RE: Character identities

From: Michael Everson (
Date: Wed Oct 30 2002 - 12:18:13 EST

  • Next message: Marco Cimarosti: "RE: RE: Character identities"

    At 17:57 +0100 2002-10-30, Kent Karlsson wrote:

    > > 1. We have all seen examples, in print, in signage, and in
    >> handwriting of German umlauts being displayed in each of those ways.
    >> Obviously the underlying encoding of them is the same, as is the
    >> intent.
    >The underlying encoding *may* be the same (if there is an encoding
    >at all...). Still, I claim, it should not be up to the font
    >designer to make a font that shows e.g. an a-with-e-above glyph
    >for a-diaeresis *without also* the font being explicitly requested
    >(via some "higher-level protocol") to do such a mapping, via a
    >"hist" feature (off by default) or whatever other mechanism. Such
    >a mapping *amounts to* a transient character-to-character mapping.

    Nonsense. Sorry, your claim is just plain silly. I have a book on my
    shelf that has a nice special font which has, oh, I don't remember
    which, it is either a capital O with the two dots inside it or it is
    a capital O with a small letter e inside it. In any case it is
    obvious that Í is the character being used, because of the language
    of the book, and it's just the creativity of the font designer that
    draws this one way or another.

    I'll say it again: ń when drawn with an e above it instead of two
    dots (COMBINING DIAERESIS ABOVE) is a perfectly acceptable if rare
    variant of ń, in German at least, and if it happens to be the case
    that you can ALSO draw that by using the COMBINING SMALL LETTER E
    ABOVE is accidental. The latter was not added in order to account for
    the trivial font-choice typography used in German menus and book

    >Just as I think an "author" (I use that in a general sense)
    >should be in charge of the spelling in a document, the "author"
    >should be in charge of what diacritics are used. Would it be
    >a good idea for a British font to change "color" to "colour",
    >"i18n" to "internationalisation"?

    Oxford recommends "internationalization". And I am sorry, Kent, but
    that is an orthographic argument about strings of characters. Fonts
    don't change the spelling. THE SPELLING IS IN THE ENCODING. If I
    write the word Karlsson and the O in my font looks like a pumpkin, I
    have not changed the spelling of your name to one using LATIN SMALL

    >AAT fonts can in principle
    >do that (via glyph index mappings executed through a finite
    >automaton, but that is beside the point), so should they? Is
    >such a font (if it did this mapping by default) a Unicode font?
    >Each item in these two example pairs are seen in print (etc.)
    >and they are "known" to "mean" the same within each pair...
    >There are signs (and printed texts) that say "G°teborg"; but
    >we usually spell that "G÷teborg". Does that mean that the
    >underlying encoding (if any) therefore must be the same (the
    >same city is intended...), and ° is just a glyph variant of ÷
    >(or the other way around), and a ((Unicode)) font may display
    >÷ as ° (without being asked to perform any extraneous mapping).
    >Say the font is made for Norwegian. Is this all up to the font
    >designer? This is an exact parallel to what we started off with.

    That is not what is done in the limited realm of glyph variants for Í
    used in German book covers and so on, which is where you see those
    glyph variants. It is a different thing from G÷teborg and G°teborg,
    which is an artifact of there being a "svensk ÷" and a "norsk °" and
    the users in question having limited typographic control (such as
    typewriters or 7-bit character sets).

    The COMBNINING SMALL LETTER E ABOVE is intended for use in a very
    limited realm, editions of medieval German literature. The use of a
    similar glyph in other contexts is a different thing entirely, and
    it's just something we are going to have to live with.

    Michael Everson * * Everson Typography *  *

    This archive was generated by hypermail 2.1.5 : Wed Oct 30 2002 - 12:55:59 EST