ma, 2006-07-31 kello 06:32 +0200, Marko Macek kirjoitti: > Eduard Bloch wrote: > > #include <hallo.h> > > * Marco Macek [Sun, Jul 30 2006, 05:24:42PM]: > > > >> Martin-Éric Racine wrote: > >> > >>> fonts.conf is attached. > >>> > >> Check if Xft is being used with ldd /usr/bin/icewm: > >> > >> libXft should appear in the list. > >> > >> BTW: where do you have the font installed? > >> Did you run fc-cache? > >> How does the font appear in fc-list? > > > > Marco, I have a similar problem with Cyrillic chars. The Debian standard > > build uses Xft and imlib, lite build uses X11 fonts and libxpm, see below. > > The problem is reproduced easily with: > > > > xtermset -n асдф-Fön -T асдф-Fön > > > > There you see only junk (like not decoded multibyte) instead of non-ascii > > chars. My fc-list output looks well, I have the default set of Windows-XP > > fonts > > integrated "as usual", see attachement. > > > > Choice of theme does not make any difference. Some time ago I had > > success using Verdana, but this does not work anymore. I tried: > > > > TitleFontNameXft="Verdana:size=12" > > > > in prefoverride. The font changes, the breakage is still there. > > I have tested Verdana with Cyrillic, see attached screenshot. Titlebar > is OK (font has cyrillic), the taskbar is not > (default font). > > Which X version (mine is X.Org version: 7.0.0)
Debian Testing currently has X.org 7.0. Anyhow, I finally got Unicode to work. The solution seems two-fold: 1) IceWM needs to use a default font with a more thorough Unicode support, because the generic sans-serif might end up calling a really spartan font, on some systems. 2) Debian has to fix the way it handles locales. Basically, if LANG is not exported by the initialisation scripts of whichever shell launches X, they IceWM will never know it's supposed to be running in a non-ASCII locale and it then flatly refuses to show 8-bit characters (even though applications running inside it might have no problem whatsoever displaying a multitude of scripts simultaneously). Ditto if, when configuring locales (or belocs-locales-bin), the user decides to NOT set a default system locale. I think that, from now on, absolutely everything in the POSIX universe ought to assume that the C locale really means C.UTF-8 and not C.ASCII. This would simplify an awful lot of non-Bushistanese people's life. If this requires updating some obscure RFC which states that C means ASCII then, by $deity, let's update this ASAP. -- Martin-Éric Racine http://q-funk.iki.fi
