I would argue that the biggest stumbling block to Linux going mainstream is actually X. Get rid of it, start something sensible from scratch that's based on current (i.e. more developed) ideas than what was leading edge 25 years ago. Seriously.
Case in point: trying to configure two screens on Linux is anything between painful to downright impossible. In 2009! Windows 95 had that ability.
As the article points out, the X you use today is the ninth revision and a total redesign at that. They thought about the problem more than you are capable of giving credit.
Also: "Linux going mainstream?" First: Who cares? Your case is also incorrect (or based off of experience with the wrong software) since I can drag and drop the dual monitors in the display configuration screen today. I was able to make this work with a few text file edits in 1998. It's been quite painless for a while and quite possible for even longer.
Try two screens that don't form a rectangular area (e.g. one large and one small screen) and still get a decent setup, i.e. one where you can see all of your desktop.
My biggest problem with your argument is that you haven't presented any reason to get rid of X whatsoever. You've pointed out that somewhere in X.org or supporting drivers there are issues, but that's not a problem with X11.
If you were to say "we need to write a new X server from scratch" I would be somewhat with you, at least your point supports that. And if you were to say "X.org needs more bug fixing and better drivers" I'd wholeheartedly agree.
But throwing out X11 alltogether is just... stupid, at least for the reasons you mentioned.
Most people don't know there's a difference between X.org (the implementation) and X11 (the protocol). I guess that's why people conclude that "X" sucks and must be replaced.
X is fine in my opinion. It's flexible and a known commodity with pretty good hardware support. The cost (in man hours) of replacing X and modifying all the associated bits that rely on it far out weighs NVIDIA or ATI getting someone to spend a few hours developing a utility to setup multiple monitors. A lot of the things people blame on X really have very little or nothing to do with it.
Strange, for me, it's very easy, which doesn't seem to fall in your range. What's wrong?
Throw out working solution after 25 years of development, with all the codebase and everything, that makes perfect sense, of course. You are arguing with one particular case, which is most likely just an implementation shortcoming of your drivers (because it works for me, and well :-P), that doesn't justify throwing out thing like this. X was extended so many times, old stuff was replaced with new extensions, and eventually, it should be scrapped because someone can't configure two screens with it! Oh.
Can you name one thing that is fundamentally wrong with X and cannot be simply replaced with a new X extension, or fixed in the implementation?
Issues with advanced configurations such as dual-screens are generally the fault of drivers, not X. For example, I can configure two screens easily using the open-source Intel drivers, but proprietary nVidia drivers do not support the standard.
So, what exactly is outdated about the ideas of X? For example, I noticed that my old law firm paid lots and lots of money for a software solution that does on windows what X does natively.
I think now that we are starting to run software from remote computers again, the ideas of X are becoming more and more relevant and are being essentially repeated at higher levels for higher cost.
NeWS was technically superior to X in many ways. I'd argue with Chrome could effectively be NeWS with Javascript instead of Postscript, disguised as a browser so as not to wig everyone out...
I actually wrote a significant program in NeWS. I ran screaming back to X. PostScript is a good page description language. It's a horrifyingly bad user interface implementation language.
X the standard and Xorg the implementation have been one and the same thing for a while.
But X is just a component for the system, and what is lacking is a vision for the system as a whole. Actually, unifying visions have emerged for many facets of the system (as proven by the success on the server and embedded space), just not for the desktop.
I think he's talking the about linux adoption not being down to X but there being no long term vision for linux as a whole. Perhaps there are not enough designers and artists on board to make the linux desktop woo business. It's not always decent APIs that win.
> Case in point: trying to configure two screens on Linux is anything between painful to downright impossible. In 2009!
I used various flavors of linux all through college and I gave Ubuntu a fair shake for the past two years on my laptop, but this is the main reason why I'm using Vista today. That and shutting down took minutes. I don't have time to dick around with text files fixing this just to have it break again on the next upgrade.
Interesting read. The part where they started to break components in the code to see if people were still using them made me raise my eyebrows a little.
From a comment: 3) Set pallette[sic] (port 0x3C8 & 0x3C9)
I grew quite nostalgic when I read that line. I have fond memories of tweaking the VGA palette back in my DOS days. Palette manipulation made it easy to do some kinds of special effects that would take a lot of GPU power today. There was actually an experimental version of tvtime that would wait on port 0x3DA for vertical sync (to work around the lack of sufficient vsync support in X). I tried to convince vektor to keep it in, because that's the best my TV had ever looked, anywhere, on any device.
Oops, I didn’t even notice that! I’m not a subscriber myself, was linked to this page from Tim Bray’s twitter feed (http://twitter.com/timbray). Regardless, it seems to be sanctioned by the site themselves, if their big advert at the top of the page is any indication.
Corbet (lwn editor) posts them here from time to time. And he posted in this thread as well, so he knows about it, and presumably would have said something if he was annoyed.
I bet they get way more exposure (and thus subscriptions) from people widely distributing those links than they lose...
Case in point: trying to configure two screens on Linux is anything between painful to downright impossible. In 2009! Windows 95 had that ability.