First 2 days I learnt swift, and tried my best to pick up Objective-C (man, isn't objc syntax weird). Swift is really a pleasant language to work with, especially after having to deal with Java for one whole semester (NUS SoC being a Java school). I also used extensively this very useful website: https://objectivec2swift.com/ to learn the mapping between Swift and Objective-C. After a week+ I think I can say I can read Objective-C code quite well.

Using Apple documentations was a little bit frustrating at first – it is rather example-scarce and the caveats are usually not stated explicitly – this is in comparison with Ruby on Rails/nodeJS documentations that I have been using when I was doing web development. I also can't look at the source code, so sometimes I just cross my fingers and use lldb extensively to test various possibilities that are not documented clearly.

As for bugs, I found some – not sure if those were my mistakes or not. Firstly, MapKit annotation does not show up if the coordinate is (0.00, 0.00). I know, it's in the middle of the ocean, but it should still show up. Changing it to (0.00001, 0.00) makes the annotation shows up. Sigh, Apple, why? Another bug that I found is that when changing the annotation, it would work if I either change the title and subtitle directly, or through this kind of code:

class MyAnnotation: MKPointAnnotation {
var locationName: String {
didSet {
subtitle = locationName
}
}
}


However, it wouldn't work if I used setter/getter instead:

class MyAnnotation: MKPointAnnotation {
var locationName: String {
set(newValue) {
subtitle = newValue
}
get {
return subtitle
}
}
}


I can't be bothered to look at how MKPointAnnotation works under the hood in Objective-C but that was an annoying bug.

And now I have encountered another pecularity. For my current project, I need to create a UITableView with around 400,000 rows. Yes, you read that right, 4 followed by five 0's afterwards. When run in the simulator, no matter where you are, scrolling is always buttery smooth, yet whenever I try it on the real device, scrolling would be smooth on top and lags starting at row 200,000 onwards. This was really weird and tracking down my code yields nothing. I tried to run Time Profiler on the real device, but apparently since XCode 9.3, you can't run time profiler on iPhones below iOS 11.3 (wow, thanks so much Apple!)

Eventually I borrowed my colleague's iPhone X, and voila! the scrolling is as smooth as in the simulator. It was running iOS 11.4, so I immediately went to time profile for later comparison. For testing, I downloaded both XCode 9.2 (which supports my iOS 10 device), and ran time profiler on it. I was suspecting that UITableView rendering is optimised in iOS 11. It's either that, or that iPhone X is so powerful that the scrolling lag simply does not exist. My former hypothesis was correct! There's one symbol that takes up more than 60% of the time when the UITableView starts to lag, which does not exist in the iPhone X. It seems that the rendering logic has been changed in iOS 11. To further confirm this, I used my iOS 9 simulator, and, wow, I could not believe my eyes, it lagged even on the simulator running on the computer. Since this is something internal, I can do nothing :( except asking my future users to upgrade to iOS 11 when they finally encounter this problem.

Here's to 8 more weeks developing for iOS.

Back to blog