mule-ucs problem

I am using Mule-UCS from 
ftp://ftp.m17n.org/pub/mule/Mule-UCS/test/Mule-UCS-current.tar.gz
and Emacs 21.3 in linux and win2k, ( The input protocol is XIM and win 
IME resp. and keyboard-coding-system is "chinese-iso-8bit") .

I use this package to edit UTF-8 encoded file with Chinese characters. I 
found a strange problem: A few of Chinese characters turned into blank 
blocks (eg. UTF-8 code: 0xE4BDA0 and 0xE6838A ) when loaded from an 
existing file. BUT these characters can be displayed correctly if you 
input it from keyboard unless you close the buffer and reload from a file.

I have no problem in displaying gb2312 encoded file.

I guess mule-ucs has a bug in showing unicode characters or I have not 
correctly configured something like fonts in emacs.

BTW, I heard that next release of emacs  will include CJK-UCS 
translation feature, is it a merge with Mule-UCS?

Thanks a lot,

-- 
Yufeng Lan
http://pub.uibe.edu.cn/~20031
Univ. of International Buz.&Eco.
O: 6449 2311