Question

I am struggling with this question since I noticed that many functional testing frameworks (like Selenium for the web or UISpec for iOS) actually simulate UI events while testing. I am asking: couldn’t it be sufficient just to check for preconditions such as that, e.g., the target and selector for a button are set correctly and then fire the selector manually? Why do I need to simulate touches? This has the con that you have to know more about the UI elements you’re testing (you have to know what makes them to behave correctly), but since I am the one writing the tests, maybe this doesn’t matter?
Could anyone shed some light on this?

Answer 1

Simulating touches can be useful for determining crashes caused by obscure or unplanned user behaviour – a particularly common one is having two items pressed simultaneously. It also allows you to create potentially quite esoteric tests: for example, random user input for a sustained period of time to attempt to crash or break your application in ways you wouldn’t expect. The level to which you’d do this would depend on your app, and how important it was to you.

Your alternative approach also has some disadvantages when it comes to multi-touch. Whilst it would be fairly straightforward to fire a button selector through some sort of automatic test rather than simulating user input, what happens if you have an app that deals with swiping, pinching, or other multiple input gestures? In those cases the desired result may not be as black and white as the on/off of the button: you may have many shades of grey and differing output that required validation.

Simulated UI testing actually has quite a long history – there’s an interesting story (well, interesting to me) about the original MacPaint and how a random UI input test was able to assist in reproducing obscure or difficult crashes here: http://www.folklore.org/StoryView.py?story=Monkey_Lives.txt