It’s part of the schema part, not the serialization part right? I don’t disagree you parse then validate, but in a schema that defines type of data, it’s not unreasonable to put limits on values.