Testing and Benchmarks

Part of the Go course

You wrote a function, ran the program, saw the right answer, and moved on. A week later you change something nearby and the function quietly starts returning the wrong thing for negative numbers. Nobody notices for a month. This is the ordinary way software breaks, and it is not caused by bad programmers, it is caused by nobody re-checking the old things after changing a new one.

A test is that re-check written down once and performed by the machine. Its value is not proving your code works today, when you have just watched it work. Its value is that in six months you can change anything you like, run one command, and get a straight answer about what you broke.