> I don't know what the right solution here is. Wrapping all values in Option > seems like an overkill. Returning 0 when NULL is returned where an int is > expected sound incorrect. Any ideas?
Ormin doesn't use the `seq[string]` idea at all, side-stepping the problem. But here's a different idea: Instead of `seq[Option[string]]` a row could be `tuple[data: seq[string], missing: seq[bool]]`. Then you can easily ignore the 2nd tuple component when you know there are no NULLs in your schema (and there shouldn't be NULLs!).
