Description of problem: I have a problem when using kdirstat. I don't run the KDE desktop, and I suspect the problem could be on a lower level in the KDE stack. But I don't know where to point. So I thought I start where I see the problem, and maybe someone can help me get a more direct test case. Out of the box, kdirstat doesn't find any files. I've used strace, and figured out it tries to use a socket under two different file names: /tmp/ksocket-göran/klauncherFSfDRg.slave-socket /tmp/ksocket-göran/klauncherFSfDRg.slave-socket Those having fought with character encodings might recognize "ö" as what you get if you convert the letter "ö" from latin-1 to utf-8. But what kdirstat tries to look up is the above paths with a UTF-8 encoding. I.e. "ö" are the bytes 0303 0266 while "ö" are the bytes 0303 0203 0302 0266. The path that actually exists is /tmp/ksocket-göran/klauncherFSfDRg.slave-socket. But it kdirstat also tries to bind() and connect() to the longer path. If I do a symbolic link frpm ksocket-göran to ksocket-göran in /tmp, kdirstat starts to work normally. So this does indeed seem to be the problem. Maybe having a non-ASCII letter in my login name is to stretch the formal standards a bit. But this looks like a bug in any case. Could this be a kdirstat problem, or is it as I suspect something lower in the stack? In the latter case, would anyone know what a more direct test case would be? Version-Release number of selected component (if applicable): kdirstat-2.5.3-10.fc12.x86_64 How reproducible: Every time
kdistart will be obsoleted by k4dirstat.