PILCH Hartmut
I can still input normally into firefox but not into xterm (or xterm-based mutt) nor into emacs.
Whenever I try to input something there, I get a string of plain 7bit characters, including control characters, e.g. "C$" for "ä", "C6" for "ö", "C<" for "ü", "g" + bell sound for "睡眠", "*c" + line deletion for "かな".
It is now alright. I can input multiple languages, including zh and ja, into xterm and emacs as well as everything else.
I had to remove the scim-bridge package.
Strange because I usually have the scim-bridge package installed and don’t run into that problem. Can you please try if you can reproduce the problem if you install scim-bridge again?
Reports like
scim-bridge経由で問題なく入力できている状態 http://lists.opensuse.org/opensuse-ja/2007-10/msg00046.html
give the impression that some people may not be able to input Japanese without this package. The docs didn't tell me anything about what scim-bridge is good for, and I don't know how this package, which seems to be dangerous for some people and vital for others, found its way into my installation -- I don't remember having selected it.
It mainly solves problems with incompatibilities between different
versions of libstdc++. See section “What is this for?”
in
/usr/share/doc/packages/scim-bridge/doc/developer/introduction.html
or for example comment #18 in
http://bugzilla.novell.com/show_bug.cgi?id=353251
Sometimes it appears to solve input problems in Firefox and OpenOffice
as well, especially when 32bit versions of these applications are used
on a 64bit system.
In the latest updates for the acroread 8.1.2 packages for STABLE/Factory
and openSUSE 10.3, we have deleted the libstdc++ which comes with the
acroread tarball. Then, acroread uses the systemwide libstdc++, i.e. the
same which is used by scim and the compatibility problem with scim
disappears. I tested that acroread 8.1.2 with libstdc++ deleted works
fine with both, the “scim” module and the “scim-bridge” module
(GTK_IM_MODULE=scim or GTK_IM_MODULE=scim-bridge).
Only XIM still doesn’t work in acroread in many locales, but that can
only be solved upstream by Adobe. See also
http://bugzilla.novell.com/show_bug.cgi?id=353251
for the XIM problem in acroread.
--
Mike FABIAN