KDE issues

[Sergey]

Hello!

Has anyone experienced segmentation faults with kicker (kde panel)? It only 
happens when I quit KDE. Also sending SIGTERM or stopping kicker with dcop 
call results in the same - kcrash is invoked. If I just start kicker and send 
SIGTERM to it - fine. If I start kicker and browse the K menu for a few 
seconds, and then terminate, it fails to shutdown cleanly... I run updated 
Fedora Core 4, and I thought it might be an issue with updates (I have 
upgraded to KDE 3.4.3 from kde-redhat project). However after I install 
clean, original FC4 on a different disk - everything is the same - kicker 
crashing...

Perhaps someone helps...

Anyway thanks. I solved a lot of KDE bugs, these are my latest researches :)

**** KDE root-level apps failure ****
If you have noticed, running root-level priveledge applications from KDE (via 
kdesu, userhelper and so on) is buggy, results in X server connection 
rejects, or doesn't work at all. Try running KDE control center, then going 
to, for example, System Fonts and choosing administration level (root). After 
being prompted for password, you might see that you haven't achieved the root 
menus, but have been put back to user menus with no root options.

This happens because /tmp/kde-$USER/ksycoca symlinks have been removed by 
rc.sysinit script on system (re)boot. And if previously KDE system 
configuration cache for root had been created, the symlink is NOT restored, 
but is essential for apps to run! At least 
removing /var/tmp/kdecache-root/ksycoca will make the kbuildsycoca maek the 
links again. Running kbulidsycoca --noincremental should also help. However 
there is more elegant solution without modifications of any RPM-produced 
files. Put the following to your /etc/rc.d/rc.local startup script.

----------------------------------------------
for i in /var/tmp/kdecache-*; do
 a=${i#*-}
 [ -e /tmp/kde-$a/ksycoca ] && continue
 [ -e $i/ksycoca ] || continue
 grep -q ^$a: /etc/passwd || continue

 install -d -m 0700 -o `id -u $a` -g `id -g $a` /tmp/kde-$a
 ln -s $i/ksycoca /tmp/kde-$a/ksycoca
done
----------------------------------------------

Alternatively just alter the rc.sysinit script so that it won't touch 
important stuff. 

**** KDE 3.4.3 (from kde-redhat project) kppp bug ****
If you have problem as I used to have - KPPP fails to update /etc/resolv.conf 
after a connection has been established, here is the solution. Make a 
script in /root/bin/kppp-compat.sh containing:

----------------------------------------------
PATH=/usr/bin:/bin

REALDEVICE=$1
LOGDEVICE=$6

umask 022

case "`basename $0`" in

ip-up.local)

 if [ "$REALDEVICE" = "ppp0" -a -z "$LOGDEVICE" ]; then
  grep -v "KPPP-compat" /etc/resolv.conf >/etc/resolv.conf-
  (for i in $DNS1 $DNS2; do
    fgrep -q $i /etc/resolv.conf || echo "nameserver $i # KPPP-compat"
   done
   cat /etc/resolv.conf-) >/etc/resolv.conf
  [ -f /var/lock/subsys/nscd ] && /usr/sbin/nscd -i hosts
 fi

 ;;

ip-down.local)

 if [ "$REALDEVICE" = "ppp0" -a -z "$LOGDEVICE"  ]; then
  [ -f /etc/resolv.conf- ] && mv -f /etc/resolv.conf- /etc/resolv.conf
  [ -f /var/lock/subsys/nscd ] && /usr/sbin/nscd -i hosts
 fi

 ;;

*)

 echo "What are you doing?!"
 exit

esac
----------------------------------------------

The scripts flushes the NSCD cache as well, as KPPP fails to do it. Now make 
symlinks to this script from /etc/ppp/ip-up.local and /etc/ppp/ip-down.local.


Good on ya!