One thing I am interested in some opinions on is whether we should make UBL
behave as EDI does, i.e., for alphanumeric elements there must be at least
one non-space character in the element. This may remove a few
implementation complexities, but will add a bit to the schema design. We
would have to create a type from string which imposed a pattern (via a
regular expression) of at least one non-space character. Certainly doable.
I am curious, though, about processing overhead. I know that parsers must
do pattern validation on numerics and dates, but I wonder if imposing it on
all strings would slow down processing appreciably. Any thoughts?
--
Michael C. Rawlins, Rawlins EC Consulting
www.rawlinsecconsulting.com
