Consensus Call on Latin Sharp S and Greek Final Sigma

Shawn Steele Shawn.Steele at
Tue Dec 1 10:57:59 CET 2009

> I think we violently ;-) disagree on what "use" implies.


> For me, it is to actually use the ß in a registered domain name,
> in an email address and a certificate, and the ability to get a ß
> back from a lookup in (for example) DNS that have a domain name
> in the RDATA field. For example an SRV record, or NAPTR. 

I agree ('cept I don't know all the records), which is why I suggested a display form :)

> And the ability for someone that do have "ss" in a domain name 
> to choose not to have that match ß (as Harald explained earlier).

Who cares?  If I care about ß, then I probably like the behavior (  If I don't it doesn't really matter.  After all, I don't get to choose that maps to, even though looks really stupid.

> For you, it is the ability for a user to "type" in a ß, and then get the
> same result as if one typed "ss". 

Qualified yes.  Obviously they differ.

> And get this by making ß DISALLOWED so that we are 100%
> sure it is never used in protocol elements, but instead "ss" is used.
> Certificates, NAPTR, DNs etc.

I think that's the problem.  The "strict" use of a name in something like a certificate may be desirable for the form in the certificate to be the display form.  The more flexible use as a general pointer to a server allows (perhaps) for another mechanism to provide the proper display form (so the "canonical form" is more like a hash value that just happens to be fairly human readable.)

And that, I think, is the fundemental difference between the two "factions" on this list :)  The "strict" people want things very rigorous and no-mapping, but still want some latitude for linguistic preferences.  The "mapping" crowd seem to want more flexibility in the comparisons.

I'm curious though: if Fuß was "correctly" spelled "Fuß", and they had a certificate stamped as such, would it really be a good idea to allow a different entity to have a certificate stamped as  (Perhaps even from the same authority?)  If it is good for a Germany user trying to understand the certificate, would it still be good for a Swiss user?  (I'm asking).  I see that it's good for the entity to be correctly identified "Fußball", but does that really help the person trying to validate the certificate?

I'll have to think about that some.  I've been thinking more from "the other direction".  Is there a way to get certificates stamped with the Eszett instead of ss (or ss instead of Eszett), but still allow queries to find the URL both from Switzerland and Germany?


More information about the Idna-update mailing list