Pop Quiz:

Does the Linux Desktop Innovate Too Much?
Yet in the middle of all these experiments, nobody seems to be asking a basic question: Does the average user want any of these things?
Personally? I really don't care. But unlike every other pundit out there, I recognize that my use case is highly specialized.

See, in my line of work I can be logged in to any one of literally dozens of systems running almost as many different variations in OS, distribution, release, particular package set, and patch level. When you have to touch this many systems, the problem quickly boils down to: how can I easily make all these differences go away?

So is new shiny going to help me? Not unless new shiny can be trivially (and I mean trivially) added to Solaris 2.6 and Red Hat Enterprise 3.10. And while it may be possible, we know it will never, ever be trivial.

So the way I end up working is this:

I have a Vista laptop. (Justification. Moving on.) From there I putty to as recent a system as I can find. On that system, I install my .screenrc, and run screen. Inside screen, I ssh to each machine to do whatever needs doing.

In the unlikely event that I actually need an X applicaiton of some sort, I have a VMware installation on my laptop, inside which I run a CentOS 5.x VM. And then I ssh with -X or -Y to the target system, and run whatever nasty X application I need to.

With this system, I've almost totally removed the need to know anything about the desktop. By picking CentOS, I've managed to ensure that I'll rarely have to learn anything new, since CentOS' upstream changes extremely slowly.

If anything useful comes up, I'm sure it will dribble its way down to my VM eventually. Years from now. Like maybe when it actually works.

Like I said at the top, this workflow isn't the typical linux user's workflow. But maybe it will give someone some insight into how some of these computers are actually used.