sed with eight bit characters in input file

I just remarked that using a sed script on a text file that contains at least one eight bit character does not work in Linux, but does on
Solaris unix.
e.g.: input file mytxt: contains
Is this déjÃ* vu for someone?

mysedcmds: contains : s/ /_/g

sed -f mysedcms mytxt
replaces nothing
but if mytxt contains “deja vu”
then the replacments take place

How to remedy this?

(tried this on openSuSE and RedHat linux)

Works fine here.

$ sed -f mysedcmds mytxt

What version of OpenSUSE? Also do you have your locale set to UTF-8?

version 10.3.
How do I set locale to UFT-8

For 11.0 /etc/sysconfig/language, variable RC_LANG=en_US.UTF-8 for example. Hopefully it’s the same for 10.3.

maybe u should install the utf-8 first and the try it again like ken_yap told it