> I agree. ACE should stand for the UCS into ASCII A-Za-z0-9- encoding. > No length restrictions, no nameprep. Though the UCS must be normalised > so that it is well defined. > That is, ACE should stands for: > - UCS normalised into Unicode form C > - then encoded into the characters A-Za-z0-9-
I wouldn't mix normalization with this. First, in the case of IDN, the normalization is KC, and it follows some character mappings, and also precedes some other code point validation (forbidden code points). In addition to that, ACE drafts do not require the normalization. The RACE one doesn't even mention it, and the AMC-ACE-Z simply acknowledges the fact that this particular ACE, used in the context of IDNA, is applied on strings that went through Nameprep first. I am for keeping the separation between the two processes. Though I see why one would want to mix them, since people typically refer (improperly I believe) to the ACE-encoded IDN and not to the ACE-endoded Nameprepped-IDN, while meaning the latter. Maybe it should be the IDNA-encoded IDN? YA
