IDNA 2008 Question Re: "Confusable" Characters in Domain Names
John C Klensin
klensin at jck.com
Sat Nov 6 00:16:58 CET 2010
--On Friday, 05 November, 2010 15:55 -0700 Mark Davis ☕
<mark at macchiato.com> wrote:
>> One thing to add to this, which is that registries and
> registrars were strongly advised, from the beginning, to not
> register names containing characters other than letters,
> digits, and the hyphen (with the first two groups expanded to
> cover non-ASCII characters).
> Where were these recommendations made?
First of all, the example and questions were about
non-ASCII-symbol.com, not some deep subdomain. For the deep
subdomains, you are quite correct; this was not publicized as
much as I (personally, at least) would have liked.
The reason it wasn't publicized by, e.g., including it in the
IDNA2003 document set is that WG at the time made an explicit
decision that such recommendations were Someone Else's Problem.
The IESG did add a note that points out that registration rules
are a necessity, but did not incorporate that note in RFC 3490.
In retrospect, that may have been a mistake too, but the
decision was very explicit and had the backing of the WG at the
For top-level domains, there are ICANN guidelines, supposedly
binding on gTLDs and strong recommendations to ccTLDs, that
match the rule I described and also advised making the
particular characters being used public in an IANA registry.
Some gTLDs and ccTLDs chose to ignore those rules, but those
rules were discussed enough in ICANN's domain name
constituencies and GAC that it would be hard to argue that any
of the registries, or even any accredited registrars, were
completely unaware of them.
And, of course, one of the reasons the fundamental restrictions
are now built into IDNA2008, rather than depending on guidelines
that people might have trouble finding, is precisely because of
the problem you describe.
More information about the Idna-update