24 comments

[ 3.3 ms ] story [ 57.3 ms ] thread
The style of this article is all wrong. It takes a tremendously long time for the author to get to the point.

There is an irrelevant history lesson of an experience jumping through distros to start with. Then the main argument of the piece is unclear. It begins with (probably justified) moaning about device drivers, but is followed up with a declaration that Linux's problems are caused by recent UI changes. The author then laments that the update cycle is too fast.

Why hasn't someone told this person to stick to long term stable distros, such as RHEL or Ubuntu LTS?

I found the comment from Jonathan Corbet of LWN (https://lwn.net/Articles/489689/) particularly telling: 'The article is a little muddled, complaining about the "we know best" attitude while saying that Linux lacks the integration seen in iOS or Android, but it's worth a look.'

That does seem like a common double standard that people hold Linux to: on the one hand people complain when it lacks coherent vision and whole-system thinking, and on the other hand people complain when it has coherent vision and whole-system thinking.

I think "has a coherent vision and whole-system thinking" is not even close to a fair description of what the author characterizes as a "'we know best' attitude." Having a coherent design does not mean you can't accommodate a wide range of needs, and similarly, lack of flexibility does not indicate a coherent design.
Having a coherent design requires making decisions, and sometimes those decisions require ignoring niche preferences. http://islinuxaboutchoice.com/
Yes, a thoughtful design might ignore some use case or another — so too might a slipshod design. The amount of care put into a design and the degree of freedom it offers are largely orthogonal issues. The point is, a desire for a coherent and intelligent system design is not irreconcilable with a desire for flexibility.
(comment deleted)
The real reason Linux lags windows is application support. Windows 8 seems to have every intention of being a train wreck, but Linux will benefit little because users will not upgrade from Windows 7. Linux is not an alternative as long as users cannot run what they want on Linux.

I only see 4 ways for linux to actually gain on the desktop for the average user. 1) Truly painless cross-platform support for most languages. 2) A huge shift to web-based applications. 3) Microsoft seppukus. 4) Wine magically becomes painless and transparent.

Two is the only alternative that has a reasonable chance of happening.

That would make sense if there weren't another non-Windows operating system with about 10 times Linux's marketshare on the desktop. But somehow Apple has managed to get 10% of the market without any of that stuff happening, so I don't see how that can really be the reason.
>a proprietary iPad-like interface called Unity Unity is proprietary? Really?
"proprietary - Of or relating to an owner or ownership."

Unity is developed by Canonical. Canonical "owns" Unity. It is a frustrating choice of word, considering "proprietary" has a generally more specific meaning within computing. (more closed, trademarked, protected, etc.)

> Linux Mint requires users to do a complete re-installation, rather than a rolling incremental update

Um, and the author's solution to this is installing Windows?

I'm sorry, but at least when Linux Mint needs updates I am free to update when I want. Windows will pop up a nag screen and interrupt your work, even threatening to perform the update if you don't delay it. Anyone who has ever done work on a Windows machine has met this inconvenience. On the other hand, Linux Mint doesn't have it.

Also, his complaint is that it's not rolling release? Uh, neither is windows. Lets see, Windows xp, Windows Vista, Windows 7, Windows 8, yeah, it's rolling release alright. At least with Linux you don't also need to buy new hardware each time you update the OS.

I'm sorry, but your post is a little disingenuous. No, Windows doesn't do a rolling release--but, three things that mitigate this:

1) Windows releases are much slower. Most Linux distros have a new-shiny problem: I can't trivially (by which I mean, click a button and emphatically not muck with the command line or PPAs or equivalent user-hatefulness) get the newest OpenOffice or Firefox or Chrome or whatever on an older version of Mint (or Ubuntu--and dist-upgrading Ubuntu is a dicey proposition if you like having a working computer).

2) For the most part (there was some wackiness with architecture changes), Windows will do something much, much closer to the Right Thing on a reinstallation. Your data's where you left it. Your applications are still installed and work. It is not, on Windows, taken for granted that you must pave your machine (yes, I know you can put /home on another partition, that's only part of the problem) in order to upgrade.

3) The idea that you "need to buy new hardware each time you update" Windows is just silly. I have a laptop sitting here. 1GHz Pentium III, 512MB of RAM. I use it for playing old DOS games. It runs Windows 7 a lot more pleasantly than it did XP (which surprised me, but there you go). It may be way more pleasant to use on new hardware, sure--but so is Mint.

And by the way--you can set Windows Update to automatically do it when you shut down, and never nag you. (That this isn't the default is probably, on balance, a bad idea, but the capability is trivially there.)

Put your hackles down, dude. The mud-slinging helps no one.

Fedora does it somewhat better, I think. Run PreUpgrade and reboot into Anaconda, which will do the right thing. At least it seems to be a more fault-proof strategy than dist-upgrade IMO.

And of course with BTRFS, you can just make a snapshot right before the process and just roll back should anything go wrong. Of course there isn’t any UI for this yet, but hopefully we’ll get that as well, once Fedora adapts BTRFS as its default FS.

Furthermore, the default installation strategy for Fedora is to keep /home on a separate partition or LVM volume, so a complete reinstall is also possible.

As to the first point, it always takes some time, but Fedora 16 with latest updates is currently using Linux 3.3, KDE 4.8, Firefox 11 and Eclipse 3.7, so it’s not bad at all. Though Intellij IDEA CE is still on 9.0.

> 2) For the most part (there was some wackiness with architecture changes), Windows will do something much, much closer to the Right Thing on a reinstallation. Your data's where you left it. Your applications are still installed and work. It is not, on Windows, taken for granted that you must pave your machine (yes, I know you can put /home on another partition, that's only part of the problem) in order to upgrade.

You must not have installed Linux Mint or Ubuntu lately, because they both do this.

> 1) Windows releases are much slower. Most Linux distros have a new-shiny problem: I can't trivially (by which I mean, click a button and emphatically not muck with the command line or PPAs or equivalent user-hatefulness) get the newest OpenOffice or Firefox or Chrome or whatever on an older version of Mint (or Ubuntu--and dist-upgrading Ubuntu is a dicey proposition if you like having a working computer).

Not sure why this is an issue, because "shiny-new" Linux distributions come out of the box with a quality web browser and LibreOffice, and the versions are always recent enough and updateable. That's more than I can say for Windows, which comes with Internet Explorer and no office software by default, unless you count wordpad, which is clearly not what we are talking about.

I understand in an enterprise setting those arguments might not make sense, but in a home-desktop setting, they certainly do.

> You must not have installed Linux Mint or Ubuntu lately, because they both do this.

I've installed both recently, thanks. Neither save your system state on a reinstall--or if they do, it's certainly not the default and isn't made clear when you're doing the install.

The only acceptable thing is to reinstall and properly reconfigure any user-installed packages when they do a pave-and-reinstall, and while that's nontrivial, it's the only way to deal with the "oh, lol, reinstall everything every six months" clusterfuck in which Linux is currently enmeshed.

It seems more likely that nobody will do anything about it and instead we can have even more self-absorbed gnashing about Linux being primed to take over the desktop.

Windows, on the other hand, does this by virtue of not screwing with installed software. Loose coupling, believe it or not, has its advantages.

> Not sure why this is an issue

I think you misunderstand me. In order to get the new and shiny applications, I have to dist-upgrade. Which hilariously and regularly breaks like fucking mad. This means that, no, I can't really stay on an LTS release (and most people won't want to), because to keep up to date, I have to either play Russian roulette with dist-upgrade or...pave the entire thing. Not the case for Windows, which is what I was saying.

(comment deleted)
"Popular linux distros are frustrating." Ok, don't use one of the "user-friendly" distros. Get deb/slack. Problem solved. Also, heaven forbid you compile your own desktop system with the settings you want. Do you have to have the latest? Why not pick the version you liked best and install it? Isn't that the point of linux?

I personally don't care for Ubuntu or any distro that tries really hard to copy apple (or worse, come up with their own "better" way of doing things), so I can't relate. I've always used Slack w/ XFCE and never ever once (save the first two weeks I spent learning it) had a problem.

Linux is a complicated system, and you have to expect to spend some time learning it. If your distro's defaults are not to your liking you can either a) change them or b) install another distro. I understand the desire to bring the Window/OS experience to the casual linux desktop user, but this article seems to pick some distros which the auther has trouble with, and apply them to linux (to be honest though, the article was a bit long-winded and I skimmed some).

These rants seem to follow a similar pattern. First the author qualifies his experience with Linux by claiming he started in the nineties with some ancient distro. Then he'll lay on the praise of how great Linux was or could of been or whatever. And then they all seem to end with the same nonsense about how Linux can't "win" or "suceed" until it is adopted by the corporate market. As if going corporate is what FOSS were somehow all about. Fine, whatever, go back to Windows then. You probably never left.
The window for Desktop Linux has closed. The age of the desktop is ending; smartphones and tablets satisfy the computing needs of many people, and more app markets and low-cost devices will only increase this.
The same was said about paper when computers were invading the office space. The desktop will not die, nor will it go away any time soon.
Oh, one of those lunatics again.
Linux distributions (and GNU, and BSDs) don't have to "win" the desktop. They already have, with one important qualification: you have to care.

Mind you, if you are looking for one-size-fits-all, Windows fits the bill perfectly. All the choices have been made. But if you value customization, Linux and BSDs are your only choices.

(I mean, for Pete's sake, you can boot a room full of machines from a KNOPPIX terminal server, running out of a virtual machine, and then migrate that running VM to a system you just booted FROM it... And shut down the original! We're not in Kansas anymore!)

Linux is certainly less appealing as a dogfood, despite concerted efforts, in part because people can't take the brands they paid money for, along with them. That's why "free is not cheap enough" for many people and efforts to make Linux "sexy" can only go so far without this.

Commercial games are the epitomy of this and Linux doesn't get big commercial games. (Maybe that's because Linux users are mostly zealots who haven't bought a game in over a decade, because they are against DRM, etc... But that's a tangent.)

I'm not sure winning the dogfood market is even a worthwhile goal.

There is actually a LOT of duplication of effort happening right now with so many distributions trying make the Linux desktop "sexier," simpler, and more like Windows. Maybe it's just me, but that feels like the wrong solution. When I need Windows for "legacy" reasons (to run that proprietary accounting software, because that's what my accountant accepts), I just fire up KVM or Wine.

I don't think the Linux (BSD, etc) camp needs to take competing with Windows so seriously. They've already won. There is simply nothing better, if you care about code. You don't have to know everything, but you have to -- ehem -- RTFM. Non-coders simply aren't the primary audience. I know there are some who would disagree.

If you run a shop that doesn't allow the user the option of installing Windows software, or can standardize on document formats, Linux distributions (etc) are an easy choice today. That's how dogfood works. People eat what's in front of them.

Trying to sell Linux on the desktop without also being a fascist in this way, will only cause people to compare things to Windows. It's a recipe for complaints, in which case you might as well just feed them what they're really asking for.

That said, I agree with the author's point about rolling upgrades. If, for example, I'm working within a single package management framework (like APT or RPM, etc) exclusively, I should be able to upgrade, down-grade, side-grade, and any other grade imaginable. Reinstalls could and should be rare.

Ubuntu is 70 to 80% Debian packages, so why can't I "upgrade" (or downgrade) from Debian to Ubuntu, or from Mint to Debian (or any other package system to any other)?

I very much wish distribution projects 1) held onto a stronger sense of individuality while also 2) cooperating more to allow this level of interoperability. Simultaneous specialization and integration are not mutually exclusive: it's what modular software design is all about.

I think it's very much a matter of resources. Windows will always win because it has the greater market share. Regardless, they still manage to produce a dud every second release; think Windows ME, Vista, Windows 8 (pre-emptive).

My work uses a linux inspired toolchain, complete with emacs setup, however, we use the Microsoft compiler and target the Windows platform specifically because our users demand it.

My home PC is an ArchLinux machine, however, I'm still plagued by the inability to play my videos without tearing on my dual screen monitor setup with hardware acceleration; something that works perfectly well on Windows. I'm sure NVIDIA could provide a suitable solution but there's no money involed so they don't.

Linux will always have it's place in the server world, but as far as end user PC's go; I feel it's days are numbered.