I have long had a prejudice against tools that help you to write apps for mobile phones, especially the ones that claim you can write an app once and deploy it to multiple different platforms. My argument was that iPhone, Blackberry, Android, Windows Phone, etc, all have different user bases who expect different things. How can you write one app that works well across all platforms?
I recognised this as a prejudice, and realised that i hadn’t actually tried out the things that i was criticising. After watching a talk by coder, hacker and lecturer Syd Lawrence about PhoneGap and Sencha Touch, i realised that the time had come to test my assumptions.
Syd Lawrence – Mr Mobbles Magical Emporium from Event Handler on Vimeo.
An opportunity came along to develop an Android version of an app that already exists for iPhone. It’s quite a simple app, and on Monday last week i made a start.
I downloaded PhoneGap and installed the library into a new Android application. It provides you a very fast way to get up and running with an HTML5 app that runs locally on the phone. There are numerous javascript functions which translate into the phone’s native language to access things like the camera, GPS, accelerometer. You can also write native code and expose it to your javascript, for example i wrote a little bit of Java code to hide the soft keyboard, which i could call from javascript at the appropriate time.
To style the app, i used Sencha Touch which is an extension of Ext.js. It’s a very nice way of declaring components in javascript that look visually somewhat like a native app. I got the basic functionality together quite quickly: tabs, network connectivity with JSONP, a scrolling gallery of pictures.
It was nice to use tools i am familiar with: jQuery and CSS3. You can run your app in a browser as you are building it, and periodically just check it looks okay on a phone.
Progress was fast. I encountered a few obstacles, most notably cross-site-scripting limitations, and limitations of the pre-release Sencha Touch version 2 which meant i had to go back to the slower and older 1.1. But i managed to overcome every obstacle and within 4 days i had a basic functioning app.
All was great apart from one thing: performance. On my Android tablet the app took 3 seconds to load. On my entry-level phone it took 7 seconds initialising PhoneGap, Sencha Touch and jQuery. Just to load a page that does basically nothing until you start typing in it. I was also disappointed by the refresh when you rotate the phone’s orientation and it jerkily redraws the screen. It’s great as a proof-of-concept prototype, but it’s not something i’d be happy to release and put my name to it.
So on Friday i thought to myself, you know what? I’m going to see what it would take to rewrite this as a native Android app in Java. I spent a good deal of the weekend working on it and by Monday i had got it to approximately the same point as the PhoneGap/Sencha Touch version. I gave my client the option which to continue with, and together we chose to go ahead with the native app. We lost the ability to deploy to multiple platforms, which was something we were hoping for, but in return, we get a far more responsive, native look and feel high quality app that we can both be proud of.
There is a place for Sencha Touch: if i’m writing a mobile friendly version of a website i’ll definitely give it another try. But for an app that has been downloaded and installed, i feel that the users are justified in expecting something better. And to achieve the quality expected, at least for Android, you have to speak to the phone it its native language.
