You parse into an applicative validation structure, combine those together, and then once you've brought everything together you handle that as either erroring out with all the errors or continuing with the correct config. It's easier to do that with a parsing approach than a validating approach, not harder.