Lectures

Let’s break down how to control test execution in Go using go test: enable verbose output with -v, precisely run tests and subtests with -run, speed up failure diagnosis with -failfast, and understand when and why the cache kicks in (and how to honestly disable it with -count=1).

We break down what coverage in Go actually measures (coverage by statements), why the percentage is only a “thermometer”, and how to use the flags -cover, -coverprofile, and -covermode correctly ( set/ count/ atomic). Using a practical example with the ParseStatus function, we show how coverage highlights forgotten branches (errors, default in switch, early return) and what typical mistakes people make when “chasing the percentage”.

We break down how to read coverage reports in Go: from generating the coverage.out profile via go test to analyzing the per-function summary ( -func) and visual source highlighting ( -html). Learn to treat coverage not as a “race for a percentage”, but as a risk map: if branches, early return, error handling, and “dead” code.

We break down how, in Go, documentation examples can be turned into auto-verified tests: how Example functions work in *_test.go, how the runner intercepts stdout and compares it to the expected text in the // Output: comment, why examples do not replace TestXxx, how to name examples to bind them to the API, and how to avoid flakiness due to nondeterministic output ( map, time, randomness).

We break down why Go needs fuzz tests, how FuzzXxx differs from a regular TestXxx, how to formulate useful invariants (idempotence, round-trip), how to set seed inputs via f.Add and run fuzzing with the go test command and the -fuzz flag. Using the educational Tasker app as an example, we build two fuzz tests: title normalization and ID parsing.