write tests, not scripts that run them

 Writing table-driven tests is usually a good idea. Adding a test case
 doesn't require adding code, so it's easy to avoid fucking up the
 other tests. However, actually going from a table of tests to a test
 that runs is non-trivial.
 .
 Test::TableDriven makes writing the test drivers trivial. You simply
 define your test cases and write a function that turns the input data
 into output data to compare against. C<Test::TableDriven> will compute
 how many tests need to be run, and then run the tests.
 .
 Concentrate on your data and what you're testing, not
  plan tests => scalar keys %test_cases
 and a big foreach loop.