Re: IDN problem.... :(

From: Adam Twardoch (
Date: Fri Feb 11 2005 - 15:06:16 CST

  • Next message: Adam Twardoch: "Re: IDN problem.... :("

    > Due to the set of avilable fonts for various scripts, it is expected
    > that various fonts will be used to display IRIs. So differences visible
    > in one font may not appear with another font, notably each time a user
    > had to select custom fonts to display pages in his own language.

    Yes, but this is also true for "rn" vs. "m". These are clearly font issues.
    Some characters in some existing fonts look alike and can be confused with
    each other -- most notably, "I", "l" and "1". If you want, you could call it
    a deficiency of the Latin writing system itself. Indeed, the visual
    differentiation between "I" and "l" is not sufficient. Since the two
    characters have different casing, in the past this was not much of an issue
    because readers would usually tell the characters apart based on the
    context. The 21st century poses new challanges since more and more arbitrary
    alphanumerical strings appear in everyday life (e.g. URLs). Humans will
    gradually adapt. Americans and Europeans have been dealing differently (but
    consistently within their areas) with the ambiguities of "1" and "7" in
    handwriting. In future, people using Latin script will need more visual
    clues to differentiate "1" and "l", "0" and "O", "2" and "Z", "5" and "S"
    etc. But this is not an Unicode issue, it's an issue of the visual culture.
    Type designers can and should address that problem, and they have proven
    that they have been increasingly aware of the problem. Ancient UI fonts such
    as Arial don't provide sufficient visual clues to tell the ambiguous
    characters apart but modern UI fonts do. Web browser vendors can of course
    help by choosing reasonable default fonts for their browsers.

    The issue of identical glyphs representing different characters is a fully
    different one. The Cyrillic "a" and the Latin "a" are *supposed* to look

    > So until windows provides to users a way to build custom virtual
    > composite fonts (like Java does in its virtual "sansserif" or
    > "monospace" or "dialoginput" fonts)
    > There's really a lack in Windows (and Mac OS? and Linux?) for those
    > virtual composite fonts, and lacks of tools to design a reliable font
    > stack to display reliably any international text.

    That's not really true. Windows 2000 and Windows XP include two mechanisms
    to provide virtual fonts: font linking and font fallback. They both work
    with Microsoft core fonts. AFAIK, a more comprehensive virtual font
    mechanism will be introduced in Windows Longhorn. Mac OS X also provides a
    font fallback mechanism.


    This archive was generated by hypermail 2.1.5 : Fri Feb 11 2005 - 15:08:04 CST