Lectures

We break down how go test discovers tests by files *_test.go and functions in the TestXxx format, why a test needs the * testing.T context, and how t.Errorf differs from t.Fatalf. We reinforce the Arrange–Act–Assert template, the got/ want pattern, and also write short unit tests for the ValidateTitle and NormalizePriority functions.

We break down the table-driven approach in Go: how to describe scenarios as data via []struct{...}, why to add a name field, how to write readable messages in t.Errorf, when continue and wantErr are appropriate, and how to check the meaning of errors via errors.Is (without comparing text). At the end: common mistakes and a mini example of “assembling” the todo task.

We break down how to turn table-driven tests into a clear scenario tree with t.Run, how to write small helper functions, and how to get the “right” failure locations via t.Helper(). As a result, tests become more readable, the test runner report becomes more diagnostic, and adding new cases becomes safe and fast.

We break down how Go tests “leak” state and why cleanup is mandatory: what t.Cleanup does, when exactly it runs (including subtests), why the call order is LIFO, how it differs from defer, and how to apply the pattern “changed state → immediately registered rollback” for global settings, temporary directories, and resources. At the end: common mistakes and how to avoid them.

We break down how to make unit tests repeatable in Go: move “the outside world” into dependencies via minimal interfaces (TaskStore, Clock), replace them in tests using stub/fake, freeze “time” without time.Time, write table-driven tests, and check errors by meaning via errors.Is when wrapping with %w.