SuSE 9.0: No Japanese characters with canna 3.6 in KDE 3.1.4
Is it okay to write in English (or German) in this list? Most of you will find my English easier to read than my Japanese ;-) . 返事は日本語でも結構 です。 I have updated a computer from SuSE 8.1 to 9.1 (new installation), and since that time I cannot write Japanese characters in KDE-based programs. Japanese input is okay in Openoffice-Writer 1.1, Opera 7.50, and gjiten, but when I try to write Japanese in KMail, Kate, or a KDE console, the following happens: When I press the Kanji key, the conversion window opens. When I type, katakana appears. When I press <space>, katakana is converted to kanji. So far fine. But when I press <enter>, the Japanese characters disappear, they are not inserted into the text. This is not a font problem. I can see Japanese characters in KMail, and I can paste Japanese characters into Kate and into a console. "cannastat -v" shows that USED_CX is increased if I convert Japanese characters in Opera etc., but it is not increased if I try to convert Japanese characters in Kate, KMail etc. I have LANG, LC_CTYPE and LC_ALL set to ja_JP in the YaST system environment, in ./profile and also in /etc/X11/xim (just to make sure). I tried the same with ja_JP.UTF-8, it doesn't work either. /etc/hosts.canna had "localhost" and "unix" entries only. I added "<MyHost>:root,herrmann" -> no improvement. "ps -eF" shows: ... /usr/sbin/cannaserver -u wnn -r /var/lib/canna ... kinput2 -xim -kinput -canna This looks okay to me, though I am not sure about "wnn". (?) I am using SuSE 9.0 with a recent update, canna 3.6p3-127 kinput 3.1-184 kdebase3 3.1.4-52 The computer is a 2.3-GHz pentium 4 desktop. I have the same system (Suse 9.0 with the same packages, the locale partly in English and German, same messages from "ps -eF") running on a 600-MHz notebook with no problem. I have almost the same system (SuSE 9.0 without the updates) running on an old 100-MHz pentium desktop, also without problem (well, it's slow ... ;-) ). As long as SuSE 8.1 was running on the new machine, there was also no problem there. Does anyone know or guess why my Japanese input is not working on that computer with those programs? Or what can I check, and what can I try? Hopeful, Michael
Dear *.* I have finally solved my problem with Japanese character input, reported here 2 weeks ago. Since the same problem may happen to other users, I will shortly describe the reason and the solution. For various reasons I had reconfigured my keyboard layout using (among others) the following commands in ~/.Xmodmap: clear mod1 clear mod3 clear mod4 keycode 64 = Alt_L Meta_L keycode 117 = Mode_switch Meta_R add mod1 = Meta_L add mod1 = Meta_R add mod3 = Alt_L add mod4 = Mode_switch In SuSE Linux 8.1 these commands lead to the following modifier bindings (see "xmodmap -pm") which looked and worked well: mod1 Meta_L (0x40) Meta_R (0x75) mod2 Num_Lock (0x4d) mod3 Alt_L (0x40) mod4 Mode_switch (0x75) mod5 Scroll_Lock (0x4e) But in SuSE Linux 9.0, the "add" command was interpreted in a different way, and the modifier bindings became: mod1 Alt_L (0x40) Mode_switch (0x75) mod2 Num_Lock (0x4d) mod3 Alt_L (0x40) mod4 Mode_switch (0x75) mod5 Scroll_Lock (0x4e) Obviously, this does not look correct any more because the Alt_L and Mode_Switch functions operate several modifier bits each and because the mod1 bit is operated by two different functions. Through some channels which I do not understand, this lead to a loss of information when characters were transferred from kinput/canna to KDE-based programs. Since I have deleted the Meta_L/R entries from my keycode lines and deleted the "add mod1" lines, Japanese input is working correctly. Michael Herrmann
participants (1)
-
Herrmann.KARC