RE: Mississippi Hißes

Shawn Steele Shawn.Steele at
Fri Dec 11 10:48:57 CET 2009

Sorry, stayed up too late, didn't really mean to rant.  I'd just like this to be over with :)  I can live with PVALID, but I don't really think there's an effective way to do transition :(.  IMO there's not much difference long-term, but continuing to associate them in the standard would at least encourage consistent behavior, at the price of figuring out how to deal with the presentation end.  I forgot for a moment that I was letting that go.

Martin's comment about the Swiss registry, for example, is interesting.  In practice, the Swiss standardized to "ss", however they might occasionally use eszett for stylistic or typographic reasons.  (Like we might have "ye olde...")  In that case it's a style decision, and not bundling would probably be even worse than in .de or .at, since the phonebook probably has the ss form of their name.  

I expect that most lower level domains will just differentiate by default because they won't bother to be aware of the problem.  That doesn't mean it's good behavior, just what they happen to do by accident.  So I see the lack of consistency as a real problem.  Sure, most languages don't get that kind of special treatment, but this character IS special, and we already had it in 2003, so it IS now special.  I'll quit before I start leaning too much in this direction again.


From: idna-update-bounces at [idna-update-bounces at] on behalf of Shawn Steele [Shawn.Steele at]
Sent: Friday, December 11, 2009 1:32 AM
To: Cary Karp; idna-update at
Subject: RE: Mississippi Hißes

DNS can convey nearly perfect language-specific information about ß really easily:  That character's only used in German :)  If it shows up in DNS, it must be German.  Any other use is pretty much "for fun".  Even a random collection of characters would still have a German-like feel just because it was included.  In ANY other language it behaves a lot like a symbol, unless you intend to mean it as a borrowed German word.  For completeness, it's nice to think that it's might be used somewhere else for some other reason, but frankly {heart} is more useful than eszett in a non-German context.

Presumably .at and .de can't tell if a word's German or English or made-up.  They have said they'd bundle though, which kind of implies they don't think there's need to distinguish for lookup, at least not worth breaking 2003 over.

Note that I've never said that either label is unreasonable at any level.  I've just said that it's not worth having them point to different servers.  That's a way different concept that I don't seem to be able to communicate.  Both are great for registering things, but there's no reason to cause a lot of pain to make them unique, because, in practice, they are NEVER going to be used for differentiation.  (One may not be used at all for non-german languages, eg:, but still, they'll NEVER be used to point to different resources.  The only time they'd point different places is if someone didn't care about German, in which case they shouldn't be using Eszett at all.)

I'm not a lawyer, but I suspect you can't even register and fuß individually.  The owner(s) of one (or the other) could just go to a German court and get the other one to cede ownership.  Maybe even a non-German court.  This is complete speculation and I'm bound to get flamed for it.


From: Cary Karp [ck at]
Sent: Thursday, December 10, 2009 11:40 PM
To: idna-update at
Cc: Shawn Steele
Subject: Re: Mississippi Hißes

 > I get that Herr Fuß and Herr Fuss are different people
 > and pronounce their names differently and all that

Last time I looked, "fuß" and "fuss" were totally unrelated words in the
dictionaries of two languages that figure prominently in the present
discussion. Either is a perfectly reasonable second-level label in any
unrestricted gTLD, and certainly in many ccTLDs as well. Since when is
there a presumption that second-level labels in, say, .DE or .AT, are
intrinsically related to German?

How many times does it need to be noted that:

- the DNS has no way of conveying language-specific information about a

- the DNS in no way requires or assumes that a label has any language

- a domain does not intrinsically bequeath any categorizable information
to its children beyond its own identity?

Idna-update mailing list
Idna-update at

More information about the Idna-update mailing list