> It may lead to inconsistent exception conditions due to > changes in language specific normalization procedure. > For example, a UCS code point A in language tag T is > excluded in Version X, as there is no blocking for any > UCS code points in place yet. Due to upgrade Version X+1, > T(A) has been newly included tagged code blocks, then > an input contains A fail in Version X, but will succeed > in Version X+1. Due to failure in X will be coded with a > defaulted ACE as a legal name, then the same input > will not be fail in X+1, an inconsistency will resulted.
See nameprep-08. it is discussed extensively how to deal with this situation. > If the WG has consensus on NOT using language tag > mechanism, it is better to voice this and let > VeriSign users go ahead. dont see how verisign testbed have anything to do with idn wg. please keep things in focus. -James Seng
