logging

The best way to handle error logging is to avoid logging and returning errors simultaneously. You should let errors bubble up to the caller and only log them at terminal boundaries, preventing duplicate logs and ensuring clean error propagation.

To choose between Info and Debug log levels, use Info for operator-facing messages and Debug for developer-facing diagnostics. This structured logging discipline creates a clear boundary, keeping logs relevant and actionable for each audience.

You should remove Warning and Fatal levels to enforce log-level discipline and reduce noise. Consolidating to Info and Debug levels creates minimalist logging, making logs more meaningful and preventing unnecessary fatal-level crashes in library code.

Yes, you can use structured logging during code review to improve observability. Reviewing logging code involves checking for contextual key-value fields, ensuring proper Info versus Debug boundaries, and verifying that library-level error logs are removed.

You need terminal error handling instead of logging directly in libraries when an error reaches the end of its execution path. Libraries should return errors to the caller, while terminal handling logs the error only at the application's top level.