Behind the App: Wallaroo

It’s been awhile since I’ve done one of these deep dives on what goes on behind the scenes during the development of an Iconfactory app. There’s a common thread to each one: I feel the need to document our work when there’s a major change in how we build user interfaces.

The first one was for the “flattening” of Twitterrific 5, a task that preceded Apple’s work in iOS 7 by six months. The next one was for Flare 2, when the Aqua face of macOS began a dramatic evolution in Yosemite.

With Wallaroo, there’s another major change that may not be noticeable on the surface: it’s the Iconfactory’s first app written completely in SwiftUI.

A Discovery

It all started while I was working on Shortcuts support in Tot. During March of this year, I noticed that there was an action to “Set Wallpaper”. I also learned how Shortcuts could be downloaded, installed, and managed using a URLs.

The Iconfactory has been making wallpaper images since the dawn of time, but it never made sense to make an app because changing your wallpaper was a manual task. Shortcuts radically changed this calculus and the idea for an app was born. I threw together a quick prototype that let you set two wallpapers. Sean and I had the beginnings of Wallaroo.

Our wallpaper prototype became one of those “we’ll do it some day” projects. Then something important happened: WWDC 2022. After Lock and Home Screen customization was announced, the idea immediately became a “we need to do this before September!” project.

Time is Tight

We built Wallaroo from scratch in a little over two months.

The project started with a couple of wrinkles: the “Set Wallpaper” action didn’t work with the new features on iOS 16, so we filed FB10377111 on June 6th (a couple of hours after the keynote ended). We placed our faith in the abilities of the Shortcuts team and decided to carry on in spite of this setback. (We wish everyone at Apple wrote release notes like they do!)

The other wrinkle was that we all had work-in-progress that needed to be finished up. We knew that the short timeframe meant that this was an “all hands on deck” situation, so it wasn’t until the end of June before we all freed up. We put the prototype on TestFlight and got to work.

Divide and Conquer

There were three major areas where we focused our attention:

  • Content – Hundreds of wallpapers had been created over the years, but resolution and aspect ratio varied widely. Things needed to be cleaned up.
  • Backend – Over years, we had done releases in an ad-hoc manner: uploading ZIP files to a Patreon account would no longer be acceptable. We needed a server to manage the wallpapers.
  • Frontend – An app to display the wallpapers: it had to look and work great. Sexy and fast were primary design goals.

Ged, Anthony, Dave, and Talos immediately got to work on the first bullet, but without a backend server, there was no place to put files and metadata. So we made a Numbers spreadsheet and shared it in Dropbox along with the source images. Our Slack channel for the project was filled with “I’m going in” and “I’m out!” to avoid write conflicts. (S.W.A.T. = Software Write Avoidance Technique)

I was responsible for the development of the backend. Importing a spreadsheet CSV file gave us our initial database and images in Dropbox let me manually generate thumbnails and other content that would be needed in the app.

Sean took the lead on the app. We’ve been holding back on SwiftUI due to its immaturity, but the changes in iOS 16 looked great, so we went all in (the only UIKit/AppKit code is in delegate connectors). The data in the spreadsheet was massaged again to give him some real data to use.

Talos took the lead on the app architecture and the wireframe was finished on July 5th. A week later we had enough working code to make a Git repository. A few days later Sean showed us his Captain Pike Appreciation App:

We were on our way, but had less than two months before an iPhone announcement. Time to kick butt.

Butt Kicking

July was a blur. Progress was quick and everyone was heads down on their app responsibilities.

Remember that bug in Shortcuts that prevented Set Wallpaper from working in iOS 16? It was still around and we were starting to get worried. When iOS 16 beta 5 dropped on August 8th, we rejoiced when our test ran. The shortcut action worked perfectly!

Our last update to the shared spreadsheet was on August 10th. From that point on, we were able to use our new content management system to add and update wallpaper. There was still a ton of work needed to clean up metadata and fine tune each wallpaper release.

With the new server up and running, we started testing push notifications. Since Sean’s focus was still in the primary user interface, I started working on the SwiftUI views and models that talked to our backend server. That work continued with integrating the Patreon API and hooking up StoreKit2 for subscriptions. I also had a blast doing the User License and Credits screens.

We started our first beta test with Patreon supporters on August 25th. We were going to make the mid-September release date!

SwiftUI FTW

Looking back on the development, I think there were two things working in our favor: experience and SwiftUI.

We’ve made a lot of apps and have an instinctual knowledge on how to build them. But no matter how little friction there is in working together, you still have to put the pieces together.

SwiftUI is incredibly good at doing that.

Keep in mind that neither Sean or I had created a full-fledged app using SwiftUI (widgets don’t count). We had to learn the idioms and best practices, but once that was overcome, development happened at a lightning pace.

We encountered roadblocks, of course. Tracking down memory leaks was harder than UIKit because of the abstractions. Figuring out how to share an image was a huge head scratcher. Implementing parallax in a ScrollView was many days of hard work. And you should see the comments in our PagingView!

But overall, the experience was extremely positive. If you’ve been on the fence with this technology, iOS 16 feels like a turning point in SwiftUI’s evolution.

Give It a Go!

Now that you’ve read about how we made it, take Wallaroo for a spin. It’s a FREE download and a fun little app for iOS 16. And a great example of what you can do with SwiftUI.

Lame, Until it Isn’t

Where there’s smoke, there’s fire. And as we approach WWDC 2022, there’s a lot of smoke around AR and VR. In some ways, this is going to be a huge inflection point, in other ways, it’s probably going to be a letdown.

Remember when the iPod was announced? Some folks called it lame because it didn’t meet their expectations.

The same thing will be true of anything Apple wants us to put on our face. It’s going to less impressive technically than any of the currently shipping products. And that’s good, because you don’t make fundamental changes by tweaking existing technologies. A Nomad audio player was a tweak. An Oculus headset is a tweak.

Everything we’ve seen to date with VR has been an attempt to bring information to a 3D world. Headsets are just a means to project that 3D environment so our eyes can see it.

I think Apple’s approach with AR will be completely different: they will bring 3D to an information world.

We all have the greatest source of information humankind has ever known in our pocket or purse. Much of that information relates to the world around us: weather, transportation, shopping, dining, etc. Relating that data to our physical space will be a powerful tool.

All the AR examples we’ve seen on Apple’s devices hint at this direction. They take the information on our phone and place it at derived 3D coordinates. Where people get tripped up in these demos is where the results are shown: on a standard screen.

I don’t think that’s Apple’s final goal, because any current screen technology will block your view of the real world. It’s also why I think 3D headsets will remain a niche technology: people have innate need to see what’s going on around them.

Our current  screens also use a lot of power. And that means batteries. And that means weight. Not what I want on my face, for sure.

Apple knows this and that’s why I think a new display system is the thing they’re taking time to get right. We may or may not see this new display at WWDC. I can remember a time when all we had for an iPad was a simulator.

The changes caused by a new display will be incremental. There will certainly be technical limitations in the product that are imposed by size and weight: Apple will improve on those things as components allow.

Other changes will happen because no one, including Apple, really knows how this display will be used by normal folks (we, I should note, are not normal folks). The first Apple Watch tried to do a lot of things: iteration got rid of things no one used, and improved the things everyone wanted.

It’s likely that a first iteration will also be a “satellite device” where the iPhone does the heavy lifting. Much like the original iPod relied on a Mac. The realityOS could be nothing more than widgets for a new display on your face.

That will feel lame until you realize something else: after two decades, the basic form factor and functionality of that first iPod is now an essential part of our lives and we call it an iPhone. Don’t underestimate Apple’s ability to iterate.

Consistency Sin

This past summer we narrowly avoided a major user interface regression on Apple devices. The story ended well, but I think it’s important to look back on the situation and ask a simple question:

Why did this happen in the first place?

My answer is something I call “consistency sin”. Understanding the cause lets us avoid similar situations in the future.

Your first reaction to this nomenclature may be, “Isn’t consistency a good thing in user interfaces?”

Absolutely! Colors, fonts, and other assets should be similar within an app. Combined they help give the user a sense of place and act as a guide through an interface. And in many, cases these similarities should be maintained across platforms. There’s no sin there.

But you can get into trouble when this consistency starts to affect the user experience.

Design is not how it looks, it’s how it works.

Steve Jobs said a lot of smart things, but I use this advice most often.

The roots of consistency sin take hold when folks disregard the inherent differences between platforms. A greater importance is placed on making sure things match visually: how a person uses that design takes a back seat.

Platform controls and interface elements can differ at a fundamental level. The mouse is optimized for indirect interaction while a screen is optimized for direct interaction.

For the most part, developers are shielded from these details through the use of standard components that conform to the Human Interface Guidelines.

Enter Safari

Higher level interactions are driven by the type and quantity of information the user is working with.

To use Safari as an example, I can have hundreds of tabs open while I work on my Mac; on iOS it’s usually less than a dozen. Safari on iOS is also a full screen experience, while multiple windows and interactions with other applications are common on macOS.

Safari’s new tab design on iOS works great for me: swiping between tabs of fullscreen content is a better interaction for a limited number of pages. The grid of pages as a fallback for selection also works well for managing what I want to keep around.

The consistency sin in Safari was to come up with a good design for iOS and assume that it would also work well on iPadOS and macOS. It practice, these new tabs were difficult to use in a different work environment.

Luckily the folks working on Safari did the smartest thing possible: they listened to feedback and fixed the issues before shipping. That’s an important thing for a product that every Apple customer uses every day on every device.

It’s one thing to make a mistake, it’s a wholly different thing to deny that anything’s wrong. So let’s take a look at another example.

Notifications, Too

Notifications also suffer from consistency sin.

I look at my iPhone Lock Screen dozens of times each day, and sometimes just to just view a reminder or some other short notification. It’s quick, simple, and minimizes distractions.

On my Mac I see the Lock Screen only once or twice per day for just a few seconds as I enter a password. That means notifications occur while I’m actively working.

Again, consistency sin looks for a single solution that ignores my needs. On macOS I don’t want a minimal solution that is suitable for a mobile device. I want options that let me quickly dismiss or defer an item that’s interrupting my work. (And I certainly don’t want to hunt around in the window for a hidden control that lets me access a function.)

Placement is also an issue: on iOS controls tend toward the bottom of the screen (for reachability). The opposite is true on macOS where they tend  toward the top of the screen so they’re closer to the menu bar and window controls. Consistency sin says that notifications should always be at the top of the screen.

With iOS, there’s a nice visual and functional separation between app interactions in the lower half of the screen and notifications in the upper half. On the Mac, notifications are just another thing fighting for real estate at the top of the screen.

We’ve Only Just Begun

The bad news is that we’re likely to have more consistency sin in our future, thanks to Electron and other cross platform frameworks.

While development teams try to attain feature parity, experience parity will suffer. Everyone who’s used an iOS app and immediately thought “this is a web page” will know what that means.

There is a long history of user interface frameworks that make work easier for a product team. Every time, these solutions end up being a least common denominator that makes it more difficult for customers. Don’t be surprised when they complain: as they did for Safari, and as they do for Notifications.

If you’re a designer or developer, it’s your job to push back on the notion of consistency when it begins to affect a user’s experience. Remember design is how it works, and work is not the same on every device. 

jsc: My New Best Friend

A friend of mine recently pointed me at a well hidden command line tool. In the JavaScript framework used by Safari and other parts of Apple’s products, there is a tool called jsc. It’s a command line interface for JavaScript that uses the same code as the rest of the system.

You can find the binary at /System/Library/Frameworks/JavaScriptCore.framework/Versions/Current/Helpers/jsc. That path is unwieldy, so I have an alias set up that lets my just type jsc in the Terminal.

So what can you do with jsc? Pretty much anything you can do with JavaScript in a browser with the caveat that there aren’t document and window instances.

If you run jsc -h, you’ll see a lot of options for testing and profiling JavaScript. It’s clear that the WebKit team uses this internally for running tests. But we can also use it for trying out ideas and running simple utilities.

A picture is worth a thousand words, so let me show you how it can be used to solve a simple problem. I recently needed to convert some strings in our Turkish localization of Frenzic to uppercase: the lowercase “i” was getting converted to the dotless version.

JavaScript’s toLocaleUpperCase() function is the perfect way to do this, so I pulled jsc out of my tool bag and got to work. The first challenge was getting input.

Luckily, there is a readline() function that takes keyboard input and returns a value. Unluckily, that input isn’t in the encoding you’d expect it to be: characters are returned in ISO-8859-1 (Latin-1), not UTF-8. Remember, there’s no document instance so the default encoding is used.

To workaround this limitation, you can percent escape the characters to UTF-16 and then decode them back into UTF-8 with this technique:

var text = decodeURIComponent(escape(readline()));

(If any WebKit engineers are reading this, it would be nice to have a command line option like --encoding=utf-8.)

Generating output is a bit different than a browser, too. You’ll be using print() instead of console.log(). To convert the text input and display it, I used this:

print(text.toLocaleUpperCase('tr-TR'));

There are a few more built-in functions that may prove useful, but so far, I’ve only needed to read and write text. It’s undocumented, but jsc also takes standard input and can be used as a shebang:

$ echo "print(1+2);" | jsc
3

Since this is likely code I’ll have to use again, I created a Turkish.js file:

while (true) {
    print('Turkish text?');
    var text = decodeURIComponent(escape(readline()));
    print(text.toLocaleUpperCase('tr-TR'));
    print('-------------');
}

I can now run this any time with jsc Turkish.js. And you also get to see how having JavaScript in a command line can be handy. Enjoy!

Codesign: The Saga Continues

I have a long history of writing about code signing in macOS. When Big Sur was released, I thought “Finally!”

I was wrong.

This time around I was tripped up by Safari, of all things. It doesn’t open app archives like other parts of macOS.

This story began with customer reports of xScope being a “damaged app” on Big Sur. This was surprising because I had been downloading and testing the app on Big Sur for several months without issue.

I was also doing all this work on Apple Silicon using the DTK. And since Google Chrome wasn’t yet working on this device, all my testing was limited to Safari. Safari’s default setting is to open “safe” files after download, so I left that alone (as most customers would).

This is where I shot myself in the foot. At no point did my downloads touch the Archive Utility. And I had no idea that Safari’s code is different than the system utility.

When I checked the signature of the app downloaded with Safari, everything looks good:

$ codesign -vvvv ~/Downloads/xScope.app
...
/Users/CHOCK/Downloads/xScope.app: valid on disk
/Users/CHOCK/Downloads/xScope.app: satisfies its Designated Requirement

Things were very different when using Google Chrome:

$ codesign -vvvv ~/Downloads/xScope.app
/Users/CHOCK/Downloads/xScope.app: unsealed contents present in the root directory of an embedded framework
In subcomponent: /Users/CHOCK/Downloads/xScope.app/Contents/Frameworks/Sparkle.framework

$ codesign -vvvv ~/Downloads/xScope.app/Contents/Frameworks/Sparkle.framework
/Users/CHOCK/Downloads/xScope.app/Contents/Frameworks/Sparkle.framework: a sealed resource is missing or invalid
file added: /Users/CHOCK/Downloads/xScope.app/Contents/Frameworks/Sparkle.framework/Versions/Current/Resources/._fr_CA.lproj
file added: /Users/CHOCK/Downloads/xScope.app/Contents/Frameworks/Sparkle.framework/Versions/Current/Resources/Updater.app/Contents/Resources/._fr_CA.lproj
file added: /Users/CHOCK/Downloads/xScope.app/Contents/Frameworks/Sparkle.framework/Versions/Current/Resources/Updater.app/Contents/Resources/._pt.lproj
file added: /Users/CHOCK/Downloads/xScope.app/Contents/Frameworks/Sparkle.framework/Versions/Current/Resources/._pt.lproj

It turns out all these folks complaining about a “damaged app” were either using Chrome or had Safari’s “safe” file handling turned off. The damaged archive wasn’t getting repaired automatically by Safari.

The root of the problem is localization in the Sparkle framework. There are two symlinks with extended attributes (the “._” is where macOS stores things like Finder information). The intent of the symlink was to say that French Canadian is the same as French, and Portuguese is the same as Brazilian Portuguese.

Since macOS automatically makes this inference, it’s safe to just delete the scripts that create the symlinks. In your Sparkle project, find any occurrences of “Run Script: Link fr_CA to fr” and “Run Script: Link pt to pt_BR” in your Target Build Phases and remove them. I had them in “Sparkle”, “SparkleCore”, and “Installer Progress”.

After you build and notarize, you’ll see that your app is “valid on disk” no matter how it’s unarchived.

I’ve also submitted this information to Apple’s Product Security group. As I said in my email, the biggest problem here is expectations:

The reason I’m writing is because Safari’s implementation for opening “safe” files is somehow bypassing a code signing check or repairing the downloaded package. The Archive Utility does not. Customer and developer expectations for unzipping archives is that they are not modified and behave the same way across all Apple products.

If you’re a Mac developer who’s using Sparkle and distributing your product via a web download, now’s a good time to check how things work in a variety of browsers. I’ve heard that we’re not the only ones affected.

And if you encounter a download that’s damaged because of these Sparkle symlinks, this quick fix in the Terminal will set things right:

$ ditto xScope.app xScopeFixed.app
$ rm -rf xScope.app
$ mv xScopeFixed.app xScope.app

The ditto command strips the extended attributes that are causing the issue. This may, in fact, be what Safari is doing for “safe” files.

All that’s left to do now is wonder what surprises codesign has in store for next year’s release of macOS…