The fix for the validation running twice here is definitely welcome, good catch.

However, I am not totally sure about the "print to stderr is also considered an error".
The way I see the validation functions, is that there is a simple contract that the developer needs to fulfil:

1. The function name needs to start with validate_
2. It prints if there is an error.

The printing is simply a way of returning a string, since such a feature does not exist in bash.
Also take into account that the printed string cannot just be an output of some external program (which may print its errors to stderr), since it must be something that makes sense in bashly scope: "docker must be installed" or "must be a number".

While I dm not totally opposed to capturing stderr from these validation functions as if it was stdout, I think it dilutes the message and may add confusion.

I am on the fence, hoping to either convince you or be convinced otherwise.

[external program may print errors to stderr]

That was also my thought against this solution.

I'm convinced that the bashly user should handle external program error messages and print something meaningful to the end user.

Instead of changing the behavior here, I think we can make the "validation contract" more explicit in the docs. Then I'm going to open a PR to bashly-book instead.

Closing this one.

Cool.

This is the documentation - bullet 3 in particular.
https://bashly.dev/advanced/validations/#custom-validations