During the FC4 cycle, there have been quite a few updates to xterm. I never use this, so set about removing it as it's a waste of time/space/CPU to keep updating it. However, "rpm -q --whatrequires" says that xinitrc depends upon it, which means that you can't have X without xterm. Looking at the code /etc/X11/xinit/xinitrc, it seems that this dependency is only needed for if /etc/X11/xinit/Xclients does not exist. But seeing as this file is also part of xinitrc, that can't happen. The other file it is referenced from is /etc/X11/xinit/Xclients itself, but in this case it is only executed if it exists (if [ -x /usr/bin/xterm ] ; then) - i.e. there is explicit allowance for it not existing. Hence I conclude that xterm is not a real dependency of xinitrc, and the dependency should be removed. It only gets called of xinitrc has itself been damaged.
I forgot to mention. xterm is only called from /etc/X11/xinit/Xclients if you don't have GNOME or KDE. If you're that kind of user, you probably know what you're doing, and will install xterm manually.
The ability to start an xterm is required for any failsafe / rescue mode session, or for being able to start X without either GNOME or KDE installed . I think the default xinitrc Xclients SHOULD contain xterm, and hence must have a dependency on the xterm package. If the default Xclients file did not contain xterm, what terminal would you suggest that it invoke ? We don't want to replace the xterm dependancy with one on GNOME gnome-terminal / KDE konsole. It would be impossible to do anything with an XSession that did not use any window manager if there were no terminal invoked, and hence Xclients must have a dependancy on a terminal program - this has always been xterm. If we removed the xterm dependency, users could end up with a failsafe / rescue mode session that did nothing and could not be exited.
"The ability to start an xterm is required for any failsafe / rescue mode session, or for being able to start X without either GNOME or KDE installed." I'd say that the first case here, "failsafe/rescue", is something 99.9% of users do in text-only mode (at the console, or via SSH), not in X. The second case, "for being able to start X without either GNOME or KDE installed" isn't a true dependency - it's an option. If X _must_ have a terminal, then the user has the option of GNOME, KDE, or a barebones situation. If they choose to not have GNOME or KDE, then fine - give them xterm. But this is by no means a real "dependency". I say, let the user choose. It's not a real dependency.
RE: > If they choose to not have GNOME or KDE, then fine - give them > xterm. But this is by no means a real "dependency". > I say, let the user choose. It's not a real dependency. Yes, but if the xinitrc package did not specify a dependency on xterm, it would be unable to contain an xterm invocation in any of its scripts, so the user would be unable to choose to run X without GNOME or KDE . > If X _must_ have a terminal, then the user has the option of GNOME, KDE, > or a barebones situation. xterm IS the "barebones situation" in this case; without it, users would not have the option of running X without GNOME or KDE. Sorry, but as xterm has always been part of every X Windows distribution, xinit must be able to depend on it - this is NOTABUG.