On top of, yes. Kind of like… let’s say the Google toolbar. It installs into Internet Explorer and enhances it, but without IE, it can’t do anything. Without X, KDE (and Gnome and so on) can’t do anything, but KDE is still a separate thing.
It’s a bit difficult to explain. Probably the best way to look at it is to…well…look at it.
If you do a CTRL+ALT+F1 it should take you to a login prompt on a different console. You can switch back with ALT+F7. From the F1 one, log in and type “xinit – :1” (note, two dashes).
This should give you a bare X, probably running an xterm, and it might by default have twm running. If you click on the background you should see twm If you get nothing you’re not running a window manager. Anyway, if you are in twm and you exit it, you’ll see what X itself is like. No scroll bars, no menus, no nothing really (you can restart twm by typing “twm &” in an xterm).
KDE just builds on this. It provides a window manager and utilities to create a desktop enviroment.
Anyway, you can switch between the blank X and the KDE one with CTRL+ALT+F7 and CTRL+ALT+F8. Exit the blank X session by hitting CTRL+ALT+BACKSPACE and you should be dropped back to the console on F1. Logout and ALT+F7 will take you back to normality.
The X Windowing System (X) is a very low-level software layer over the graphical hardware (your video card, your network*, and whatever else you physically need to get an image on the screen). It only handles the most basic requests (nothing as complex as windows, even) and relies upon the software that sits on top of it to handle everything else. X hands most keystrokes and mouse actions off to the window manager with one exception: Control-Alt-Backspace (yes, a key chord, like Control-Alt-Delete in the MS-Windows world) will kill X dead, dropping you into a text-only mode.
*Real propeller-heads can run graphical programs installed on a distant machine (yes, with full graphics and everything on the machine they’re sitting at) if they can get a network connection and sufficient permissions on the distant machine. (And, of course, X must be running on both systems.) You don’t really need to worry about how this works, just know it’s an option.
So, what defines what you see? Well, the window manager helps. What does a wm do? Well, it manages the windows. It tells X (not the hardware) how to draw windows and the background and everything else the wm wants. If you want your windows to look a certain way (for example, like MS-Windows draws them, or like NeXTStep draws them), you talk to the wm.
Now, you can do real work. Running just X and a wm is perfectly fine from most viewpoints, but people from an MS or MacOS world will want more. That’s where the desktop environment comes in. The DE integrates everything and provides a nice facade over the whole shebang. KDE, for example, provides a standard navigator that doubles as a web browser (kinda like Explorer in MS-Windows, only the KDE version doesn’t suck ;)), some basic functions common to most programs (the right-click context menu), and an integrated look and feel.
Now, what can each of these do alone? Diddly-shit. Without X, nothing can talk to the graphical hardware. Without a wm, X can just barely paint a picture on the screen and a DE is truly up shit creek. Without a DE, you can get things done but you probably won’t feel truly `at home’. So, out of the three, only the DE is truly expendable. You may find that you don’t need a DE, in fact, as I have, but you certainly aren’t obligated to give anything up.
Ah, ok, thanks for all the explanations guys, it’s sinking in.