Reproduce, then fix
A bug is first reproduced with a failing check. Then it is fixed, and the same check is run again and passes.
At Blume, we do not fix a bug we have not seen fail. First a check that fails for the reason the user reported. Then the fix. Then the same check, passing. The pair of runs, red then green, is the proof.
What goes wrong without it
Fix first
A user reports: “Invoices show the wrong date.” The agent finds a date formatted without a time zone, changes it, sees the test suite pass and reports the bug fixed.
But the suite passed before the change too. The user’s problem was that an invoice created at 23:30 in Oslo shows tomorrow’s date, and that lives in a different function.
Reproduce first
The agent asks which time zone the user was in. Oslo. It writes a check: an invoice created at 23:30 Oslo time shows that day’s date.
The check fails and shows tomorrow’s date. After the fix, the same check passes, and at review you see both runs.
What to do
- Get what you need to reproduce: the steps and the conditions that matter, such as time zone, language and screen size. Ask, do not guess.
- Write the check as the expected behaviour, not “the bug is gone”.
- Run it and watch it fail for the reported reason. Keep the failing run.
- Fix.
- Run the same check again, then the rest of the checks.
If you cannot reproduce it, say so plainly and go back to the reporter with what you tried.