3 * http://wiki.laptop.org/go/Ethiopian_Setup
4 * http://wiki.laptop.org/go/Amharic_language
8 To enable the Amharic Ethiopian locale, the /etc/sysconfig/i18n
15 To alleviate boot overhead with UTF-8 locales, we'd like to
16 move this setting to a later point in the boot process.
17 olpc-dm would be a good candidate.
19 * TODO: I'm going to make some tests and report back.
20 * DONE: LANG=C in /etc/sysconfig/i18n, LANG=am_ET.UTF-8 later on
24 F7 glibc does not even boot when am_ET is enabled, but it's not
25 obvious why. Suffice to say that the output of ls comes out
26 sorted by strlen() when am_ET is enabled!!!
28 glibc-2.6 from F8 is ok, but I can't find an obvious patch to
29 backport. The only recent addition to the am_ET locale has
30 been to discard accents in string comparisons.
32 Moreover, the glibc-2.6 rpm post-install scriptlet fails silently
33 on jffs2, corrupting the locale archive. It happens because
34 the build-locale-archive tool tries to use a shared writable
35 file map which is not supported.
37 The glibc-2.6 rpm is available from OlpcBernieRepo.
39 * DONE: Of course we don't want to upgrade to 2.6 at this time, so I'm asking the glibc hackers for insight.
40 * DONE: http://sources.redhat.com/bugzilla/show_bug.cgi?id=5033
44 A good candidate that provides high-quality Ethiopian glyphs is the
45 Abyssinica-SIL.ttf, which is not packaged in Fedora.
47 I grabbed the TTF file from the Debian package ttf-sil-abyssinica-1.0
48 and temporarily installed it in /home/olpc/.fonts/
50 * DONE: abyssinica-fonts RPM package is in Fedora, and tagged for OLPC-2.
52 == ET keyboard layout ==
54 The updated xkeyboard-config RPM shipped with latest builds
55 already contains the "et" keyboard layout.
57 This can set in xorg.conf, or loaded for testing like so:
60 setxkbmap -v -model olpc et
63 The "=" key is not working in current builds, but Sergey
64 has already fixed it upstream and we'll pick it up with
67 * DONE: Send a new olpc patch upstream to resync
68 * DONE: I'll update our RPM from a CVS snapshot of xkeyboard-config
69 * DONE: https://bugs.freedesktop.org/show_bug.cgi?id=12202
73 Sergey's Compose file for am_ET is already upstream, and it
74 is required for XIM-baded composition to work in all applications.
76 I prepared an updated RPM and staged in OlpcBernieRepo.
78 Even with this new RPM installed, XIM does not appear to work in
79 Abiword and other utilities I tried.
81 Furthermore, the Compose works by pressing a vowel, followed
82 by a consonant, which seems to be a less convenient way
83 of producing glyphs (see below)
85 * DONE: Sergey said he'll check if CONSONANT+VOWEL is possible with XIM.
90 GTK contains an Amharic input method which is currently outdated.
91 I'm in contact with the author, Daniel Yacob, who'll soon port
92 his latest patch to the current version of GTK.
93 This probably means we'll have to fork the gtk2 package too.
94 I'm unable to tell how important these changes would be for users.
96 Lidet Tilahun, a professor of Ethiopian at Harvard, tried typing
97 on the XO with this IM and confirmed that this would be the
98 preferred method. It's basically CONSONANT+VOWEL, with a way
99 to switch in a CONSONANT-only mode.
101 Additionally, it seems this IM requires the "us" keyboard.
102 I couldn't get it to work with the "et" keyboard loaded.
103 Daniel says it shouldn't happen.
105 * TODO: waiting for a patch. Progress is being tracked here:
106 http://bugzilla.gnome.org/show_bug.cgi?id=101360
110 We couldn't get glyph composing to give correct visual
113 On F7, Abiword works fine, but it's an old version that still
114 used Xft directly. The version we use now switched to pango
115 and apparently regressed on this.
117 I asked uwog about it and he suggested me to file a bug
118 upstream. Without this fix, the IM is still usable, but
121 * FIXME: bug filed upstream: http://bugzilla.abisource.com/show_bug.cgi?id=11190
122 * DONE: pinged the above bug...
126 == Other Gtk applications ==
128 I tried installing gedit on an XO, and I couldn't get neither the
129 Amharic, nor the XIM input methods to work with it.
133 * TODO: Needs investigation.
138 System stability appears to be somewhat compromised when all
139 the above changes are enabled.
141 I've seen activities crashing randomly and funny rendering problems.
143 * TODO: Before we even start filing bugs on this, I'd prefer to
144 see some of the above components land in the builds, otherwise
145 we might be chasing bugs that are even harder to reproduce.