[Fontconfig] Why do asian fonts contain greek glyphs?

James Cloos cloos at jhcloos.com
Thu Sep 29 03:41:43 EST 2005

>>>>> "Simos" == Simos Xenitellis <simos74 at gmx.net> writes:

Simos> Why does Greek show like that? The characters are always
Simos> slanted, big space between them, and in general they are
Simos> terrible.

I'm not completely sure about the historical reason, but the
pre-unicode CJK character set standards all included greek
characters.  I suspect this was for rendering math, given
how the glyphs are typically shaped.  Cyrillic glyphs are
also included in the sets and could also be an issue.

Emacs has a similar problem with greek and cyrillic text pasted
from other X apps, due in part to the fact the the original i18n
extension for emacs, MULE, was written by and for Japanese users.

Simos> a. How can we instruct fontconfig (fonts.conf) to
Simos> ignore any Greek glyphs found in those Asian fonts?

The locale should affect font selection when looking for backup
glyphs.  It should help to ensure that a font which matches the
user's primary desired locale is the top preference for any given
font alias, such as serif, sans or monospace.  Beyond the locale,
the apps can tag the text with a language.

Simos> b. Is ignoring the Greek glyphs in those Asian
Simos> fonts the correct path in solving the problem?

It would probably be a good idea to fall back to a CJK font only
in a CJK locale or for text marked as C, J, K or V.  How best to
accomplish that is an /interesting/ question....

Ignoring only fonts that include a specific set of language matches
(such as the regex "^(zh|ja|ko)(-..)?$") would be enough for this
specific issue, but seems a bit kludgy, yes?  

James H. Cloos, Jr. <cloos at jhcloos.com>

More information about the Fontconfig mailing list