Re: [tlug] Freely distributable Japanese capable utf8 font?

["Stephen J. Turnbull" <stephen@example.com>]

Dave M G writes:

 > In any case, is there no such thing as a font that can handle any and 
 > all scripts?

Microsoft Arial does.  There are a couple of others, I believe there's
a free-as-in-beer one by Bitstream.

However, there's a fundamental problem with the unifont approach.
Specifically, Chinese, Koreans, and Japanese strongly disagree about
what glyphs look good for kanji.  Even Europeans; different cultures
have different preferences for exactly how high the accents appear
(thus have I read, I don't have first-hand experience).  There are
also characters that have not been standardized in Unicode (ask any
Ukrainian, although that particular Russian oppression may have been
remedied by now).

 > How does Google manage it? It seems possible to come up with searches in 
 > any language and all scripts seem to display properly.

Google doesn't provide fonts, as far as I know; your system does.

The way it works is that each font may be queried to determine both
the languages it can be used with and the character repertoire it can
handle.  Language usually implies "full repertoire" for the language,
but doesn't always.  The character repertoire is given as a set of
Unicode character codes.

So, when you say "this document is in Times Roman" and you put the
kanji for "ichi" in the document, first Firefox tries to find U+4e00
in the character repertoire of Times Roman, then it looks for a
Japanese font and checks that for U+4e00, and finally it widens the
search to all fonts.  The last option is an algorithmically
constructed "font", the glyph for "ichi" looks like

                +--+
                |4E|
                |00|
                +--+

:-)

There are mechanisms for downloading fonts, AFAIK, but I've never had
need to think about it.  I doubt they're implemented very widely,
especially not on keitai and PDAs.

You can also render the document (or individual characters)
server-side and send them as images.