RE: Question regarding U+00A6 BROKEN BAR

From: Philippe Verdy (verdy_p@wanadoo.fr)
Date: Sat Dec 22 2007 - 07:52:14 CST

  • Next message: James Kass: "Acceptable alembic glyph variants"

    > De : unicode-bounce@unicode.org [mailto:unicode-bounce@unicode.org] De la
    > part de Karl Pentzlin
    > Envoyé : jeudi 20 décembre 2007 18:25
    > À : 'Unicode List'
    > Objet : Question regarding U+00A6 BROKEN BAR
    >
    > Some standard keyboard layouts, like the British and the Swiss German
    > ones (at least on the layout files supplied with the German edition
    > of Microsoft Windows XP Professional) contain the character U+00A6
    > BROKEN BAR besides the character U+007C VERTICAL LINE.

    And French keyboards typically display a broken bar glyph on the key for
    AltGr+6; note however that this key generates the ASCII U+007C VERTICAL
    LINE, mostly used by developers for programming languages and to provide
    compatibility with ASCII.

    The broken bar character was never mapped directly, the vertical line was
    not perceived as equivalent to the box drawing vertical line, because of its
    unsuitable width and the fact that it does not join vertically, even in
    monspaced fonts. The broken bar or ASCII vertical line are both perceived as
    equivalent, just variants of each others, both glyphs being permitted when
    displaying the ASCII character (but the preference is given for the unbroken
    glyph).

    I don't understand then, why French keyboards really need to display a
    broken bar instead of the vertical line.

    One reason I can see is that the broken glyph makes it visually more
    distinctable from a capital letter I or lowercase L, so the broken glyph is
    a visual hint to help seeing a vertical line symbol rather than a letter (or
    digit 1).

    Due to the separate encoding of the box drawing vertical line (that should
    be centered in a monospaced font, unbroken and joining vertically), I
    suppose that the unbroken vertical line symbol is ASCII is the worst symbol
    invented, and the broken bar glyph should better be used for the ASCII
    symbol U+007C. The fact that it was encoded separately in ISO 8859-1 was a
    result of an a priori incorrect understanding of its need, when it should
    have remained unified with ASCII. So ISO 8859-1 made a decision incompatible
    with the ASCII usage as a symbol for denoting some logical expressions.

    Anyway, the disunification in ISO 8859-1 is the main reason why it remains
    now disunified in ISO 10646 and Unicode. Originately considered as glyph
    variants of the same symbol, they are now given distinct codes, and this has
    changed the preference for displaying the ASCII character from a broken bar
    to a vertical line (this was confirmed when ISO 10646 assigned it the
    normative name "VERTICAL LINE" to U+007C...)

    So one question remains: is it still considered valid to display U+007C as a
    broken vertical line? I think so, because it is useful in some circonstances
    (like on French keyboards, as a visual hint), and because if you really
    don't want to display a broken glyph, there's still a box drawing vertical
    line for your intended purpose.

    This question is quite similar to the use of a visual glyph for denoting the
    space (and for this purpose, several special characters were encoded
    allowing applications to choose their preferred glyph).

    Personnaly, I do think that the encoding of U+00A6 in ISO 8859-1 and ISO/IEC
    10646 and Unicode was not necessary. If we were trying to create it today,
    it would not be encoded, given the now existing font technologies that allow
    selecting variants when needed for some contexts (such as the "visual
    control" mode in editors), instead it would be the default glyph for the
    ASCII symbol (preferred because of its clear distinction from a letter I or
    l or digit 1), and the unbroken vertical line would be just an alternate
    variant.



    This archive was generated by hypermail 2.1.5 : Sat Dec 22 2007 - 10:24:00 CST