Stupid U-label question

Frank Ellermann hmdmhdfmhdjmzdtjmzdtzktdkztdjz at gmail.com
Tue Aug 19 20:02:59 CEST 2008


Paul Hoffman wrote:

> I can't tell whether or not you are asking about
> IDNA2003.

My error and the following "xn--xn--" experiments
used <http://josefsson.org/idn.php> cross-checked
with <http://www.imc.org/idna/>, that is IDNA2003.

But my question was about IDNA200X:  I *think* we
have it clear that LDH-label means what the name
says, a DNS domain label (1..63 octects) limited
to Ascii leters, digits, and hyphen; not starting
or ending with a hyphen.

We also had a clear U-label definition:  At least
one UTF8-non-ASCII (or actually any encoding that
can be transformed into UTF-8) where encoding it
as A-label and decoding the result "works" and
yields u = A2U( U2A( u )) for a valid U-label u.

What I've done was this:  Let u1 = ä (umlauted a),
that's A-label xn--4ca.  Then let u2 = xn--4caä,
and extract u3 = 4caä to bypass an IDNA2003 rule.

For u3 the A-label is xn--4ca-cxa.  Therefore in
theory u2 could have an A-label xn--xn--4ca-cxa
IFF that is not prohibited somewhere in IDNA200X.

> Step 5 of section 4.1 seems as concise and clear
> as it could be.

Yes, RFC 3490 4.1 (5) is clear.  I had no doubts
about Simon's or your IDNA2003 code, I'm just not
sure where it will be covered in idnabis-protocol:

At the moment 4.3.2.1 covers it, but I *think* we 
don't want to exclude "--" in positions 3 and 4.

Now we need "something" covering RFC 3490 4.1 (5),
because so far all agreed that it was a good rule.

 Frank



More information about the Idna-update mailing list