does XIM work in ja_JP.UTF-8 locale?

XIM works find for Japanese when XEmacs is started with LC_CTYPE=ja_JP.eucJP xemacs But when starting XEmacs like LC_CTYPE=ja_JP.UTF-8 xemacs I can switch to Japanese input mode using Shift-Space and enter Japanese, but when committing the Japanese string with RET, it is inserted as mojibake into the XEmacs buffer. I tried (set-coding-priority-list '(utf-8)) (set-coding-category-system 'utf-8 'utf-8)) (set-keyboard-coding-system 'utf-8) (set-terminal-coding-system 'utf-8) (set-default-coding-systems 'utf-8) but none of these settings seems to have any influence on that. When I try: mfabian＠gregory:/tmp$ echo '-*- coding: utf-8 -*-' > test-file mfabian＠gregory:/tmp$ LC_CTYPE=ja_JP.UTF-8 xemacs -q -eval "(require 'un-define) (set-keyboard-coding-system 'utf-8) (set-terminal-coding-system 'utf-8) (set-default-coding-systems 'utf-8) (set-buffer-file-coding-system 'utf-8) (set-coding-priority-list '(utf-8)) (set-coding-category-system 'utf-8 'utf-8)" test-file then enter 漢字 using XIM and switch to hex mode with M-x hexl-mode RET I find: 00000000: e8b2 8def bda2 e89f 84ef bfbd 2a2d 2063 ............*- c 00000010: 6f64 696e 673a 2075 7466 2d38 202d 2a2d oding: utf-8 -*- 00000020: 0a . I don't understand what has been inserted here. If I use M-x set-input-method japanese-canna RET in the last example instead of XIM to insert 漢字 I get: 00000000: e6bc a2e5 ad97 2d2a 2d20 636f 6469 6e67 ......-*- coding 00000010: 3a20 7574 662d 3820 2d2a 2d0a : utf-8 -*-. "e6bc a2e5 ad97" is the correct UTF-8 sequence for 漢字. But what is "e8b2 8def bda2 e89f 84ef bfbd" which is inserted when using XIM? Any idea what happens here? -- Mike Fabian <mfabian(a)suse.de&gt; http://www.suse.de/~mfabian 睡眠不足はいい仕事の敵だ。