But the ICANN normally has a signed contract with the registry that it wil operate for the general benefit of the communities, using *fair* commercial practice. The same contract then assumes that this will be observed by registrars who should also operate fairly.
If a registrar sells a name that is not functional, according to any approved protocol, that sale is not fair, the purchaser pays for a name that will not be functional with most applications for most users of these applications. May be the buyer does not care and just wants to harvest a security hole found in some softwares that fails to also check the validity of such domain name (in that case it is very likley that the buyer violated the rules on purpose just to create problems to others). Otherwise, this was just an error from the buyer, that will be very dissatisfied when he will see that the domain he just bought had an error that the registry should have detected. So the registrar just stole the money by negligence... Will the buyer get the right to renew its application for a domain name, or be refunded ? In all cases, those registrations are just pollution of the worldwide DNS and will cause unnecessary traffic and disputes. ICANN should really investigate how the RFC's are enforced by registries at least, so that no registrar can't pollute them by bad names. If later there's any desire to moentize some new sets of possible names (in domain names or in other protocol field names), this will require first a serious investigation about compatibility issues (at least) and the applicable naming rules, as well as the condition of sales (who can sell them, who fixes the prices due to the registry, how much will go to ICANN or ISOC or IETF or others, including the Unicode consortium if it applies as a candidate hold to hold some registry or wants to act as a registrar to some classes of names usable in some protocols...) Note that the ISOC is not the only one to rule various protocol registries. There are tons of protocols that have been opened for use by other with extension mechanims requiring a registration of some name or identifier, in a registry that the initial protocol owner will maintain. Plus many mappings used to map a name used in one protocol into another protocol (such as URNs). Le 28 février 2012 11:52, Stephane Bortzmeyer <[email protected]> a écrit : > On Sun, Feb 26, 2012 at 09:55:11PM +0100, > Jeroen Ruigrok van der Werven <[email protected]> wrote > a message of 14 lines which said: > >> So I guess they did not check the codepoint categories in their >> validation step then? > > Probably. People are free to ignore RFCs (or UTRs). > >
