Windows Anonymous

I've mentioned it before, in these pages and elsewhere. There is a movement to port Linux apps, an entire desktop environment, and, indeed, the kernel itself, to Windows. Slowly. Insidiously.

See, Windows sucks. There's no opinion involved here. It's slow, inefficient, insecure, unreliable, offers little-to-no flexibility, tries (and fails) to be everything to everyone, and to beat all you're expected to pay for it. It is proof of what happens when you try to make something idiotproof. Over in X11-land, however, we have some of the most brilliant programmers in the world, doing what they do best, often for free. For a long time, the goal was to get 3rd-party Windows apps ported over to Linux/BSD/Mac and everyone began lobbying software vendors thusly. Once developers realized that you get better results by yelling at a brick wall, they went about making replacements. Not surprisingly (to Linux users anyway), the replacements turned out superior to their Windows counterparts. Then a funny thing happened: In an odd reversal of fortune, Windows users wanted what Linux users had in terms of applications; they wanted the simplicity, functionality, and security that Linux users have had all along because Linux users were smart enough to use Linux in the first place. Windows users got all excited when they realized that the very same concept of open-source which benefits us could theoretically benefit them the same way; at the same time, the developers of some of these apps saw an opportunity to preach the open-source gospel in the hope of perhaps getting some Windows users to switch full-time. What still hasn't sunken in with any of them, however, is that's not how it works.

Exhibit A: A company called MobaTek is hawking a couple of apps meant to bring Linux-like functionality to Windows. We have MobaSSH--which, let's face it, is almost pointless since most Windows users won't go near an SSH client, much less a server--and MobaXterm, which is "an enhanced terminal with an X server and a set of Unix commands (GNU/Cygwin) packaged in a single portable exe file." Well, that it is; in addition there are plugins to give is more functionality such as GCC, G++, Xnext, Vim, Perl, and so on (sorry, no Python as of yet). It is a portable .exe but can be installed (and should be in order to use the available plugins). In common usage it works like Cygwin on steroids, not in itself a bad thing. It can start an OpenBox session right on your Windows desktop. It's even capable of accessing a remote X-server with Xdmcp. Does it work? Of course it does, it's *nix. All this comes at a price, however: Windows, being inefficient, doesn't give up resources easily, and if you get creative enough this whole thing will slow your machine to a crawl; this particular road-test was done on a Windows 7 Professional install with an Intel Core 2 E6600, 2GB DDR2-800 and 128MB VRAM, not shared, so I wasn't exactly holding back. I'm not saying it's impossible to run this on anything less than the latest-and-greatest, but if you think of this as a way to skirt the hardware requirements of conventional virtualization, you'd be mistaken.

The next example, KDE for Windows, might be familiar to some of you; it's KDE's attempt to bring KDE to Windows. Evidently their previous attempt to do things the other way around (by making KDE4 look like Windows) wasn't working out. Anyway, KDE has gotten somewhat bloated and resembles Vista, so this might be a good match. You get two install options: End-user and package manager. End-user is for the typical Windows user who doesn't know anything outside Microshaft's walled garden and likes not having to think for very long; package manager mode is for someone who can at least handle Synaptic in Ubuntu without suffering an aneurysm. There are 3 compiler options, and packages are binary-incompatible so if it was compiled with one, it's invisible to the other two; for this test I went with MinGW-4. I installed the typical setup needed for an average KDE install such as kdelibs, kde-workspace, kdebase, and let the installer pull the requisite freight-train of dependencies; in this case it came to a total of 52 packages.

I'm here to tell you, it's downright eerie seeing Konqueror open in Windows 7, with my home directory shown as "C:/Users/chris"...forward slashes on a Windows box. Just. Eerie. Keep in mind that this just gives you the KDE apps; it doesn't reinvent Windows as a KDE-themeable environment. At least I don't think it does; KDE's Settings manager threw an error about no views defined and no one is answering the bug report at KDE's website. You can always install Stardock; with all this other jazz installed and running you'll bring the whole deal to a screeching halt, but it'll look pretty.

Anyway, so now we have both an xterm, GCC, and KDE on Windows. But what if we could have the actual Linux kernel installed? Well, if you're a true masochist, you can. It's called coLinux, and it aims to put a Linux kernel inside Windows. In its current incarnation, the installer lets you download "images" of various Linux installs, all hopelessly out of date: Fedora 10? ArchLinux-2007? Ubuntu-7.10? Fedora 10 isn't even supported any more. Once installed, everything is done command-line style--the one thing Windows users hate--complete with passing params and editing .conf files...you know, like in Linux. Messing with this on a Windows guest in VirtualBox will require specific settings or it'll crash. The tagline at the top of coLinux's homepage says "If Linux runs on every architecture, why should another operating system be in its way?" Indeed. The rational response to this would be to simply replace Windows with Linux and that would be that. However, this project doesn't answer the question; Windows is still in the way. Rather, it's the answer to a question no one asked, namely, if you have the resources to run 2 operating systems side-by-side, why aren't you just using VirtualBox? Hell, even Wubi will at least give you the latest version of Ubuntu, and the rest of the distros available for coLinux are available as a liveCD or liveUSB; on top of that, they'll also be current, stable, supported versions.

The concept being employed in all of these projects is that of putting a Honda Civic on the back of a truck in the hopes of improving the fuel-efficiency of the truck: It's retarded logic, and it doesn't work. If you want to use Linux, then by all means use Linux. If you want everything to "just work", use *buntu; if you want to really learn what goes on with your machine, use Arch or Gentoo. You have liveCDs, liveUSBs, virtual machines. But if you put a KDE logo on your Windows install, it's still Windows. If you use any of these apps expecting to get all of the benefits of using Linux, you're in for a rude awakening, and this is where I have a problem: Someone (or many someones) will try one (or more) of these expecting the security and stability they've heard about Linux, gain neither, possibly end up with a hosed Windows install, and blame Linux.

Tell me it can't happen...I really want to believe it, but I know better.

To the developers of these apps, I have 4 words for you...silk purse, sow's ear. Especially when that sow is Windows.