Problem: You want to provide additional information and context to an error you receive before returning it as another error.
Solution: Wrap the error you receive with another error you create before returning it.
There are a couple of ways to wrap errors. The easiest is to use fmt.Errorf again and provide an error as part of the parameter:
err1 := errors.New("Oops something happened.") err2 := fmt.Errorf("An error was encountered - %w", err1)
The %w verb allows you to place an error within the format string. In the example, err2 wraps err1 . But how do you extract err1 out of err2 ?
The errors package has an Unwrap function that does precisely this:
err := errors.Unwrap(err2)
This will give you back err1 .
Another way of wrapping an error with an error is to create a customized error struct like this:
type ConnectionError struct { Host string Port int Err error } func (err *ConnectionError) Error() string { return fmt.Sprintf("Error connecting to %s at port %d", err.Host, err.Port) }
Remember, to make it an error, the struct should have an Error method. To allow the struct to be unwrapped, you need to implement an Unwrap function:
func (err *ConnectionError) Unwrap() error { return err.Err }