Hello, Just installed mule-enabled 21.5.9 on Solaris 8 and noticed that no longer can input japanese via xim. Instead of valid characters the garbage is input, that seems to be raw euc-jp characters as set my locale. The comment in event-Xt.c suggest using composed-input-coding-system variable to determine coding system for convertion. After initializing this variable to coding system for current locale in startup file, everything seems to work perfectly. No side efects noticed yet. Not sure how conceptual this change is, anyway patch follows. (setq composed-input-coding-system (get-coding-system-from-locale (current-locale))) Beta 21.5.9 Solaris 8、euc-jp ですが　XIMで　日本語入力して文 字化けになりました。"composed-input-coding-system"のヴァリアブ ルでXIMの出力を変えて使えます。 93a94,95 1065c1067 < (XLSTREAM (fb_instream), Qundecided, CODING_DECODE, 0); --- 3555a3558,3563