In this multipart blog post I want to introduce you in the world of property-based testing and how to do this in C#.
In the first part(this one), I’ll give you an introduction on what property-based testing is, why it useful and how it can help you write better code. In a second post, I’ll show you a concrete example on how to start writing Property based tests in C#.
Disclaimer: Property-based testing is hard! And although I would strongly recommend it for critical parts of your codebase, I would certainly not use it everywhere.
What is Property-based testing?
Property-based testing was introduced in 2000 by Koen Claessen and John Hughes via the Haskell library QuickCheck.
Mark Seemann uses the following definition in his Pluralsight course on Property based testing:
Property-based Testing is an automated testing technique where you incrementally zero in on the correct behavior of a system by describing its properties or qualities in general terms and then use randomly generated test data to execute deterministic tests.
I have read the definition above multiple times but I still find it hard to understand. So let us try to explain this further…
Most of the test I see developers write are Example-based tests. We try to think about what would be representative input for the functions we want to test and use these examples to create some unit tests. The better developers don't only think about the happy path and try to also think about what could be incorrect input to see if our functions handle failure conditions gracefully.
The problem is that when your functions get at a certain level of complexity, it becomes quite hard to think at all possible variations(examples). And although your test coverage could be 100%, you are still not 100% confident that your tests are covering all edge cases.
Property-based testing can help you deal with the shortcomings of example-based tests. Instead of finding relevant examples, we try to understand and represent the relationship between our inputs and outputs. We search for the rules, specifications, “properties” that allow us to describe this relationship in a general way without specifying exactly what the input or output should be.
Sounds still abstract? Let’s give an example:
To test the ‘add’ method above, we can try to come up with representative examples but there is an infinite list of possible combinations. Instead let’s try to describe the add method as a relationship between inputs and outputs:
- Relationship 1: It doesn’t matter in which order we provide the x and y input, the output is the same.
- E.g. 1+7 is the same as 7+1
- Relationship 2: Having zero as either the x or y input is the same as doing nothing (the output doesn’t change).
- E.g. 7+0 = 7 or 0+7 = 0
These are the obvious ones to come up with, a third property that is also relevant is the following:
- Relationship 3: If I call the add method multiple times, the order in which I do them doesn’t matter.
- E.g. 7+(4+5) = (7+4)+5
It is not a coincidence that this perfectly aligns with the mathematical description of an additive operation:
- It doesn’t matter in which order we provide the x and y parameter, the result is the same. –> Commutativity
- Having zero as one of the parameters does nothing. –> Identity
- If I call the add method multiple times, the order in which I do them doesn’t matter –> Associativity
What’s nice about these properties is that they work with all inputs, not just special magic numbers. These properties are in fact the specification of the additive operation.
So far for the theory. In my next post I’ll show you how to write Property-based tests in C#.
If you want to learn more about Property-based testing, here are some interesting resources for you:
- The lazy programmer's guide to writing thousands of tests - Scott Wlaschin
- Midwest.io 2014 - Property-Based Testing for Better Code - Jessica Kerr
- Introduction to Property-based Testing with F# by Mark Seemann on Pluralsight