Ingo Molnar on what ails the Linux desktop

The reason I stopped recommending Linux to "normal users" is _because_ of the concept of distributions.<p>Coupling the updates of single apps with the updates of the whole desktop or framework and libs, is just plain wrong. Having to upgrade the whole distro (including all the other installed apps you dont want to upgrade) just to install a new version of one single app you _want_ to update is a nightmare. Total bullshit. Users. Dont. Want. That. Users dont want one update to trigger another update, or even to trigger the upgrade of the whole desktop.<p>The blog post by ESR is one prominent example: <a href="http://esr.ibiblio.org/?p=3822" rel="nofollow">http://esr.ibiblio.org/?p=3822</a><p>He basically wanted to upgrade just one (obscure) app, and the process triggered the automatic removal of Gnome2 and installation of Unity. Just _IMAGINE_ how nightmarish this must look for normal users. You simply dont remove somebodys installed desktop sneakily from under their feet. You simply dont. That feels like the total loss of control over your computer.<p>I personally had, during the last 10 years, people go from Linux (which I talked them into trying) back to windows, _precisely_ of this reason, of having to upgrade the whole distribution every few months just to be able to get new app versions. They dont have to put up with this insane bullshit on Windows, why should they put up with it on Linux?<p>This "distribution" bullshit is not what is killing desktop Linux, it is what _already_ killed desktop Linux.<p>The other reasons why desktop Linux never made it (no games, no preinstallations on hardware) are imho just consequences of the distribution concept and the 6-month planned-obsolescence cycle. Nobody wants to bother with something which will be obsolete half a year down the road. Nobody wants to develop for a target that moves _that_ fast.<p>Windows installations, once installed or preinstalled, run for a decade. Develop something, and it will run on a 10 yr old Windows your grandparents use. Most people encounter new Windows installations only when they buy a new computer. PC manufacturers know that customers will hate it when their new computer OS is obsolete within half a year and that they wont be able to install new apps, so they dont preinstall Linux, it's as simple as that.<p>If anybody _ever_ really wants to see Linux succeed on the desktop (before the desktop concept itself is gone), he will have to give up on the distribution concept first.

The solution for this is quite simple: split the distro to a core, and a separate 'apps' repository, and let app authors control (but with a mandatory QA step) when their software gets released or updated. We did this with Maemo (and later MeeGo), and it has worked great: <a href="http://bergie.iki.fi/blog/application_quality_assurance_in_linux_distributions/" rel="nofollow">http://bergie.iki.fi/blog/application_quality_assurance_in_l...</a><p>Cross-distro app repositories are also a possibility, thanks to the Open Build Service (<a href="http://openbuildservice.org" rel="nofollow">http://openbuildservice.org</a>). And since MeeGo's community apps service is open source (<a href="https://github.com/nemein/com_meego_packages" rel="nofollow">https://github.com/nemein/com_meego_packages</a>), all software needed for this (including an app store client app) already exists.<p>What is needed is a major distribution to make the first move on this.

The Linux desktop is dead. Face it. Accept it.<p>How do I know this? Go to a developers' or tech conference, and what is the prevailing OS? 9/10 times these days, it's MacOS.<p>As much as Apple may be control freaks, they did desktop UNIX right and developers have flocked to them.<p>Meanwhile, the Linux desktop community is still having the same debates, the same difficulties it had back in the mid-90's. Dependency issues. Lack of compelling use case software; and that which does exist is versions behind current. Fragmentation.<p>The only areas in which the Linux desktop has moved forward is UI and user-oriented management, and Ubuntu deserves much of the credit for the latter.<p>Add to this the gradual movement away from the desktop paradigm to mobile. The desktop paradigm will still be around for several years (especially for creators--developers, media folks, etc), but the end-user is moving more and more towards smartphones and tablets for consumption. Extend those with external keyboards and monitors (think docks) for light creation work (documents, spreadsheets, etc), and you've the future.<p>Nope. The Linux desktop is dead.

I've been using Linux since 1993, and back then, the Linux OS and desktop were far superior to Windows 3.1.<p>Then Win 95 came out and that had a decent desktop. I remember when the KDE people started talking about a desktop for Unix and people didn't get it, but when we saw the beta it was like... Wow!<p>Then Red Hat Linux didn't like the license of KDE, so they had to create Gnome. As a result, rather than having one good Desktop, the average Linux has two half-baked desktops. This fork has wasted people's energy and been a distraction away from an excellent experience.<p>Another example of this is sound. I don't know how many incompatible sound APIs exist for Linux now, I know it's more than the fingers on one hand. The consequence of it all is that often sound doesn't work and unless you're a crazy enthusiast you might never get it to work.<p>I was a Linux zealot until 2003 or so when I had a job that had me using a Windows machine a lot, and by that point there was Win XP which was a huge improvement over Win 95.<p>I still use Linux on servers, but desktop Linux has largely disappeared from my life. Every so often I try to install it here or there, but I typically find the experience disappointing. I was a Fedora fan for a long time, but Fedora became increasingly finicky about where it would install. I switched to Ubuntu, but every installation ends up having some serious problem.<p>For instance, Ubuntu installed just fine on my PPC Mac Mini with the exception that the fan runs full speed all the time and the machine sounds like a vacuum cleaner.<p>Windows and Mac OS have been on a general trajectory of improvement -- sometimes there are changes you don't like, but the overall direction is good. Linux did, after years of struggle, get a stable multiprocessor kernel (2.6) but other than that I get the feeling Linux has been going backwards not forwards.

The author is right mostly, the problem is friction and market dynamics. Adding software in a given linux distro is too nerdy. If you don't have an "app store" like gui, you've already lost most users. Command line = friction for most users, even a lot of developers who grew up with windows.<p>As a linux nerd, I'm fine with command line + synaptic, but look at how well people have used the iTunes store, Amazon store, Google Play, etc... All of those have much less friction to find and download the right software than most linux distro's have. Ubuntu's market is close but...<p>WHERE IS THE NON FREE SOFTWARE?!!<p>If linux wants to do well for humans, paid, proprietary software NEEDS to exist on the platform. Ubuntu Software Center comes close, but it still kind of sucks.<p>Also, as a dev, it wasn't until VERY recently that you could even sign up to publish an app that was commercial in nature. It is hard to build a real marketplace when you're asking developers to give away all their work for free so that you can sell more operating systems (or support contracts) without the software dev seeing a dime.<p>As a software developer I can't feed my kids with free downloads on an open source operating system used by people who don't like paying for software.<p>Make it easy for devs to build software that people will pay for, then get operating system users who will buy that software for real money and you'll have fixed the Linux Desktop problem.

Ten years ago I was trying to persuade people that two things were very important for Linux to succeed in the desktop:<p>1) Distribution of software as a cross-distribution package that just has everything it needs inside a directory, libs and so forth. If you said this N years ago you were an asshole because "duplication of file blabla" and so forth. The typical example was "an user should simply go to some web site of some application, download a file, and click on it to execute the program".<p>2) Device drivers with a well specified interface between the OS and the hardware, so that different versions of the kernel could use the same driver without issues.<p>People complained a lot with technical arguments, about why a different approach is better than "1" or "2" from some kind of nerd metric. So the reality for me is that Linux does not succeed in the desktop because it is "run" by people with a square-shaped engineering mind. There is no fix for this.

Molnar's point about the <i>political</i> and <i>procedural</i> difficulties of adding new applications in the official repositories of most distributions is true (although this is changing -- witness Canonical's Ubuntu Software Center, PPAs, and "universe" repositories).<p>But his reasoning breaks down when he says the relative dearth of <i>commercial</i> applications for the Linux Desktop is due to this issue. That's not true.<p>The main reason why OSX/iOS, Android, and Windows attract more <i>commercial</i> developers is because those platforms have a much greater installed base!

I am running on linux desktop for already 5 years together with 50 friends, peers etc. It does not quite seem write to compare 90 cent mobile applications, which are 'mostly' few bunch of screens interfaced to a service otherwise given by a web site or simple 80s area arcade games, to packages in a linux distro. the author is totally making a terrible mistake here. mobile apps lack in size, complexity and who said they are great? they are mostly consumables which perish in few days or months (I exclude some ).<p>Ubuntu is already trying to be more flexible with software center and applications, however open source is not single walled garden which can adhere to tight monotonous architectures found in mobile.<p>In an linux distro the apps are diverse, and programmed in multitude ways by all possible programming languages (from python to lisp to C) and environments. That is reality and life, and what must be done must be done by being aware of that fact. and No, no one can force the broad, diverse open source world to a tight control a la Apple.

Igor's point about distros trying to "own" 20K packages is well-taken; it's simply not possible.<p>In my own life the 'solution' I have found is to use FreeBSD; I get a stable, well-maintained core with a sharp distinction between core, userland and third-party (the ports system).<p>I have found the ports system to be a lightweight, agile alternative to GNU/Linux package managers:<p>When you install FreeBSD you are left with a kernel, standard UNIX command-line utilities and everything you need to hammer the system into a finely-honed tool.<p>Right now, I'm using it exclusively on my servers because I'm willing to accept the trade-offs of Ubuntu (beta 12.04 on my dev box, XUbuntu 11.10 on my netbook) on the desktop; a little instability and fully-automated updates is OK in exchange for not having to fiddle with graphics drivers, sound, Flash, etc.

I'm just an ordinary user, but I'd put my 2¢.<p>Nowadays, any developer can create a packages for his application (and any of its dependencies) and publish it in self-hosted repositories. Users can easily add such repository to their system's sources list and bureaucracy problem's over. Well, some developers are doing this already - the only thing that keeps the rest of them is either ignorance or complexity of the packaging process.<p>I believe there's no need for an Android market clone (which is yet another centralized repository). What users may need is just a directory, pointing to external repos. Ubuntu market seems somehow promising (at least I remember seeing some dialogs like "you need to enable this source to install that package").<p>Content duplication is not a problem. A real problem is keeping the system up-to-date when you <i>want</i> to update some library, because of an important feature or bugfix. And it's nearly impossible with every application's bundling their own copy of that library, with some copies being actually incompatible forks (and a lot of copies being just different builds - think of different compiler versions - of exactly the same sources).

UNIX in general never transitioned well from single binary applications that could just be placed in /bin /usr/bin or /usr/local/bin as appropriate to multi-file applications that needed an array of libraries, resource files etc. That coupled with a file system layout that nobody could ever agree on made it a mess once UNIX moved outside the curated environments it was typically found in prior to Linux.<p>NextStep made a good stab at the problem with bundles (.app .service .framework etc) but once you moved outside of the abstraction layer you were right back into the mess. Most Linux distributions seem to want to emulate SunOS circa 1992 and any attempts to "improve" on the solution in drowned out by fundamentalism.<p>I actually though the author nailed it in the last paragraph of part 2 of his post. The free software movement needs to start looking forward and not try and emulate what worked 20 years ago. There is definitely potential for a brave organization that is willing to try and tackle the challenge.

My impression is that the problem is more fundamental: a lot of open source and free software which finds its way onto linux is made by people who don't seem to care about the end-user experience. If it works for them, there is no need to improve it.<p>Not all open software projects are receptive to changes, improvements or bug reports from strangers so nothing is really resolved without forking by which adds its own complications.<p>Of course there is still good open source / free software. It's just hard to come by.

So, how does the fact that any person or company can host their own packages or even full repositories that can be added <i>with a single click</i>¹ and are not dependent on hierarchical organizations fit in that?<p>GNU/Linux distros, at least APT based ones, are perfectly distributed if the person wants to.<p>¹ If you have apt-url installed, which Ubuntu has <i>by default</i>

I disagree. The author seems to forget the fact that packages are only a layer of added ease of installing new software. The act of installing new software from source or binaries distributed by the author is still as free as ever.<p>I admit that I do install about 90% of my programs as packages, but the problem of central authorities responsible for patching and distributing software and taking too long to do it isn't present in every distribution. I've used Arch Linux for years now, and it solves this problem by separating the packages into an 'official' channel of reliable maintainers testing and releasing new versions on the package system and a user repository where anyone can add packages.<p>To me, this seems like the optimal solution. Community-maintained packages can be promoted to official ones, from what I can see new versions are released from testing within days and if you're not satisfied with how others maintain the packages, building the packages from the newest versions yourself is almost as easy as installing binaries from the repository because anyone can use the build and packaging scripts used by the maintainers themselves.

There are "core" linux distros, like Slackware. There are many attempts at autonomous apps distribution like zero-install and openpkg that mostly work, and probably would work fine with a reasonable effort backing them.<p>[1]: <a href="http://0install.net/" rel="nofollow">http://0install.net/</a> [2]: <a href="http://www.openpkg.org/" rel="nofollow">http://www.openpkg.org/</a><p>Of course the problem is that the big distros (redhat, debian) can't be bothered to care.

Yes, I see the point being made in the original article. I tend to use LTS Ubuntu and conservative distributions (currently PUIAS) and so see a slower rate of change.<p>Another Hacker News thread is discussing the new release of Audacity.<p><a href="http://news.ycombinator.com/item?id=3714766" rel="nofollow">http://news.ycombinator.com/item?id=3714766</a><p>and it occured to me that I would like to try to compile a <i>statically linked</i> build of Audacity that could work on any version of GNU/Linux from Ubuntu 12.04 down to (say) CentOS 5.7. Just a big binary blob that I could copy and run.<p>How would I find out how to do this? I've compiled little things before (dwm window manager, qalculate)

It really frustrates me that authors are not responsible for compiling and distributing their software, only then will we have software upgraded regularly and quickly.

This is basically why Ubuntu has PPAs and what they are trying to turn Software Center into right?

I've been happily using the stable branch of Debian for about 4 years now. Whilst I agree bureaucracy and politics should be avoided where possible, it seems to me that handing everything off to a 3rd party to focus on the "core packages" harms overall system performance.<p>Time and again I hear people in the lab complain about all these bugs in their applications, running Debian I can honestly say I don't have this problem. Ofcourse, I don't have the latest software either, but for me that's the price I pay for a stable system.

Most of the complaints here don't even have any suggestions, just a complaint that can be easily translated to "dependencies in software are hard to manage". Well, no shit. How about some interesting ideas around this from the "hackers"?<p>The ideas that are proposed are mostly things that have already been tried but have failed because of social/manpower reasons (they would take enormous amounts of effort and time from all involved for little benefit) or technical (they don't work)<p>A large percentage complain that the distro in question updates <i>too</i> frequently, when there are clearly distros that cater to stability (they just aren't the "cool ones").<p>Some of the complaints are demanding some mythical OS that allows you to install it once, never have it update or change but still have access to all the newest software. That would be wonderful. No one has figured that out yet, not windows, not mac, not any *nix.<p>I know bitchy comments aren't helpful, or likely to be well received, but there must be some other people of my ilk still on HN. Now that it's Product Guy News, where should I be going? Where is this story posted with people actually talking about interesting ways to improve things who actually understand the problem and could be called hackers without the technically skilled people laughing?

Part 2, where he discusses his solutions, is more enlightening I think.<p>He's dead-on about the impassable problem of package updates affecting other packages. Having everything sandboxed with a general permissions system for directories instead of per-file is also better (this is how MacOS wanted to work before OS X). A free and open mesh network with reputation-based security is also the future.<p>But hey, linux is wide open, if these are the changes that are needed, we will see them.

Isn't the problem that the distributions are downstream to the apps and libraries they install ?<p>Quoting android and ios ecosystem is well and good, but in those ecosystems the os comes first and the apps are developed downstream. You simply cant have that in a linux eco system.

Why do linux distros keep trying to recreate the gaudy, confusing experiences of the proprietary systems? I want my drivers to work, and a window manager that wraps the file-system - allows interaction with the files. In an ideal world, someone would restructure the four bin directories so that system stuff was in one area, and user stuff was somewhere else so we could easily open what we want by navigating the tree.

Just for another POV -- I love the fact that I can apt-get a version of practically any FOSS project of note, and within a minute or two, I have something that works with the rest of my system. If I need bleeding-edge, I go to the project page and download a later binary or source, but 90% of the time that's not necessary.<p>For me, it's a perfect combination of a vetted ecosystem (ok, somewhat closed but closed in the way I like -- no crapware, all legit source distros and mostly mature projects) with the ability to go outside that system at any time, at my own risk.<p>Anyone can set up a repository to add to/compete with Canonical's, and of course they do. So with my willow garage repo's, I can keep up with their concept of what's stable, etc. It works nearly perfectly, IMSHO.

Isn't the real issue simply that the desktop distributors simply lack the manpower required to out together a stable desktop release, party due to the massive fragmentation?<p>I assume MS and Apple have huge teams dedicated simply to making sure that all of this software works together nicely.<p>If there was a desktop distro that cost $100 a throw and that money was re-invested in testing the desktop platform more thoroughly and making it past and future proof that this would solve many problems?

Sounds like he wants a distro agnostic Ports system, one that hides the dirty work of watching compile time crap fly across the screen and just gives the user a suitable package for their distro.<p>Either that or he's essentially suggesting we move to static compiled packages, which while tremendously inefficient from a space and security standpoint would alleviate at least some of the headaches of trying to do cross distro binary offerings.

Maybe Android will conquer the desktop then?<p>All it needs is mouse and keyboard support in the interface and higher-resolution apps, which will come for tablets anyway.

Good point. But most of these apps are written, because they can make money for the author. There just isn't a large enough user base in Desktop Linux to make that money, and hence much less people willing to invest the time to write apps. And the much smaller userbase is split up between Gnome and KDE and now Unity and Xfce and so on. So there is even less incentive to write apps.

Linux is being used by more developers that I see than ever and even regular people are finding out about it and trying it out.

The author mostly talk about maintaining packages. Some people say that it is the missing games and other desktop applications. However, we all know that Linux is best OS for development. It has best set of libraries, languages and utilities. Yet, why there not many cool Desktop applications?<p>I believe the answer is, Linux doesn't have an IDE for Desktop development. Look at windows, they have .net Platform and IDE, Look at Mac, they have Xcode and Android uses eclipse for IDE.<p>Most of the application developers want and IDE, which is what is lacking in Linux. I hope someone makes an IDE for Linux desktop development.

Part II of Ingo Molnar's blog post: <a href="http://news.ycombinator.com/item?id=3719719" rel="nofollow">http://news.ycombinator.com/item?id=3719719</a>

I have to wonder if people's expectations are just a little to high for something that is free. I think a large amount of people complaining about Linux did not use it when it really really sucked (back when Slackware and RedHat were the only options).<p>Linux is a hackers operating system. When people try to make it into a desktop operating system it starts to suck (Gnome 3).

Honestly, Linux is a lot like the tiptronic setting on my cars gearbox. I have thought it's a cool feature and I'm glad it's there but really... I never use it 'cos it's kind of fiddly and dumb.

what I prefer in this posts is the interesting discussion coming with the article. It seems attention whores and trolls have not yet reached g+, it looks pretty sane for now.<p>mode religious:on please g+ try somehow to make a good karma system to keep the NSR low.