Table-driven tests

func TestAdd(t *testing.T) {
    tests := []struct {
        name string
        a, b int
        want int
    }{
        {name: "positive", a: 2, b: 3, want: 5},
        {name: "negative", a: -2, b: 1, want: -1},
    }

    for _, tt := range tests {
        t.Run(tt.name, func(t *testing.T) {
            if got := Add(tt.a, tt.b); got != tt.want {
                t.Fatalf("Add(%d, %d) = %d; want %d", tt.a, tt.b, got, tt.want)
            }
        })
    }
}

Tests live in files ending with _test.go; test functions start with Test and accept *testing.T. Prefer small inputs that explain the behavior over a large fixture that hides it.

Commands

go test ./...
go test -race ./...
go test -coverprofile=cover.out ./...
go tool cover -html=cover.out
go test -run '^TestAdd$/negative$' ./path/to/package

Coverage shows which statements ran, not whether the assertions were meaningful. Use the race detector in CI where its extra time and memory are acceptable.

HTTP and cleanup helpers

Use httptest.NewRecorder for a handler in isolation and httptest.NewServer when client behavior and the full HTTP stack matter. Register cleanup with t.Cleanup; use t.TempDir for per-test files and t.Setenv for temporary environment variables.

Call t.Parallel() only after checking that the test does not share mutable state, fixed ports, process-wide configuration, or a database fixture with other parallel tests.

Further reading: testing and net/http/httptest.