2 ms·

The argument would be that tests that are easily broken by relatively safe changes are bad tests.

• Evbn 14y ago
  Or that the tests are correctly telling you that your proposed change will break your clients.