Re: Latin letter GHA or Latin letter IO ? (was: Pre-1923 characters?)

From: Philippe Verdy (
Date: Sat Jan 03 2004 - 17:23:32 EST

  • Next message: Michael Everson: "Re: Latin letter GHA or Latin letter IO ? (was: Pre-1923 characters?)"

    From: "Michael Everson" <>
    > The glyphs are not normative.

    I thought that you were exactly promoting the reverse. That's why I wanted
    to moderate things about glyphs. But Unicode still states that they are
    "representative", and it publishes them as they are necessary for correct
    identification of character identities.

    But if you want to insist more with your position, why not simply dropping
    completely all glyphs from the Unicode standard? I'm sure many people would
    not be happy, because the character identity would only be demonstrated by
    the normative properties, which include their often misleading assigned

    I maintain that if you remove the glyph shown for latin letter oi
    (considered only as informative and not mandatory in any of its aspects),
    and just keep its normative name, then many people will think that the
    encoded character really represents a letter named or pronounced "oi". Which
    is completely wrong in our case. But would allow people to use the assigned
    code point to represent the L-shaped character "i with lower-right hook"...

    Can't you admit the problem here with a so extreme position? You have here a
    position which is much more modulated in the published Unicode standard,
    which states that the published "representative" glyphs are representative,
    which means that at least the exhibited glyph can be safely used in ALL
    cases (yes, even if actual glyphs often have contextual forms, sometimes
    mandatory in some scripts or languages) to represent the character (also
    even if other glyphs are possible) without causing interpretation/reading
    problems in the rendered text.

    The problem we were discussing here is that only the informative and
    non-normative properties are giving the appropriate identity of the encoded
    letters, but NONE of the existing normative properties... So I really give
    some credits to these glyphs (and ISO/IEC 10646 too... as it just encodes
    repertoires of characters for the need of allowing unambiguous conversion of
    many legacy encodings into a common and unified repertoire). I did not say
    that these glyphs are mandatory for conformance, but the exact way these
    variations in glyphs are allowed is not described. (meaning that font
    designers must be extremely prudent before creating variants of these
    glyphs, as it may produce confusive texts in some tricky cases or with some
    languages for which the font design was not tested).

    This archive was generated by hypermail 2.1.5 : Sat Jan 03 2004 - 17:56:10 EST