Question

I am trying to figure out how to properly write unit tests for number conversion to have 100% test coverage and to cover all possibilities.

Let’s say, I have a function which converts integer to roman numeral. It works from 1 to 3999:

function integerToRoman(integer) { ... }

How can I make sure it actually works? Writing 4000 unit tests for every number between 1 and 3999 doesn’t seem like a good option. On the other hand, if I just pick some random numbers and write tests for them, then I cannot be 100% sure the function is working.

Answer 1

100% coverage doesn’t mean testing every single possible input value. But it means that you:

1. Cover with test every function line
2. Cover with test normal (generic) input values (not all possible, but from the different ranges if they are processed differently)
3. Cover all edge cases you can think of
4. Cover various invalid input values (including no input values as well)

That’s it.