Re: i18n and non-latin fonts

[Changwoo Ryu <cwryu debian org>]

Dmitry Mastrukov <dmitry@taurussoft.mtu-net.ru> writes:

> Hi all!
> I've been finding trouble in many application (GNOME and GTK as
> part) periodically. The trouble is in displaying strange (for me of
> course) CE symbols in place of expected russian (sorry, but I speak
> Russian ;) ). The reason is in hardcoded in program body font
> strings like "-adobe-helvetica-*...*-ISO8859-1" (worst) or
> "-adobe-helvetica-*...*-*-*" (little better). The first form is ugly
> and is not discussed. The second form is not sufficient because it's
> not true fact that "adobe" font collection has appropriate
> codeset. For me, "adobe" has not koi8-r codeset. But koi8-r codeset
> has "cronyx".
> IMHO all hardcoded font string should have form
> "-*-helvetica-*...*-*-*" (or courier or others). I find and replace
> font strings with suggested form (and so tired to do this) and this
> work fine almost everywhere. Sometimes there are another trouble. My
> "cronyx" family has not all "adobe" fontsizes. But this is the
> problem of the font collection. The i18n code decide which font to
> use right. I think the form "-*-family-*...*-*-*" is more clear and
> right for i18n.

I try to find and submit a bug for these hardcoded fonts and
gdk_font_load()'s.  But if the program uses the font selection dialog,
it is not so easy.  I think we need a fontset selection widget (we
could steal ideas from the Mozilla or Abiword ones).

And/Or I think we can make a routine for the font selection, very like
the gtk fontguess patch by a Japanese hacker (but I don't like it to
be included into GTK+).  But I don't remember who made it or where it
is...

The patch modifies gdk_font_load(), so that gdk_font_load tries to
load the reasonable fontset even for
gdk_font_load("-adobe-helvetica-...-iso8859-1").

-- 
Changwoo Ryu