Paul Hoffman / IMC �g�J�G > Solving the Traditional-Simplified problem in the IDN protocol would > have given more uniform results for users.
Agree. For example, LDH naming constraints are also built into the DNS instead of registration policies which might vary from registry to registry. > The intense discussions > over the past 18 months have shown that not only is it unlikely that > there will ever be an IETF-based solution, and it has also become > clear that a registration-based solution will likely give Chinese > users more sensible naming that anything than the IETF could > standardize. A written document presenting detailed ideas would be helpful for further discussions. I appreciate it if you could give a pointer to one such document. But, I am not sure whether you're referring to a global registration policy of equivalent names based on Han characters, or various inconsistent registration policies among all the registries including TLD, SLD, etc.? A hybrid approach combining both IDN protocol and registration policy is indeed an effective approach for the Chinese domain name problem. Without implementing basic equivalence matching function in IDN protocol, I am not sure if registration policy along would make sense. Chinese domain names service is not only provided by CNNIC, KRNIC, JPNIC, and TWNIC, but is also provided by other ccTLD and gTLD. Second level domain name holders can also develop their cDN registry business by selling the third level names to other parties. Thus, it is essential to build a standard Han variant table into the IDN protocol so that users are guaranteed to receive consistent resolution of Chinese domain names independent of the many registries. We also believe that the developing IRNSS protocol will add more equivalence matching capabilities to Chinese domain names, especially for those with context sensitive variants or language-specific variants. We believe that with international collaboration, there is an opportunity to develop a more reasonable domain name technology. Please do not rush into a standard to allow registries to dump premature products into Chinese Internet community. > For example, in the IDN WG, the proposals from CDNC > members only addressed part of the Traditional-Simplified problem > (the 1:1 mapping), while the registration solution can address all of > the problem in a manner that will help Chinese users. I understand that the 1:1 mapping table does not solve all the problems. But, I am not sure if, "the" registration solution had ever been proposed, and it's going to solve all the problems. > Due to the efforts of some CDNC members, the IETF is now well aware > of the "2^n" variant problem that is unique to the use of Han > characters in the Chinese language. Fortunately, using registration > guidelines will solve the problem. The only people seriously affected > by the size of DNS zones created by the "2^n" problem will be very > large, flat zones such as those managed by CDNC members, but they are > able to compensate for the size of zones with no changes to the DNS > protocols. There is obviously serious scalability problem associated with this large, flat zone solution. Though it seems to work fine for the time being. But, its management is complicated. Furthermore, with various dictionaries of Han variants, registries may choose to adopt different policy of name equivalence. Consistent resolution of domain names is thus hard to maintain among the registries unless character variant equivalence matching is part of the IDN protocol. Without supports in IDN protocol, do you have a feasible solution to mandate a consistent registration policy on Han characters among all registries interested in commercializing Chinese domain names, including TLD, SLD, etc? > Of course, the character variant problem exists in many different > languages, not just in Chinese. Because of this, each registry or > zone that includes IDN names needs to make decisions about name > equivalence (and therefore multiple registrations) before they enter > any IDN names into their zone. If you mean to say that each registry may decide its own name equivalence, then you are wrong. Language is international and thus name equivalence MUST be consistent among all registries. > Based on this input during Working > Group last call from the CDNC, it seems clear that we need to explain > these points more in the IDNA document. No, explanations are not good enough. The IDN protocol has serious roblems in dealing with Han ideograph. We need to pay more attention on those problems. > > --Paul Hoffman, Director > --Internet Mail Consortium
