There are output requirement routines that can be used to avoid the output being proposed. In this case you won't get any NAST record, so nothing to process.
Other validations can be done in the output program, naturally. But normally if a validation no passed then the error status would be set and log updated with the error text. The message then can be repeated after the errors were addressed. I guess if you create a custom output program you can do whatever you want in it. But wouldn't it be very confusing to the users when the message is not processed yet there is no error message? How would anyone be able to tell a message that was not touched from the message that didn't pass a validation? You might want to think this through.
