For anybody who cares, I've updated my notes on adapting CLFS trunk to use UTF-8 (this probably breaks things in non-latin1 locales).
They are at http://www.linuxfromscratch.org/~ken/fonts/CLFS/CLFS-UTF-8 (yeah, the fonts directory is probably not the ideal place, but that's where it is). Working for me on all my available platforms (although it's a while since I tested x86_64 multilib). Differences from trunk: add libidn to glibc (with a sed for checkinstall.pl, I think I've worked these out for the arches I don't have), build ncursesw instead of just ncurses (based on lfs, but some differences for how we build and install it), for man recode the supplied messages and pages to UTF-8, add groff-utf8 to be able to read the UTF-8 man pages, a couple of seds for vim to install its UTF-8 pages in the normal directories, and change the console bootscript to match LFS - so, set up /etc/sysconfig/console just like in LFS. The fully UTF-8 system appears to work well - in an xterm (I use urxvt, but gnome probably thrives when fed UTF-8) I can even see the japanese and taiwanese man pages from shadow. For latin and cyrillic alphabets it all seems to work fine in the console (provided your console font can display what you need). I'm not yet ready to suggest this should go into trunk as optional - I want to play with CVS groff instead of using groff-utf8, but I don't have time at the moment (gcc-4.2 and ppc64 needs attention, after I've worked out why my toolchain multilib buildscripts break even with gcc-4.1.2, and I want to look more deeply into kbd-1.12 - now that I've got my own console font, I want to add non-latin1 dead keys to my uk console keymap). But, if anybody wants to debate any of this, feel free. Groff-utf8 can be found in my fonts/other/ directory. ĸen -- das eine Mal als Tragödie, das andere Mal als Farce _______________________________________________ Clfs-dev mailing list [email protected] http://lists.cross-lfs.org/cgi-bin/mailman/listinfo/clfs-dev
