draft-liman-tld-names-00.txt and bidi
John C Klensin
klensin at jck.com
Sun Mar 8 05:20:08 CET 2009
--On Saturday, March 07, 2009 20:38 -0500 Lyman Chapin
<lyman at acm.org> wrote:
> If you type "http://3184.108.40.206" into an old version of
> Internet Explorer (if I remember correctly, Windows 95
> vintage), you'll get to the main IETF web page as surely as if
> you had typed "http://220.127.116.11" (or "http://www.ietf.org"
> for that matter). This is just an artifact of the way in which
> software interfaces interpret what users enter, not an
> observation about the DNS or the encoding of data fields -
Well, exactly. I'd go a step further and suggest that it
illustrates one of the problem that can occur when software
interprets something as an address that couldn't possibly be an
address and then tries to treat it that way. And that is an
example why I'm opposed to any model that would consider "666"
as a valid TLD because it couldn't possibly be an element of an
IPv4 address. If something construes it as "an address" because
it is numeric, two separate bad things happen: (1) It will
either be rejected completely because the code finds out that it
isn't a valid address, or the code will start truncating,
dropping bits, or doing modular arithmetic to treat it as an
address, possibly resulting in an unintentional and false hit.
And (2) it will never be looked up as a domain name in spite of
the fact that it cannot be an address.
It is not, in any event, a DNS problem, although ICANN could
certain make it one.
More information about the Idna-update