If 'lax' and dfdl:textNumberRep is 'standard' then grouping separators can be omitted, decimal separator can be either '.' or ',' (as long as this is unambiguous), leading positive sign can be omitted, all whitespace is treated as zero, and leading and trailing whitespace is ignored. Also the exponent is also optional and assumed to be '1' if not supplied
The part that seems odd: "all whitespace is treated as zero" So according to this, if the data is: " 9 9 " then with textNumberCheckPolicy="lax", the leading/trailing whitespace is trimmed and the middle whitespace is converted to zero, so this is equivalent to "909". This feels like a spec bug to me, but I wanted to have others confirm this. Note that I don't think this behavior is consistent with what icu4j does. It simply ignores the middle space and converts the data to "99". It's possible we're not setting an ICU4J flag, though, I don't see an obvious ICU4J flag that would control this. Mike Beckerle | OGF DFDL Workgroup Co-Chair | Tresys Technology | www.tresys.com Please note: Contributions to the DFDL Workgroup's email discussions are subject to the OGF Intellectual Property Policy http://www.ogf.org/About/abt_policies.php