RE: Case mapping of dotless lowercase letters

From: Kent Karlsson (kentk@cs.chalmers.se)
Date: Wed Dec 17 2003 - 08:24:30 EST

  • Next message: Michael Everson: "RE: [OT] CJK -> CJC (Re: Corea?)"

    > The difference here is that Germans recognise ss and sharp s
    > as variant spellings in the same words,

    Not altogether, taking into account spelling rules.
    They are *ordered* the same, but that is another matter.

    > whereas in Turkish i and dotless i are
    > quite different letters, just as in Swedish, Turkish and
    > German o and o
    > umlaut are quite different letters. I know Germans tolerate o umlaut
    > written as oe,

    No, again an ordering rule, not a spelling rule. It has been used as
    fallback too, like ss for ß. But it is not correct spelling.
    (I will not go into the German spelling reform, since I'm not well
    familiar with it.)

    > but I don't think Turks do. But surely the whole point of
    > getting away from ASCII-only domain names is to respect national and
    > language-specific alphabets. What is needed for Germany and Sweden
    > should not be denied to Turkey.

    There was never an intent do deny Turkey anything. The thing was that
    the uppercase of i is I (usually) and the uppercase of ı is also I, so i, I,
    and ı used to be folded together (to i) in the drafts for IDN. Apparently
    that was deemed to harsh and was modified. (I think I complained at
    some point, but it wasn't modified then, but apparently much later.)
    Still for IDNs there is no language dependence in the case folding, as
    there are for the case *mappings*. So "I" is turned into "i" (not "ı") also
    for Turkish for IDNs. On the other hand, domain names are most often
    written in lowercase anyway.

                    /kent k



    This archive was generated by hypermail 2.1.5 : Wed Dec 17 2003 - 09:21:28 EST