From: "Roozbeh Pournader" <[EMAIL PROTECTED]> > On Mon, 28 Apr 2003, Mark Davis wrote: > > > BTW, the ICU demos have been all upgraded to Unicode 4.0, on > > http://oss.software.ibm.com/icu/demo/. > > They include: > > > > [...] > > > > IDNA Demo > > This simple demo performs IDNA transformations as described in RFC 3490. > > But isn't the IDNA repertoire limited to Unicode 3.2?!
Probably for now, until new characters are accepted in IDNA, which has always used only a approved subset of the Unicode/ISO10646 defined set, even with Unicode 3.2. IDNA defines, for security reasons, some other equivalences that are normative for use in domain names, even if they are not included in NFC/D and NFKC/KD data. (Look for exemple the additional equivalences for dashes, and Unicode standard characters whose usage is forbidden within IDNA, or equivalences added between similar letters belonging to distinct scripts within Unicode). IDNA really adds restrictions based on the fact that two Unicode characters that are not canonically equivalent or compatibility equivalent, still have too similar glyphs in most fonts, and thus must be "unified" with additional equivalences. A domain name is not an unrestricted general text, so it has additonal constraints that are out of scope of the Unicode standard which does not consider the security requirements for each application that uses it as a base standard. Unicode 4.0 is still too new for IDNA. Give some time to IDNA authors to study the additions in Unicode 4, and publish a new subset of acceptable characters and new IDNA-canonical equivalences, or prohibitions. There's nothing wrong however in the ICU announcement (IDNA is not a Unicode specification but a separate IETF/IEC technical standard), because it lists all demos that have been upgraded to the latest versions of each specification.
