I've heard about such crashes related to scim as well. The original reporter mentioned #342198 [1] in scim, which is a bug caused by the libstdc++ allocator transition. But as it's reported in testing and scim 1.0.2-3, it looks to me more related to #328421/#323216 [2,3], which is about the libstdc++ SONAME transition.
In both cases, though, the source of the problem should be the same, i.e., libstdc++ has unversioned weak symbols. This is known both by upstream and the Debian GCC maintainers, and seems won't be fixed upstream until at earliest GCC 4.2. This is also mentioned by SDiZ Cheng in #343060 [4]. scim 1.4.2-1 and friends should be entering testing soon, so this problem will be covered up. As the scim maintainer, I agree with Matthias Klose that adding Conflicts in libstdc++6 is not a very good idea. As the workaround of setting GTK_IM_MODULE="xim" exists, it probably should just be well documented. 1. http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=342198 2. http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=328421 3. http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=323216 4. http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=343060;msg=37 Just my two cents, Ming 2005.12.27 -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
