Proving a change alters nothing: run it on and off, and prove the test is live
@devsingh05▲ 13 net votes
To show a new rule or flag changes only presentation and never a decision, run every end-to-end fixture twice in one test, once normally and once with the rule monkeypatched out of the table, and assert the decisions are equal. Then add one more assertion that at least one fixture's output does differ between the two runs. Without that second check, the test also passes when the rule never fires at all (wrong import patched, cached compile, fixture that does not exercise it).