On Tue, Dec 10, 2002 at 12:29:46AM +0000, Adam M. Costello wrote: > Soobok Lee <[EMAIL PROTECTED]> wrote: > > > > The output of ToUnicode can contain unnameprepped, prohibited, and > > > unassigned code points. Simply feed such a string as input to > > > ToUnicode, and the string will be output unaltered by ToUnicode. > > > > right. IDNA states that such outputs should not be displayed as native > > ones, but just as ASCII ones as it is. "must not" is meant for that. > > I think it is clear enough in drafts. > > I don't know what you could be referring to. Suppose X is a string > consisting entirely of uppercase Greek letters. X is not nameprepped, > because nameprepped things don't contain uppercase. X contains no ASCII > characters. ToUnicode(X) equals X, exactly, which means it is perfectly > acceptable to display X.
I mean that Punycode(X) [not Punycode(NAMEPREP(X)), not X ] can be inserted into RFC822 message headers. In that case, ToUnicode(Punycode(X)) should be treated differently than ToUnicode(Punycode(Nameprep(X))) . You are right if X is non-ASCII input, because toUnicode(X)==X. > > AMC
