Logger Failure

Here's the code I used to create Linko's logger:

func initializeLogger(logFile string) (*log.Logger, error) {
	if logFile != "" {
		file, err := os.OpenFile(logFile, os.O_WRONLY|os.O_CREATE|os.O_APPEND, 0o644)
		if err != nil {
			return nil, fmt.Errorf("failed to open log file: %w", err)
		}
		multiWriter := io.MultiWriter(os.Stderr, file)
		return log.New(multiWriter, "", log.LstdFlags), nil
	}
	return log.New(os.Stderr, "", log.LstdFlags), nil
}

func run(ctx context.Context, httpPort int, dataDir string) int {
	logger, err := initializeLogger(os.Getenv("LINKO_LOG_FILE"))
	if err != nil {
		fmt.Fprintf(os.Stderr, "failed to initialize logger: %v\n", err)
		return 1
	}
	// ...
}

Notice that if an error occurs when opening the file, I return an error from initializeLogger, and then in run() I write a message to os.Stderr and return a non-zero exit code. If you used log.Fatal or log.Panic instead, you might have a couple of problems in your code that would:

• Make it impossible (or very difficult) to test that behavior in a unit test.
• Prevent the program from running any deferred functions or doing other cleanup.

Instead, I prefer to let the caller of the initializeLogger function decide how to behave in the event of a failure! Then, when it's time to handle the error (in the run function), this is one of the few times it's okay to log without a logger (by using fmt.Fprintf) because it was the logger itself that failed to initialize!