Re: Deseret Unicode

From: Erik van der Poel (erik@vanderpoel.org)
Date: Tue Apr 26 2005 - 10:56:42 CST

  • Next message: Erik van der Poel: "Re: Deseret Unicode"

    Jukka K. Korpela wrote:
    > On Mon, 25 Apr 2005, Curtis Clark wrote:
    >>Actually, it is user-selectable, and at least in the case of Firefox, on
    >>somewhat of a per-language basis.
    >
    > On Opera, too. But there are two kinds of settings, unless I've missed
    > something:
    > a) mapping serif, sans-serif etc. to specific fonts
    > b) selecting the font to be used when a page makes no font suggestion
    > (even using a generic font name) or when the browser has been set
    > to ignore such suggestions.
    > The latter is possible on IE, too, and the default font can be set
    > differently for different writing systems. Presumably, this means that
    > when processing text with no font suggestions, the browser determines
    > which writing system each character belongs to and selects a font
    > accordingly. Things are probably rather confusing and confused when
    > we get into details. For example, the second kind of settings might
    > look like being for _language_ dependent settings; and to add to the
    > confusion, it probably partly _is_ dependent on the declared language.
    >
    > I've noticed that if I set e.g. lang="ru" for a piece of text containing
    > Russian as transliterated (in Latin letters), browsers may think (from the
    > lang attribute) that the text must be in Cyrillic letters, so they use
    > the font selected for Cyrillic script. This is somewhat understandable,
    > since browsers cannot really recognize the writing system - they need
    > to make a guess.

    It does get a little confusing when you get into the details. I don't
    know exactly how Opera does it, so I won't comment on that, but I did
    the initial implementations for the open source Windows and Linux/Unix
    versions of Mozilla, so I'll say a little about that.

    The CSS generics (serif, sans-serif, etc) are set on a
    per-language-group basis. One of the groups is called Western, and this
    is selected e.g. when no language has been specified and the charset has
    been set to a Western one like iso-8859-1 via HTTP, HTML or browser
    menu. So the language group concept is based on charsets, and this was
    done because the early browsers set fonts on a per-charset basis, and a
    lot of the Web at that point had come to rely on that.

    Of course, when the language _is_ specified, Mozilla uses that to select
    a font, by first mapping it to a language group, and so on. A while ago,
    I wrote up a more detailed description for this mailing list:

    http://www.unicode.org/mail-arch/unicode-ml/Archives-Old/UML025/0107.html

    The approach and early intentions are written up here:

    http://www.mozilla.org/projects/intl/fonts.html

    Another bridge that we had to build from the old font preference world
    to the new was from (proportional vs fixed width) to (serif vs
    sans-serif vs monospace). This one was addressed by another UI element
    to choose serif or sans-serif for proportional text. (Fixed width _is_
    monospace, of course.)

    Erik



    This archive was generated by hypermail 2.1.5 : Tue Apr 26 2005 - 10:57:24 CST