Standardizing on IDNA 2003 in the URL Standard

John C Klensin klensin at
Thu Aug 22 03:50:47 CEST 2013

--On Wednesday, August 21, 2013 17:30 -0400 John Cowan
<cowan at> wrote:

> John C Klensin scripsit:
>> That would have meant no separate code point for a final
>> sigma in Greek;
> See my earlier post for why that's not possible.
>> no separate code points for final Kaf, Mem, Nun, Pe, or Tsadi
>> in Hebrew;
> This is even less possible.  In Hebrew, a pe at the end of a
> word is always /f/.  But in Yiddish, there is a contrast
> between /p/ and /f/ in final position that Hebrew does not
> have, and in that case a non-final pe in final position is
> used for /p/, sometimes but not always with a dagesh (embedded
> dot).
> In short, Greek and Hebrew positional variants require AI-hard
> algorithms.


As I said, I know a rather long list of reasons why the model I
tried to describe was not reasonable or desirable and suggested
that people more expert would have an even longer list.  

Your note only reinforced that impression.

The point was merely to suggest that there are no simplistic and
still fully general solutions to the relevant set of problems.
Your note reinforce that suggestion too.



More information about the Idna-update mailing list