As you might know, Firebase Dynamic Links got deprecated as stated in an FAQ entry:

Note: Firebase Dynamic Links is no longer recommended for new projects. In the future, the Dynamic Links service will shut down, but you will have at least 12 months from the announcement date to migrate. We will announce more information in Q3 2023.

That’s why we’re going to build a replacement for Firebase Dynamic Links (FDL) with Flutter in this article.

But first we’ll need to understand what FDL offers and how it works, otherwise we’re not able to build a replacement. Furthermore, we want to understand why it got deprecated, though, since there are no official reasons mentioned, the reasons mentioned by me are just educated guesses. With that being said, let’s jump straight into the teardown of FDL.

FDLs value proposition

[…] are links that work the way you want, on multiple platforms, and whether or not your app is already installed.

At a first glance, this sounds a lot like Android’s App Links or iOS’ Universal Links. Both of those technologies enable deeplinks into your app, just like FDL. Contrary to FDL, those are two mechanisms are baked right into their respective operating systems and SDKs. Interestingly enough, FDL actually predates App Links and Universal Links. FDL is about 7 years old, whereas App Links and Universal Links are younger technologies.

The biggest advantage of FDL over App Links and Universal Links is, however, that FDL has a mechanism that covers deeplinks for the first app start. It’s called deferred deeplinking. This is very nicely explained by this image

Taken from https://blog.pragmaticengineer.com/google-shutting-down-firebase-dynamic-links/

On iOS, this first app start deeplink works by copying the deeplink information to the users’ clipboard on the webpage. After installing and starting the app, the app reads the users’ clipboard, parses it and then allows the developer to use that information to open a specific page in the app.

On Android, I assume that Google Play Install Referrer are used for deeplinks for the first app start. Though, the same clipboard solution can be used where Google Play Services cannot be used.

Let’s take a closer look at the clipboard solution.

We’ll start with the part that’s happening in the browser. We can make use of the navigator.clipboard APIs in order to write something to the clipboard. Nice! Well, kinda. The documentation for it states that we have to request the users’ permission first, otherwise we’re not able to copy something to the clipboard.

In the app, reading from the clipboard causes a popup on iOS and Android, notifying the user about it. This makes the app looks like it’s maliciously reading or even spying on its users when it’s not properly explained.

Other similar services, like branch.io, build their deferred deeplinking solution by saving the destination on the backend for the IP of a specific. This has the drawback, that you can only do it, if the user allows it. Otherwise, you’re violating GDPR.

Additionally, FDL provides functionality for shortlinks and analytics.

Guessing reasons for FDL being shut down

Now that we’ve understood what FDL offers and how it works, we’re able to do some educated guess as to why it’s getting shut down.

  1. The deeplinking functionality is now baked right in into the mobile operating systems. There’s no need anymore for a third party service.
  2. The deferred deeplinks for the first app aren’t as seamless anymore as they’re now plagued (for a good reason) by permission requests for clipboard writing, and hints for clipboard access. This turned the whole first app start experience from magical experience to a somewhat nightmarish experience, as the app looks like it’s spying on you right on the first app start.
    Solutions which work by saving a user’s IP need the consent of the user. An IP is typically considered personal data. So this approach is also not really seamless.
  3. It’s easy enough to build the open app store functionality yourself, if you’re already using App Links and Universal links, so the shortlinks kinda loose some of its value.

Building a replacement for FDL

We need to decided how important which parts of FDL are for our app.

  • Do we value the deferred deeplinking functionality?
  • Do we value the regular deeplinks?
  • Do we value short links?

The important part can be any combination of the above questions. However, I’m just explaining point 1 and 2, as there are already a lot of tuturials out there explaining shortlinks, and this is a mobile focused article.

Of course, you can also just use any available third party service like branch.io, Adjust or whatever. But where would be the fun it?

In order to be forward compatible and not rely on a third party service which could be shut down, we’re using App Links and Universal Links. An additional benefit from that is, that Flutter already has built in support for those kinds of deeplinks. This already explained well better than I could on the Flutter website, so I recommend checking their documentation:

Now, moving on to the fascinating part of FDL, deferred deeplinks. When I investigated into this topic, I was really fascinated by the approach, it’s genius.

On the website, we assume that App Links and Universal Links are in use. That means anytime the user is actually landing on the webpage, it means he’s either not using a mobile phone, is on a mobile but hasn’t installed the app, or is on a mobile phone and disabled deeplinking. To discern between a mobile phone and desktop browser, we can check the UserAgent of the browser. That also helps us in to know whether we should link to the App Store or to Google Play.

const getMobileOS = () => {
  const ua = navigator.userAgent
  if (/android/i.test(ua)) {
    return "Android"
  }
  else if (/iPad|iPhone|iPod/.test(ua)){
    return "iOS"
  }
  return "Other"
}

// ...

const os = getMobileOS()
if (os === 'Android') {
  window.location.href= "market://details?id=<packagename>";
} else if (os === 'iOS') {
  window.location.href= "https://apps.apple.com/de/app/<app-id>"
}

Okay, now we also need to copy the current URL to the clipboard, in order to be able to read it in the app.

const getMobileOS = () => {
  const ua = navigator.userAgent
  if (/android/i.test(ua)) {
    return "Android"
  }
  else if (/iPad|iPhone|iPod/.test(ua)){
    return "iOS"
  }
  return "Other"
}

// ...

const os = getMobileOS()
// The following line got added. You may or may not neet to request a permission to do this.
navigator.clipboard.writeText(window.location.href);
if (os === 'Android') {
  window.location.href= "market://details?id=<packagename>";
} else if (os === 'iOS') {
  window.location.href= "https://apps.apple.com/de/app/<app-id>"
}

Moving on to the Flutter part:

Future<void> main() async {
  WidgetsFlutterBinding.ensureInitialized();
  final deferredDeeplink = await _getDeferredDeeplink();
  runApp(MaterialApp(
    initialRoute: deferredDeeplink,
  ));
}

Future<String?> _getDeferredDeeplink() async {
  if (await Clipboard.hasStrings() == false) {
    return null;
  }
  final content = await Clipboard.getData('text/plain');
  if (content == null) {
    return null;
  }
  // now you need to check whether the content matches
  // the format of your associated website.
  if (content.data.startsWith('https://example.org') {
    // The initial route property of MaterialApp expects the deeplink 
    // to start with a '/'
    return content.data.replaceFirst('https://example.org', '');
  }
  return null;
}

Additionally, the deferred deeplink check should only happen on the first app start, not on any of the consecutive app starts. It’s not shown here, but it can be done based on a persistent data store of your choice.

I want to note, that this is a very naive approach as is, as I want to show how it works. In a real world application, you should put emphasis on explaining the users why you’re using the clipboard. Otherwise, you might look like you’re spying on your users.

This can be done by either explaining this to your users, or by explicitly asking your users if you’re allowed to do so. This however makes the whole experience not as seamless anymore. Which, as I explained above, is also why I believe FDL got deprecated.

Additional resources, which we’re quite helpful during my research for this article, were the following:

Originally published on Medium.