aa.net.uk Broadband - Broadband you can work with

Skip to Navigation / Skip to Content

Knowledge base International domain names

Domain names were designed to contain just letters and numbers (and hyphen) but with the internet spreading somewhat there is demand for more complex characters to be used. The main thrust of this is for domains in totally different character sets like Chinese or Greek, etc.

Well, there is a way of making such characters using UTF-8 and the domain name system (DNS) protocol allows any characters including UTF-8. Even the popular DNS servers like bind handle UTF-8 characters. Most applications just pass the domain to a resolver library without checking it and most resolver libraries just put that in the DNS request without checking it. So this should be simple - all that was necessary was an RFC saying domains could now use UTF-8 characters. Then a few libraries and resolvers may have to have unnecessary checks removed or changed and pretty much everything would just work.

Sadly not! The way this has been done is unnecessarily complex and will mean IDN support is likely to be flaky at best for many years...

Puny code

The decision is that every application must understand international domains and convert them. This means browsers, mail clients, command line tools (like ping, dig, whois, telnet, ftp), and - well - just loads and loads of programs have to change. At present it is pretty much just the main popular browsers that understand international domains and that is it!

The domain is converted to one that only uses letters, numbers, and hypens, using punycode which is a complex algorithm but at least well defined. Then that is looked up instead. So the domain aa®.com is converted to xn--aa-9da.com.

Preparation for punycode

Of course it is not that simple either. The domain has to be prepared, and this is a complex mapping. Some characters are mapped to others. So, for example, ½ is converted to a 1 and a fractional symbol and then 2 and then that is converted to punycode.

This means that it is not actually at all simple for applications to convert a domain simply to the punycode equivalent. Why this is not done in the standard resolver libraries is beyond me. It also means there will be mistakes as one application will follow the specification more correctly than another. Worse still is new unicode characters that come along may mean additional mappings are needed which means that all applications then have to be updated!

Limiting what you can register

Another real annoyance is the main registries like .com and .net have very complex rules on what you can and cannot register. You cannot mix different character sets and cannot use a whole load of special characters that would be sensible, useful or even fun.

Do AAISP handle international domains

Yes we do. We will try and register a domain if you ask for one as normal. Our control pages are smart enough to display the domain as it should be, i.e. converting the punycode to UTF-8 for you.


We have the domain aa®.com which converts to xn--aa-9da.com. You browser should cope with <a href="http://aa®.com/"> links like this or this.