Mutating Test Scenarios

Most testing (both automatic and manual) aims to increase predictability. This is usually done by:

• Following a predefined set of steps
  
• Removing ambiguity by removing static
  

In this paper we make the point that these assumptions originate from manual testing. If we blindly follow these assumptions in automated testing, we will not be providing any additional advantages apart from a larger capacity of work. It is time that test automation helps us perform tasks that cannot be done by manual testing.

Scenarios that follow a predefined set of steps are too simplistic, and they should be leveraged to go beyond their original intent. We also show that if we can inject selected 'noise' in the execution of a scenario, we open new opportunities in test domains.

We call this approach 'Mutational' test scenarios. In these methods, the scenario we defined will, when needed, step out of the original intent and process, to 'mutate' to address the challenges that are put on it.

In my talk I will describe different levels of mutations, including data, events and permutations (combinations mutations).

Learning outcomes include:

• Writing better tests
• Controlling Chaos
  
• Taking new unexpected paths
• Upgrade Testing
  
• Event-Based Testing
  

Baubak Gandomi

Baubak studied at the Mid-Sweden University, Universität Leipzig (Germany) and finally graduated from the university of Stockholm, Sweden with a degree in Computer Sciences. He is currently a test automation architect at Adobe, a position he has held for the last 9 years. Apart from upgrade tests he is very interested in continuous delivery methodologies, cross-product/teams testing and finding measurement methods for assessing functional and transactional coverage. When not solving Quality and Test related problems, he practices the martial art of Ninjutsu, and promotes the ugliness of the world through his blog, Ugly City Guide.