Because many floating point numbers can’t be represented exactly in a computer, most testing frameworks provide a way of checking that a floating point result is “close enough” to the expected value. In a recent Python project, I learned that pytest has a solution to this problem that I’d never seen before.
