The X protocol is only really necessary for the opposite case - for allowing powerful computer A to render the display of a weak computer B. If B is powerful, it can render the display itself and send the rendered images, as I was mentioning earlier.

Yeah, no, you have it backwards.

The idea was I sign in to my underpowered workstation and run spice (or whatever) on the more powerful server. I would like the outputs to display on my weak computer. That's the whole point of this, and labs still use it religiously.

Sure, but RDP and VNC can do that just fine, and don't need any special buy-in from the GUI toolkit to do so.

The only technical advantage of doing it the X Server way is if the machine running the program doesn't have the resources/hardware to render.

I know it's not a usual case - which is why I'm saying this doesn't need to be built into the window system at all.

No, they don't! They're not even remotely comparable. Have you actually tried a setup like this?

not really a convincing reason, the interface doesn't have to be rendered by the same process that's running the intensive computation

> doesn't have to be

unless you're volunteering to write the code to make it not, that's how it is