Currently, if there are multiple constraint violations that identify the same starting column position on a line, then only one is including in the list of violations returned. This reduces the clutter from a cascade of error messages, but which violation is selected is arbitrary and is often not the one that is the real cause of the error.