Perl's testing system maintains a test counter. You either put the number of tests you're going to run at the top of the program, or you declare that you're not going to track it. This guard against your test exiting prematurely, but there are other guards so it's not all that important.
Every once in a while somebody tries to write a program to count the number of tests for you. This is, of course, defeated by a simple loop. They plow ahead anyway, doing more and more elaborate tricks to try and detect loops and guess how many iterations there will be and solve the halting problem. Usually they declare that it just has to be "good enough".
Here's a particularly elaborate example.
