Cross Platform Mobile

29 April 2011

With the rise of so many mobile platforms, each with a different UI, many people are looking at cross-platform toolkits. These allow you to write a mobile app once and then deploy it to a range of mobile devices. Are these toolkits worth using?

A cross-platform toolkit is alluring. There are a bunch of mobile devices out there, and it's a lot of work to write applications for every one. It would be so much easier if we could write once and run anywhere.

Of course we've been here before, with desktop applications. Over the years there were lots of attempts to create cross-platform environments for desktops. Java is perhaps the best known, but it was neither the best nor the first (ask Smalltalkers to tell you about VisualWorks). But you may notice that cross-platform apps have, on the whole, not been successful. [1][2]

The first issue is handling the variations between UI controls on different platforms. You have two broad strategies: you can either use native components on each platform or emulate components using more primitive graphics - effectively creating your own UI system. In the Java world this was the difference between Swing (emulation) and SWT (native controls).

Both approaches have problems. If you use native controls you have to deal with the fact that similar controls on different platforms have slight differences in how they work, which makes it hard to come up with a good abstraction. Furthermore you have to decide about what to do with capabilities that are only available on some platforms - do you end up with the lowest common denominator of your platforms?

So emulation becomes attractive, either as an entire approach or to make up for controls you can't get on some platforms. Emulation has a couple of difficulties: firstly it's hard to get an emulated UI to perform responsively enough - which is a big deal for UI controls. Secondly it's very difficult to get them to perform exactly like the native controls. It's easy to get trapped in an uncanny valley where things work mostly like the native controls but there are just enough tiny differences to throw users off. With UI controls you have to be really anal to get the behavior "just right".

Getting good UI controls are almost impossible, but that's not the biggest issue - which is more about the overall user experience. Different platforms have different ways they expect you to use them that alter the entire experience design. If you use a Unix-based system, a good experience design will make extensive use of the middle mouse button, since that's the way of Unix UIs. But try that on a Mac and you'll struggle, since the Mac thinks even two buttons are too many.

Mobile devices have an even bigger burden here as they have a greater disparity in how they approach the overall user interaction. Recently a client asked us to take an iPhone app we'd developed for them and make a version for Android. Our initial experiments showed us that we had to completely rethink the experience design from scratch to get a decent Android app. The test cases for the iPhone app made an extremely valuable check-list for the capabilities of the Android app, but the whole feel of the app had to be different.

So for these reasons I think cross-platform mobile toolkits are a dead-end. It's just too hard for them to really mimic the native experience. If it's worth building a native app, it's worth building it properly, including an individual experience design for that platform. [3]

So where does this leave people who want to support lots of mobile platforms, but aren't prepared to deal with the cost of native apps for each one? Fortunately we do have a single platform that runs on any worthwhile mobile device - the web. Web apps can be very functional and capable these days when written by people who know what they are doing. The biggest issue here is offline use. If you can live with online all the time, then this won't be a problem, but if you need offline you'll need to explore the various local storage options.

When you do the web app, don't try to make it look and feel like a native app - make it look like a mobile web app. It'll still be as usable as an emulated app, but will avoid plunging your users into the uncanny valley. This mirrors what has happened on the desktop, where people who want to support multiple desktop platforms have found the web to be an effective deployment platform. The key to its success here is that people expect web apps to behave like web apps, so they don't expect them to be like native apps - avoiding the problems of different user expectations on different platforms.

To summarize:

  • Don't use cross-platform toolkits
  • For maximum reach: built a web app that looks like web app
  • To appeal to a particular platform: build a native app for that platform, with a experience design based on that platforms interaction style

Follow-Ups

Some thoughts and reactions from emails and other comments

One correspondent talked about how his company creates native apps that are thin wrappers around web apps. This provides ease of launching and getting to common links, while keeping most of the behavior within the cross-platform web.

Gunnar Peterson argues that differences in security models also discourage cross-platform toolkits.

Notes

1: These issues are issues about cross platform apps with a graphical UI. Libraries without a UI work very well in a cross-platform manner. As a result often the best strategy on a desktop is to combine cross-platform libraries with a custom UI for each platform.

2: A notable exception is Jetbrains's family of development tools. I don't have an explanation of why they are such an exception.

3: I see one path that might prove me wrong. In this scenario you use a cross-platform toolkit - but you write a different app, with a different experience design, for each platform you build for. The gain over doing this with native code is that you have a single platform for your developers to use and can get some reuse of common code (particularly non-UI code). This strategy doesn't address the problem of dealing with UI controls, and even if it works, it's only worthwhile if the developer-understanding and code reuse benefits are significant.