Rewriting SpeakerClock in SwiftUI | Cocoanetics

After I began out creating iOS apps, 11 years in the past I put a number of apps on the App Retailer. Since they turned fewer and fewer because the earnings from them didn’t warrant updating them. Amongst these my most profitable one was iWoman, which I offered in 2015. My second-most-valuable (when it comes to income) remained my beloved SpeakerClock, the final app standing.

I had left SpeakerClock on-line for the primary purpose that it saved producing like a mean of $100 per 30 days, even with out me doing something on it. For that purpose, I didn’t need to make it free, however quite put it to a comparatively excessive price ticket of $5. There may be additionally an In-App-Buy of one other $5. I figured “why kill the cow whereas it nonetheless produces some tasty milk”.

The opposite aspect impact of those value tags was that – I consider – solely individuals who actually wished what the app was providing would truly buy it. My philosophy with this talking timer was to have the largest LED digits attainable, with the performance that helps the talking model of TED Talks, which traditionally have defaulted to a most size of 18 minutes.

Some crashes launched by new iOS variations brought about me to do small bug fixing releases (for iOS 3 in 2010, iOS 5 in 2011, and 2017 for iOS 10). Additionally, trying again on the launch notes of these variations, I had made this precise promise:

“We’ve completely modernised the code base in order that we will convey you some thrilling new options within the subsequent main launch”

However I didn’t lie with this assertion, a “subsequent main” launch would have been model 2.0. However I didn’t ever dare to show the model quantity up that prime. I solely elevated the third digit of the model quantity.

Apple did pressure me to do a brand new construct finally, once they cracked down on apps which weren’t up to date in too lengthy a time. And the newest replace they did themselves, when the Apple certificates had expired and so they re-signed my app on their servers with out me doing something.

Enter SwiftUI

Over the previous couple of months, I’ve grown very keen on SwiftUI. Being a developer on Apple platforms for greater than a decade made me fairly uninterested in having to maintain writing the identical MVC code numerous occasions. And that may solely get you want normal performance, nothing really thrilling. So I jumped on the probability when one in every of my shoppers requested me to implement a brand new iOS Widget in SwiftUI, within the fall of 2020. Apple had turned to SwiftUI as the one manner you might create such widgets due to SwiftUIs capacity to provide and protect a static view hierarchy which the system might present to the consumer at sure factors in a timeline with out substantial energy utilization.

My consumer was pleased in regards to the end result and so I used to be tasked with the subsequent degree of SwiftUI growth. I wanted to implement a watchOS app, additionally solely in SwiftUI. Improvement was fairly just like the widget, however this time I additionally wanted to take care of consumer interplay and communication with the iOS counterpart app. That each one took some a couple of months greater than the widget, however once more elevated my SwiftUI expertise tremendously.

After having delivered the watch app, I had a bit of further time accessible to do one thing for myself. I do have another concepts for apps, however my ideas turned to SpeakerClock. I figured that this extremely customized UI would lend itself properly to be carried out in SwiftUI.

Paths in Shapes

A very powerful asset within the legacy code was the drawing of the large pink LED digits and the way they prepare themselves in portrait versus panorama, in a pleasant animation. So my first SwiftUI view was one which had a Path aspect with the SwiftUI instructions including the trail components to make up the person bars of the LED. My first error right here involved utilizing a GeometryReader to find out the scale of the trail. The LED digits have a set facet ratio and the drawing coordinates are primarily based on these.

struct LEDDigit: View
{
   var digit: Int? = nil
	
   var physique: some View
   {
      GeometryReader { proxy in
         let (w, h) = proxy.unitSize

         // high horizontal line
         Path { path in
            path.transfer(to: CGPoint(x: 24 * w, y: 7 * h))
            path.addLine(to: CGPoint(x: 60 * w, y: 7 * h))
            path.addLine(to: CGPoint(x: 62 * w, y: 10 * h))
            path.addLine(to: CGPoint(x: 57 * w, y: 15 * h))
            path.addLine(to: CGPoint(x: 24 * w, y: 15 * h))
            path.addLine(to: CGPoint(x: 21 * w, y: 10 * h))
            path.closeSubpath()
         }
         .activeLEDEffect(when: [0, 2, 3, 5, 7, 8, 9].accommodates(digit))
         ...
}

Whereas this produces the right output, it causes the person Paths to animate individually when rotating the system. I solved this drawback by shifting the person path’s code right into a Form the place I’m including the bars solely primarily based on whether or not I’m on the lookout for the lively or inactive LED components. The trail(in rect: CGRect) perform fingers us the required dimension, so we don’t a GeometryReader any extra.

struct LEDDigitShape: Form
{
   var digit: Int? = nil
   var isActive: Bool
	
   func path(in rect: CGRect) -> Path
   {
      let w = rect.dimension.width / 73
      let h = rect.dimension.top / 110
		
      var path = Path()
		
      // high horizontal line
		
      if [0, 2, 3, 5, 7, 8, 9].accommodates(digit) == isActive
      {
         path.transfer(to: CGPoint(x: 24 * w, y: 7 * h))
         path.addLine(to: CGPoint(x: 60 * w, y: 7 * h))
         path.addLine(to: CGPoint(x: 62 * w, y: 10 * h))
         path.addLine(to: CGPoint(x: 57 * w, y: 15 * h))
         path.addLine(to: CGPoint(x: 24 * w, y: 15 * h))
         path.addLine(to: CGPoint(x: 21 * w, y: 10 * h))
         path.closeSubpath()
      }
      ...
}

That is used such:

struct LEDDigit: View
{
   var digit: Int? = nil
	
   var physique: some View
   {
   ZStack
   {
      LEDDigitShape(digit: digit, dot: dot, isActive: false)
         .activeLEDEffect(isActive: false)
      LEDDigitShape(digit: digit, dot: dot, isActive: true)
         .activeLEDEffect(isActive: true)
   }
}

The 2 members of the ZStack draw all of the inactive LED components behind the lively LED components. It nonetheless wanted to be two Shapes as a result of one form can solely have a single drawing model. The inactive components are merely stuffed in a grey. The lively components are full of pink and have a pink glow round them simulating some radiance.

With this method a digit is all the time drawn in its entirety which lends itself to easy resizing.

Format and Orientation Woes

The following step was to combination a number of LED digits and lay them out over the display screen with completely different positions for panorama and portrait orientations, with a easy animation whenever you rotate the system.

I’ve mainly two layouts:

1. Hour digits, Colon, Minute digits (in a HStack)- in horizontal structure with the outer sides touching the secure space insets
2. A VStack of Hour digits and Minute digits – in vertical structure

Sounds straightforward, however my makes an attempt with HStacks and VStacks failed miserably. Firstly of the rotation animation the digits would all the time get a really small body increasing into the ultimate one.

I can solely think about that by some means the SwiftUI structure system doesn’t keep in mind that these are the identical views. So I attempted giving them static identifiers and I additionally tried geometry matching. However I couldn’t shake these animation artefacts. There have to be some piece lacking in my understanding about view id.

In the long run I got here again to doing my very own structure inside a GeometryReader, setting body’s width/top and acceptable offsets (i.e. translation) for particular person components. This works very properly and likewise lets me have a separate animation for the opacity of the colon.

The colon sticks to the precise aspect of the hour digits and disappears in portrait structure. By sorting view modifiers in a sure manner I used to be capable of get this impact that the colon fades in with a slight delay.

var physique: some View
{
   GeometryReader { proxy in
			
   let digitSize = self.digitSize(proxy: proxy)
   let colonSize = self.colonSize(proxy: proxy)
   let centeringOffset = self.centeringOffset(proxy: proxy)
   let isLandscape = proxy.isLandscape
			
   let timerSize = self.timerSize(proxy: proxy)
			
   Group
   {
      LEDNumber(worth: mannequin.countdown.minutes)
      .body(width: digitSize.width * 2, top: digitSize.top)
      .animation(nil)
				
      LEDColon()
      .body(width: colonSize.width, top: colonSize.top)
      .offset(x: digitSize.width * 2, y: 0)
      .animation(nil)
      .opacity(isLandscape ? 1 : 0)
      .animation(isPadOrPhone ? (isLandscape ? .easeInOut.delay(0.2) 
                              : .easeInOut) : nil)
				
      LEDNumber(worth: mannequin.countdown.seconds)
      .body(width: digitSize.width * 2, top: digitSize.top)
      .offset(x: isLandscape ? digitSize.width * 2 + colonSize.width : 0,
              y: isLandscape ? 0 : digitSize.top)
      .animation(nil)
   }
   .offset(x: centeringOffset.width,
           y: centeringOffset.top)

You may see that I’m particularly disabling animation with .animation(nil) for probably the most components as a result of I discovered that the animation in any other case is all the time out of sync with the rotation resizing animation. The LED colon however has its personal animation with an extra delay of 0.2 seconds.

The second purpose why I explicitly disabled animations is as a result of on the Mac model these animations would lag behind the resizing of the app’s window. This resizing additionally switches between each layouts relying on the way you drag the window nook, type of like “responsive design” as we’ve seen on HTML internet pages. Extra on Mac issues additional down beneath.

Multi-Modal Buttons

One other problem that had me attempt a number of approaches involved the preset buttons (high left) and site visitors gentle buttons (middle backside). These buttons have a unique perform for a single faucet (choose) versus an extended press (set).

The principle drawback is that you just can’t have a easy .onLongPressGesture as a result of this prevents the traditional faucets from being dealt with. One method is to have a .simultaneousGesture for the lengthy press, however then the faucet motion is executed proper (i.e. “simultaneous”) after the lengthy press motion if you happen to elevate the finger over the button. The opposite method is to make use of a .highPriorityGesture which once more disables the built-in faucet.

I ended up with the next method which makes use of the gesture masks to selectively disable the lengthy press gesture if there is no such thing as a lengthy press motion and to disable the faucet gesture if an extended press was detected.

struct LEDButton<Content material: View>: View
{
   var motion: ()->()
   var longPressAction: (()->())?
   @ViewBuilder var content material: ()->Content material
	
   @State fileprivate var didLongPress = false
	
   var physique: some View
   {
      Button(motion: {}, label: content material)  // should have empty motion
      .contentShape(Circle())
      .buttonStyle(PlainButtonStyle())   // wanted for Mac
      .simultaneousGesture(LongPressGesture().onEnded({ _ in
         didLongPress = true
         longPressAction!()
         didLongPress = false
      }), together with: longPressAction != nil ? .all : .subviews)
      .highPriorityGesture(TapGesture().onEnded({ _ in
         motion()
      }), together with: didLongPress ? .subviews : .all)
   }
}

This method makes use of a customized TapGesture in tandem with the LongPressGesture. A @State variable retains observe of the lengthy press. We do must reset didLongPress to false or else all subsequent faucets would proceed to be ignored. I discovered that I don’t want a dispatch async for placing it again to false.

I consider that the explanation for that’s that the primary setting of the variable causes the physique to be up to date and thus the together with: to disable the faucet gesture whereas in progress. Thus the faucet doesn’t fireplace upon releasing the lengthy press. Good to know: The .all permits the gesture and the .subviews disables a gesture.

Opposite to different approaches I’ve seen on the web this method preserves the usual habits of Button for highlighting, Whilst you press a customized button like this, it makes it barely clear.

A Mac Model – For Free?

The massive promise of SwiftUI is that you’d get a Mac model of your app for little further work, successfully “free of charge”. So I made a decision to place this to the take a look at additionally produce a macOS model. I set the focused gadgets to iPhone, iPad, Mac and selected the “Optimize Interface for Mac” as a result of that sounded to me like the higher end result.

This optimized mode brought about some points for my customized buttons, as a result of they obtained changed with empty spherical rects destroying my customized look. You may forestall this modification by including .buttonStyle(PlainButtonStyle()).

Other than this my code actually did run as a local Mac app fairly properly. Behind the scenes although it’s all Mac Catalyst. As I perceive it, meaning UIKit continues to be on the helm, on Mac only a macOS model of it.

I left the code signing settings alone as I wished to have customers be capable of set up the Mac and iOS variations with the identical buy. This “common buy” is enabled by having the identical bundle identifier for each variations.

Some very minor tweaks had been required for adjusting some minimal and most button sizes. There’s a bug on macOS that stumped me for some time. Solely on Mac I discovered that after I tapped in sure spots in my app this is able to trigger gestures to cease working. Then after I triggered a brand new structure by resizing the window, every little thing returned again to regular.

My workaround for this was to connect the Pan Gesture (for setting the timer) solely to the LED digits. This manner there is no such thing as a interference and all buttons proceed to work usually. The system would possibly get confused by having too many conflicting gestures on high of one another.

A side-effect of the Mac model is that you just begin to connect keyboard shortcuts to buttons. This was additionally a purpose why I wished to get Button to work with faucet and lengthy press versus making a customized view that isn’t a button.

let title = "(index+1)"

PresetButton()
.keyboardShortcut(KeyEquivalent(title.first!), modifiers: [.command])

This manner you’ll be able to set off the preset buttons additionally with COMMAND plus quantity. And never only for the Mac app, however that works for iPads with hooked up keyboard as effectively.

That obtained me pondering, that perhaps it will be nice to permit the house bar to cease/begin the timer, like we’re used to from video gamers. For that objective I’ve an empty utterly black button behind the LED digits:

Button(motion: { mannequin.isTimerActive.toggle() },
       label: {
          Rectangle()
          .foregroundColor(.black)
          .body(width: timerSize.width, top: timerSize.top)
          .onTapGesture(rely: 2) { mannequin.restoreGreenTime() }
       })
.keyboardShortcut(.house, modifiers: [])
.buttonStyle(PlainButtonStyle())

This button permits me so as to add a keyboard shortcut for house to behave the identical as a faucet. Curiously having a two-tap gesture hooked up to the Rectangle() poses no drawback.

I submitted the Mac construct proper after the one for iOS however initially obtained a stunning rejection:

The consumer interface of your app is just not in step with the macOS Human Interface Pointers. Particularly:

We discovered that the app accommodates iOS contact management directions equivalent to faucet and swipe.

The explanation for that was that I put again the assistance display screen with a textual content I had beforehand written with iOS in thoughts. I wanted to exchange mentions of swiping with dragging and as a substitute of tapping you might be clicking. I’ve laborious coded the textual content and formatting for now and with and #if I can change the textual content between a model for Mac and one for iOS.

Group
{
   Textual content("Setting the Timer")
   .font(.headline)
   .padding(.backside, 5)
						
#if targetEnvironment(macCatalyst)
   Textual content("To regulate the timer, click on on the LED digits and drag horizontally.")
   .font(.physique)
   .padding(.backside, 5)
#else
   Textual content("To regulate the timer swipe left and proper.")
   .font(.physique)
   .padding(.backside, 5)
#endif					
}

As soon as I had made these modifications the Mac app was accredited in a short time.

Conclusion

I’ve skilled first hand how I can rewrite an app in SwiftUI and the nice pleasure that may be had from deleting all of your crufty Goal-C code when doing so.

SwiftUI is my new love and this fashion my app is not a “baby from one other mom”. This restores some enthusiasm in me to really lastly actually add some long-promised “thrilling new options”. For starters I’m pondering of getting a watchOS companion app which reveals the timer and permits you to distant management it. One other concept is likely to be to retailer my presets on iCloud in order that they’re the identical on all my gadgets.

I might love to listen to from you what you consider the method of re-implementing components of apps and even entire apps in SwiftUI.

Additionally revealed on Medium.

Tagged as: SpeakerClock

Classes: Updates