Greek Casefolding sigma

John C Klensin klensin at
Fri Mar 28 17:51:15 CET 2008

--On Friday, 28 March, 2008 13:22 +0200 Vaggelis Segredakis
<segred at> wrote:

> Dear John, Patrik and Harald,
> I would like to ask you to remove the UTF character 03c2
> (Greek small letter final sigma) from the list of disallowed
> characters of the Draft "The unicode Codepoints and IDNA
> draft-faltstrom-idnabis-tables-05.txt, Feb 18, 2008" and
> place it in the PVALID list of characters.
> Please excuse me if this is not the proper way to ask for an
> Internet Draft to be changed.

--On Friday, 28 March, 2008 13:44 +0200 Sotiris Panaretou
<panaretou.sotiris at> wrote:

> Dear all,
> I would like to express our position here in Cyprus (.cy),
> about the final sigma case. We are in a process of opening the
> registration of IDNs using only Greek characters and we are
> informed about the changes of the corresponding protocol. 
> To give you an example, I will not be able to register my name
> which is σωτήρης and I will be forced to
> register σωτήρησ or σωτήρη which is
> incorrect.
> Please consider the point of view of the two ccTLD Registries
> .gr and .cy, which represent two European countries and would
> like to use in a proper way an official European Language.

Dear Vaggelis and Sotiris,

First of all, this way of requesting a change is entirely proper
and the information in your two notes is very helpful.  Please
note that the comments below are my personal opinions only;
there will soon be a working group whom you need to convince. 

As Patrik pointed out in his recent note, this character is
classified this way, not because of a decision we have made, but
because Unicode case folding rules dictate that outcome.  With
IDNA2003, that same Unicode case folding rule dictates that
final sigma be changed into medial-form lower case sigma which,
as you point out, is not a correct transformation (no matter who
does it) except possibly for comparison purposes.

In some ways, you can consider the classification produced by
the Unicode rules to be just a default that can be changed if
there is adequate reason for doing so.   However, the IETF has
traditionally been reluctant to make character-by-character
decisions and the consequent exceptions.  In addition, if you
have been following the mailing list in the last few weeks, it
is probably clear that there is also great concern about
introducing changes to how language characters (since final
sigma is mapped out by IDNA2003, treating it as a normal
character would constitute an incompatible change).

Again noting that this is a personal opinion, I see two ways for
you to proceed:

(1) You can try to persuade the Unicode Consortium to change the
casefolding rules.  I think it unlikely that you would succeed,
if only because I believe they have various stability rules that
would prevent such a change, but getting the issue resolved at
the source is, at least in principle, the best option.

(2) You can explain to the working group, not only why the
change in interpretation is correct and appropriate but either
how you propose that the incompatibility be dealt with or why it
is unimportant.   I personally believe that you have adequately
explained the "correct and appropriate" part in your two notes
and earlier materials posted to the list, but others may have
different opinions on that subject.   For the second, it would
be useful to know exactly what your registries, and any other
registries that might have registered labels in Greek, propose
to do to deal with differences between the IDNA2003 behavior
(map to lower-case medial sigma) and your proposed IDNA200X
behavior (treat final sigma as a normal character).

best regards,

More information about the Idna-update mailing list