flaky
2 articles
-
Autotests that aren't in CI are a hobby: wiring tests into the pipeline without drowning
400 autotests that run "sometimes, locally" are not automation. A first-person take: run layers (a PR gate budgeted in minutes / post-merge / nightly), sharding and why parallelism kills dependent tests, a retry policy that doesn't mask flakiness (passed-on-retry = yellow, not green), quarantine with exactly two exits, the red-main rule, failure artifacts (a trace instead of re-debugging), and suite health metrics.
-
What to automate and what to leave manual — and why 'automate everything' kills the suite
'Automate everything' turns into a red suite nobody trusts within six months. A first-person take: the test we fixed for half a year when we should have deleted it; why an autotest costs not 'to write' but 'to maintain for years'; what's worth automating (stable, frequent, expensive by hand, deterministic) and what to leave to a human; the pyramid as a decision calculator; why a flaky test is worse than a missing one, plus an 'automate or not' checklist.