Shared publicly  - 
miguel had a pretty cool post on why the linux desktop failed.
Daniel hernandez's profile photoPierre Sauter's profile photoJeaMY Lee's profile photoLuís Fonseca's profile photo
Alan Cox
"The second dimension to the problem is that no two Linux distributions agreed on which core components the system should use."

That made me laugh. There was KDE and Miguel then came along and created the very confusion he's ranting about. He was also core to ramming CORBA down peoples throats which then had to be extracted slowly back out of the resulting mess that blighted Gnome 2.x and occupied vast amounts of developer time.

He's dead right about the way the Gnome people keep breaking their compatiblity eveyr time not just with the apps but with the UI, with the config (which is still worse now than in Gnome 1.x !) and so on.

However it's not an Open Source disease its certain projects like Gnome disease - my 3.6rc kernel will still run a Rogue binary built in 1992. X is back compatible to apps far older than Linux.

As for his ranting about the audio I blame +Lennart Poettering 8)  - the kernel audio hasn't broken compatibility, it even has OSS compat layers back to the beginning of Linux audio support. Actually blaming Pulseaudio is mean too (but it's fun blaming Lennart and that's what he exists for) - it also has compat stuff designed to make ancient apps just work 8)

Gnome isn't really a desktop anyway - it's a research project.
The gnome people claiming that I set the "attitude" that causes them problems is laughable.

One of the core kernel rules has always been that we never ever break any external interfaces. That rule has been there since day one, although it's gotten much more explicit only in the last few years. The fact that we break internal interfaces that are not visible to userland is totally irrelevant, and a total red herring.

I wish the gnome people had understood the real rules inside the kernel. Like "you never break external interfaces" - and "we need to do that to improve things" is not an excuse.

Or "different users have different needs". The kernel was - and is - happy to support both the SGI style thousand-CPU machines and the embedded vendors with cellphones and routers. The fact that they have different needs is very obvious.

I personally think that one reason that the Linux kernel has been so successful was the fact that I didn't have a huge vision of where I wanted to force people to go. Sure, I wanted "unix", and there are some very high-level concepts that go with that (fork,exec,files etc), but I didn't want to enforce any particular world-view outside of that very generic pattern.

In fact, Linux pretty much did what I envisioned back in 1991 when I first released it. Pretty much all subsequent development was driven by outside ideas of what other people needed or wanted to do. Not by some internal vision of where things "should" go.

That's exactly the reverse of the gnome "we know better" mentality, and "We will force Corba/.NET down your throat whether you like it or not, and if you complain, you're against progress, and cannot handle the change".

Some gnome people seem to be in total denial about what their problem really is. They'll wildly blame everybody except themselves. This article seems to be a perfect example of that.
Linus, My involvement with  Gnome stopped about five years ago, and I merely stayed in the periphery because I used Gnome as a user and we built C# programs that used Gnome libraries.  So it is unfair to the Gnome guys to attach my position to their project.  I have not talked to them in a long time, and have no idea if anyone there even espouses my opinions.

Although you might have a strict policy for the kernel ABI, which is commendable, and I cherish that post of yours where you make this case forcefully on some mailing list, my opinion is that the attitude of the kernel developers influenced the way the FOSS community built software.

There have been entire talks and heated debates devoted to the question on binary drivers and why you guys consider breaking that ABI fair game.   The problem is not whether you guys had that right or not, but that the entire mood in the space was "we do not keep junk around".

You have a strong personality, and so does a lot of the people that surround you and your strong personality, like it or not, influenced the attitudes of people.

One example where this has transpired was the mood on the kernel lists (this is what I remember from about 1999-2000, so this is dated).   My take is that you are brilliant, clever and funny, and you can also be mean and harsh.   Many people tried to imitate you, but they were neither brilliant, clever or funny.   They just turn out to be mean and harsh and this attitude spread on the mailing lists.   

So the message that most people heard was we do the right thing, even if we break software.  And they did.

From APIs, to printer subsystems, to audio systems, to init daemons to bus systems all of these very fine changes to the stack have caused problems for 3rd party ISVs that want to support the Linux desktop.

Supporting Linux desktop for proprietary software developers is just too expensive and the market is both small and deeply fragmented.

As for Gnome, personally, I would like them to make some changes, and I even agree with some of your complains about the Gnome Shell.   But it has not bothered me as much as it has bothered you.

You take on a couple of tangents, and I wanted to reassured you that nobody is going to force anything down anyone's throats.

When it comes to CORBA, both the KDE guys and ourselves, out of our own naiveness embraced it to solve a series of problems that we thought we would have, that we ended up not having.   Feel free to lambast me for advocating my poor choices at the time.   Smarter heads prevailed and CORBA went out the window.  What can I say, I was young, and the KDE people were young.   In both cases, the mistake was corrected, and there is no CORBA left for you to suffer.

You do not have to worry about .NET either.   Mono is not part of Gnome, and no apps shipping with Gnome use it, so you are safe there too.
Alan, love you too.

I am surprised that you do not seem to remember that you were involved in the launch of Gnome, that you encouraged us to develop Gnome on LinuxNet, that you had a problem like we did with the Qt license, that you contributed to Gnome, and that you even participated in the very first Gnome meeting at pre-IPO Red Hat.
Alan Cox
I haven't forgotten that - but you can't argue it both ways. The split created part fo the problem and you created it (for reasons I think were probably justified at the time). You are trying to have your cake and eat it.

One desktop I think btw has stayed true to its origins, tidy clean and fast - thats E. It's changee a lot - the code is now beautiful instead of looking like the ravings of an undergrad on excess caffeine but it's stayed true to its goal.
I dunno, I remember E started off as wm, and then it started feature creeping into a desktop pushing out the release as it kept going.  It never seems to be done enough to put out a release.

Yeah I heard EFL is pretty nice.  Glad to hear the code looks good.  Does it have a lot of misspellings? :)
Alan, I used the collective "we" in my whole piece to enlist me as one of the parties responsible for our problems.   

Perhaps you and Linus felt that I was unjustly attacking Linux, and that is my bad.  I should have titled the piece "What Killed The Open Source Desktop".

We tried to make it happen, we lacked the discipline, we failed to deliver the system that we envisioned.  The piece focuses mostly on the failures of userland.   Perhaps it can serve as a warning of what to avoid next time someone wants to create a desktop.

But the failures were not entirely technical.   There were political forces behind undermining the LSB, there was always "one more thing to fix" attitude that prevented Debian from shipping quickly, and all this freedom created Linux architects that envisioned very different versions of the system.  

We have the unlimited options, it was just not what third party developers that were used to ship-and-forget wanted.

I think it is too late for the open source desktop to become a mainstream consumer OS, but it will continue to be a solid workstation OS.

Linux has succeeded beyond my wildest dreams in the form of Android.   And perhaps Linux as a kernel for a web browser will be the next big thing, but it wont be what I had desired.
the EFL code is beautiful? I assume +Alan Cox has been reading it off some other (private) version control system, because the one in SVN it still looks like the product of an undergrad on excess caffeine.
Alan Cox
+Rodney Dawes E always tried to be very compact, very fast and very themeable - it still does that nicely.

+Emmanuele Bassi maybe its a matter of taste - I think the code is nicely done.
I have heard that E will get releases now. Again they are too lade for anything. Maybe they can get some of the (old) GNOME users.
+Garry Lyndon I don't understand your comment.  What your describing is a general high tech blabber that occurs regardless of what OS you pick.  If you're around technical experts, then you're going to get similar conversations regardless of community you're in.  If you adopt a distro and then stick to the distro communication channels you'll be fine.
Sorry, Garry, but your comment begs this question: have you never read a manual? Really? Did you learn to use the clothes washer (or the VCR) without either asking or reading the manual? Sooner or later you will have to read the manual (be it through reading or through asking/imitating someone knowledgeable). It happens on all walks of life, so please don't use that excuse. You have been RTFMing your whole life: using a computer shouldn't be a reason not to.
You were told Linux is ruled by bullies. Let's suppose, for a second, that it is. Does this mean you must avoid downloading even a simple Live CD? Come on, man, you have probably done many more dangerous things than that and suffered no harm even when people told you not to do them. And no, Linux isn't ruled by bullies either, the same way normal life isn't: it just happens that bullies are usually noisier.
Then again, it's your choice, so please don't take this as more than a rebuttal of your arguments and some suggestions.
+Garry Lyndon im not a bully, but I read the manual before blame someone of something. For adopt Linux just use it, read it and learn from it.
Ingo Molnar
+Miguel de Icaza

The Linux kernel project was always very strict about keeping its external ABIs, intended ABIs and even accidental ABIs.

This is how it's possible that Alan Cox's 20 years old Rouge game binary built in 1992 is still compatible with and working on today's latest Linux kernel unmodified - and will still work 50 years in the future.

Breaking applications was never acceptable for the Linux kernel, now or 15 years ago. It's a binding promise which we keep at any cost - in the past we've reverted major kernel features that broke apps.

There's no kernel feature or cleanup or "grand new vision" worth having if it's not binary compatible. People have paid us by porting to Linux and using Linux - and we pay back by keeping binary compatibility. Full stop.

(Kernel drivers that link deep into the kernel were never applications - hence they are not using any Application Binary Interface. Citing kernel driver policy when talking about applications is a dishonest argument.)

Your claim that Linus's attitude somehow influenced Gnome to break apps with every release is frankly either revisionist history or plain ignorance.
Hi +Garry Lyndon, It's actually normal to have 'fan boys' in any environment. By the way, if you are not a Linux bully, then you are a Windows or an OSX bully. Garry, if you have ever used Windows, you would most likely feel right at home on a Xubuntu flavor, no manual required.

The only problem I could see you having, would be Audio problems. But still, much less tinkering than a fresh install of Windows. I have found that, generally, things just work without intervention; yet still giving you the freedom to tinker if you wish.
E is great! I think it will go well now - embedded gui of choice, in many cases... Rasterman for president! ;)
+Gene Wildhart: 'The problem with E is, nobody uses it.' Well, that's exactly what they used to say about Linux... a very, very long time ago :)
+Linus Torvalds “That's exactly the reverse of the gnome "we know better" mentality” — You mean you run surveys to make popular decisions in the kernel development? :) Did I just wake up in a parallel universe where Linux kernel development is a frigging democracy? :)
+Miguel de Icaza you are not right. Linux desktop isnot dead (but Gnome did all for kill Linux as os for masses) It is mostly because of Kde - the real desktop - not research project. And other things like e, fluxbox. Theyare never say "hey! I will say you how you will work, if you dont likeme, shut up andgoto apple win etc".
Hope Linuscomes backto Kde.
sorry - i amfrom tablet
+Miguel de Icaza So, if I understand your argument correctly. You are no longer associated with the GNOME project for a very long time, yet you believe the problem they are dealing with is too much arrogance. Which is probably true (see for example the discussion between them and Mark Shuttleworth). But you blame the linux kernel guys for that?! I mean, come on! Doesn't seem fair to me.

Linux desktop use didn't go where some of us long time ago did think it would go. But boy, did it go places we weren't expecting. Almost all new embedded devices with GUIs, a hugh chunk of the smartphone market which is much bigger than all desktops combined (Android).

As a developer workstation it's workable. I especially like what ubuntu did for example with the 12.04 release. But it is still a bit painful to go to the process of learning the new shortcuts at my age. I hope they will never change them again... or commit mass suicide instead.

On Mac, the shortcuts are still the same as in my early days of computing. In Linux, the basic copy-paste shortcuts aren't even the same in the 3 major programs: Firefox, Terminal and OpenOffice. And in some programs they are missing altogether. Come on. In 2012!
The gui is the way the user interact with the computer. Not the graphics themself need to be 100% stable, that can be nice to upgrade, but the actual keyboard and mouse inputs should be as stable as how the kernel deals with userspace: rock solid.

In the mean time, the desktop environments have become so overbloated without even focusing on the very basics, that they become less useable on everything but the best performing embedded systems or recent workstations.

And, +Miguel de Icaza, that is in my opinion the reason why the Linux desktop isn't better. I'm not 100% sure if it would be a lot more popular if it was better. But it would be a nice try.

Macintosh Human Interface Guidelines where already published over 20 years ago (and I remember as a kid that it was a good read), and tell developers which widgets to use for what. Leading to consistency across the board, and - for example - no 'creativity' with respect to keyboard shortcuts. And - believe me - keyboard shortcuts are VERY important for your power users. Which we all are...

If GNOME would - even in an arrogant manner - protect those principles, I would yell: 'Way to go!'. Enforce consistency, enforce user interface principles with a great deal of respect for POWER USERS. Sometimes they say that GNOME resembles MacOS. Because of its simplicity. But sadly enough the GNOME developers then don't know anything about all the power user shortcuts embedded in the MacOS gui!
Garry Lyndon
When you was in a school, did you know anyone that can be described as a bully? Was a school to blame for? Did school policy protected them to any extent? I don't think so.

There is very little difference in Linux communities. There are good people in majority and bullies in minority. There is no netiquette that protect cyber-bulies to any extent.

In majority of Linux bully cases that is just wrong identification inspired by posts like yours. People come to Linux and think that as soon as they know 1+1 they can use RTFM, but they learn very fast they can't. In openSUSE RTFM is not used even with people that are really hard to help. It wasn't always like that, but in the last few years those that don't like it have strong voice compared to those that find it a suitable joke.
That would be ideal, +Alexandre Prokoudine. The words we use need representation before the numbers at the kernel level. One makes sense,  so does solid-state symbol tables, especially if we know the value of an entry is same across the cores in more than one instance. Between the kernel and desktop we need some way to sort these static votes.

Maybe hope for wireless electronic paper or electronic paper that only communicates with the pen, and sign-off on it.
Martin Owens
+Miguel de Icaza Sounds like someone has mid-life cynicism. Do you need another hug?

See, we're not a business Miguel, we're a community. There's a different meter of success. I think your frustration is more about our inability to attract apps and attention from the proprietary universe. Understandable, but not always related to number of users. Sometimes it's just perception of the health of a platform that does it in, I haven't heard you being hopeful about the Desktop for many years so why should anyone else?
Steve A
Linus and Steve Jobs have this in common - they had strong visions for their projects, and they both succeeded wildly. Linux is the best OS kernel around, and OS X is the best consumer desktop.

As a happy linux user/dev, i find it a bit of a shame that none of the Linux Desktop projects have ever quite had similar success. (I've finally been won over by KDE-3 - it's still not a patch on OS X as a user friendly desktop.)

I suppose this is the frustration behind Miquel's rants, He has a point, but he is not a constructive force in the FOSS world anymore, and should probably pull his head in imho.
+Jonathan Ballard I've no idea what background you have in software development, but it looks like you have a close to zero experience dealing with masses of users.

Well, let me give you a perspective of a +GIMP team member.

All the great free end-user software projects I can think of became great, because developers were communicating to users who thought along the same lines.

And it's the best way to work on a project, because you keep interacting with people and improving your work, while still belonging to yourself.

What happens when you let democracy in? Ugly mess. Suddenly people start treating you like you owe to them and should bow to their wishes.

— Hey developer, I used to use X application on Windows. I want the Y  feature to be like in that app. What do you mean, it's supposed to work differently? Well, make it an option, you idiot.

— Hey developer, there is this app for Mac that's a bit like your app, except it's for a different target group, different use cases and different task applications. But I want one of its tools implemented verbatim anyway. And I want it now. Not going to? Well, I'm a user, and you should be listening to me!

That's just bullshit. Please keep your democracy to yourself.

If you want some free software to change, learn to encourage, learn to make well-fitted proposals that make sense, learn to understand design decisions, but also learn to accept that the developer is the one who has the final saying, because (s)he's the one who's responsible, not you.
I think Gnome did the right thing. A bit late infact. They should have started earlier and made the transition smother. I can not categorize myself as scared of new things and finds the new workflow better than before. The extensibility is easier and more powerful than ever and pretty much at no extra resource cost. That is not bad... or is it? This debaite can sometimes feel like talking to IE fanboys back in the days when Firefox was new. Gnome-shell is great and will become even greater as the arrival of new iterations continues. The 3.0-3.2 was not ready for public release... but from 3.4 and onwards i have liked it every day.

Linux is growing on the desktop and i bet will be growing even more soon. Am thinking on Valve and such.. The finger of Linux seams to have worked as Nvidia are upping their work on Optimus. And Ubuntu are gaining some traction. We will soon have really cheap devices for anyone to use... those will need a kernel. In a way it has all ready started with RaberryPi... but it is just the beginning of that segment. I know a lot of people wanting and are trying Linux on desktop because of this. It started with  Linux on smartphone (Andoid/Maemo/Meego) and will become even greater in the future. It is a type of circle back fenomena... people are not as afraid of Linux any more.
+Ingo Molnar """Kernel drivers that link deep into the kernel were never applications - hence they are not using any Application Binary Interface. Citing kernel driver policy when talking about applications is a dishonest argument."""

Kernel drivers might not be applications but they still are enablers of application and desktop use. Without a stable ABI for drivers you get subpar desktop device support.

Furthermore, what Miguel said was something different: that this attitude towards driver compatibility in the kernel influenced how people also designed framework and application level ABI compatibility.

That the kernel maintained strict external ABI compatibility is a HALF MEASURE. It should have maintained strict driver ABI compatibility too to set the right tone. 
+Alexandre Prokoudine That first paragraph is quite offensive, as I didn't judge you, and I certainly didn't ask for it on this thread. I am in this world that used printers and paper practically everywhere, so I know it is only paperweight versus massive users of it. I think facebook knows that, but they are in the republic state of california, despite whatever you think democracy does here, ideally.
+Valdos Sine That Linux "is alive" doesn't matter at all. The discussion was about Desktop Linux. Desktop Linux set circa 1998 to capture the market share off Windows and it failed miserably to this day. Doesn't matter if it's "great" or if you use it and like it. Alive in the context of this discussion means: with big user share, a vibrant application ecosystem, and third-party support. In these fields, desktop linux has failed miserable.
Making my living on various Linux machines for years, I nowadays have no trouble with the kernel nor with sound nor with video, ... I do have issues e.g. w/ Unity breaking some GUI applications I have to use in work, so I'm now on Xubuntu. That's about all my problems w/ Linux.

PS: Machines I do install for myself do get cleaned from Mono.
+Nickos Venturas

Every single bit of the kernel that executes on my desktop system is a "critical enabler of the desktop" - without them the machine would simply not function.

Claiming that internal, application-invisible kernel interfaces should be treated the same as external Application Binary Interfaces is disingenuous...

It is fair and valid to compare Gnome's behavior towards external applications to the Linux kernel's policy towards external applications.

It is not fair and not valid to harp on kernel internal interfaces and mix up these starkly different domains - Linus was not criticising Gnome-internal interfaces either...
Gnome dev is optimistic about gnome. Shocking. Guess what. There is a large and ignored collection of us whom have quit using GNOME after years of usage. GNOME hasn't acknowledged us.
The issue with RTFM is that it is often treated as if it was MTFM (Memorize..). I'm always amazed by the energy and time people will invest that provides no constructive value and yet makes other people feel bad. I think people who are tempted to write "RTFM" should instead "STFU".
+Shlee Paradiso No, actually we still love you like a long lost brother.

+Garry Lyndon Didn't have to RTFM with MS-DOS? Really? All the commands just popped into your head all of a sudden, didn't they? :)
+Jonathan Ballard Thanks, you've just demonstrated the exact issue with GNOME criticism: looking for offences instead of sitting down and listening for just one minute.
Having just bought a MBPr I realise that actually the linux desktop has come a very long way!  I will go back to linux as soon as there is a decent dumb-user-friendly install process for my MBP.  I should also add that if my girlfriend who utterly hates technology can now live with a linuxmint machine, then perhaps this entire conversation is about 2 years to late.  My gut feeling is that the linux desktop is already a success, it just needs better marketing and forums/posts like these don't help.
Dunno, but, in my house: all Linux Desktop
Where I work: 40 Windows, 1 Mac, 10 Linux
Im going to talk as a user:

I just want to say I love Linux. I've used it for the past 10 years. I'm running Crunchbang on my old IBM Thinkpad, Arch on my desktop and others on my servers.

Where my DEs are concerned I admit I've had to tweak things to get them working in the past, but it's never been a heart ache.

What I love about Linux is I can get it working on anything I own and I've been liberated from corporations forcing me to play the game their way and have a DE a certain way. I can have it my way in Linux. - I've been provided for very well by FOSS applications, so thank you very much to all devs, and documentation writers.

When I was a kid in school we were taught how to use MS Windows, not computers. That really held me, and everyone else, back IMO. I was being trained to be a secretary (tyvm UK curriculum)! I have learned so much more about computers and technology since using Linux, either by enthusiasm or necessity.

And technology is playing such a large role in our lives now and kids are so in tune with it, they'll be fine. It doesn't take much effort to get on with Bash and such like.

I also perceive that the generations below us who get the opportunity to have Linux on what ever hardware they can afford/get their hands on, or a distro which suits them, only serves to help them get into technology in a more practical way and help the progression of our future prodigies and evolution of tech.

So aside from this debate, I hope this minute insight into my perception of Linux shows its promise.
+Garry Lyndon See? Instead of talking about real experiences you go looking for offences and of course you find them wherever you look.

No, actually I'm prepared to believe that you really think you didn't RTFM in the days of MS-DOS. You see, personal computers have changed a lot since then, so maybe you've just grown to know software to just work so much that you think it was ever so.

But no, it wasn't. If stating this makes me a troll, then I feel really sorry for you.
Little Garry was out of breath.  He had tried to escape, but now he was backed into an alleyway with no way to escape.  He turned back the way he came and froze-- a round, pear-shaped silhouette stood the way he came, blocking off his only exit.  His heart began to race with fear.

Beads of sweat pooled beneath Garry as the round figure waddled its way towards him.  It made its way towards Garry, walking down the dark aisle like a battered Bobo doll.  It was close now.  This was it.  Garry's body tensed in anticipation of punishment.

As the figure neared, a beam of light revealed the sinister countenance of his pursuer.  He could see into those cold, hard eyes--eyes that revealed no thought, blank, empty eyes that hid hate and malice.  But worst of all was the tell-tale beak, the beak of a penguin.  And that beak parted, and the figure spoke.

"Today, I think my lunch is gonna be free... free as in beer."  

And the penguin waddled off with poor little Garry's lunch money as Garry sat there sobbing with his hand over his black eye.
We could encourage Desktop Linux to proxy OpenGL ES and disable OpenGL, so we can follow standards much easier than ad-driven SDKs, which so many people like. Maybe we can inspire people for the embedded user interface that is capable to work with +GIMP .

Thanks for the insight on how +GIMP users are treated and how developers should be treated. +Alexandre Prokoudine 
Has anyone seen/tried running Android 4.0+ on a Desktop? or a TV? Not so great, but getting there. If Google nails the UX/UI by 5.0, make it run well with Intel, merge ChromeOS, Android as a desktop, with more or less the same Linux Kernel, will have a good shot.
Linux Desktop failed because there is no such thing as The Linux Desktop. There is a huge variety of different systems, all competing to be a desktop. Linux as a desktop OS will never even be as successful as OSX (~4% market share) because there are simply too many variations. Commercial software simply can't get a foothold because the ground is so unstable. 

Case in point, I'm fighting to get a big commercial software package ("BCSP")  working on Ubuntu 12.04. There is no standard way for BCSP to hook into the desktop (a la Windows' "Programs -> BCSP"). The installer was a shell script and had to be run from an open terminal. And I had to debug the script.

Doesn't matter what the app is. Binary graphical applications on Linux are a complete nightmare. KDE, Gnome, OpenStep, WindowMaker, on and on and on. Qt? Gtk? Motif? Tk? WxWidgets? How many others?

What is success? Perhaps others consider Desktop Linux to be a success without commercial applications. All I know is, I CAN'T USE IT. If the apps I need won't run, I can't use it. I want to use Linux. I love using Linux. But I'm constantly forced back to Windows.
+Garry Lyndon Thanks for proving the exact point that +Sriram Ramkrishna was making. Bullies exist in every community, and you just outed yourself as one of them. You commented on this thread for the purpose of being a bully (by calling all the Linux supporters bullies who tell you to RTFM), and then by replying again to everyone who stated how wrong you were, by calling them bullies rather than understanding anything; and trying to use your list of skills or accomplishments as some excuse for how you should be treated better than someone else.
Bad article. Biased and wrong.
I'm not a programmer. However, this guy is talking about things he may know, and NO windows user knows about windows, and perhaps 90% of the programmers that make "for windows" software don't know either (layers, codes, sub-cores, etc.).

And MacOS ecosystem is a one song record, with absolute control from one company that was, most of the time, a one person project. No surprise there is no diversity or compatibility issues!

It looks like he forgets newer kernel versions have been cleaned and made quite neat, as well as some packages (Libreoffice, for example).

The so-called fragmentation can be a problem, depending on the side of the coin you look at; the other side being diversity leads to more than one intelligent solution and prevents malware and other cr@p to be created and spread inside the Linux ecosystem.

Incompatibility, by "top dogs" or lesser dogs is not a real thing, I bet.
Even more: the hardware or software industry can easily opt for doing something just for Red Hat and family, or Debian and family, or even Ubuntu and derivatives, and thus develop something for "some linuxes". If they don't develop it's not because of the "fragmentation", it's because they don't really want to!! Because if they do develop, for instance, a game 100% compatible with Ubuntu, the community will make it compatible with many other distros in some weeks! Problem solved.

Hey, Icaza, those programmers are lazy morons. Full stop!
Diego - sad to say but yes.. someone discovered this thread,  and you know, internet happens.  Next time I'll be a little more careful on distribution. :-)
I found the post somewhere else. The ways of the Internet are mysterious, Sriram :-)
By the way, there are some guys discussing here about Linux and Gnome development and they don't use and won't use GNU/Linux ever, yet they "have an opinion (bad and wrong)" and criticize?!? CRAZY WORLD!
I think that the Linux desktop is just beginning. I've used it since 1994. With Steam and other gaming platforms coming to Linux, the move of a lot of applications to browser-based platforms, and the locked-down developer environments of Mac OS and Windows 8, Windows Surface tablets angering and alienating a lot of hardware manufacturers - to me it seems a perfect storm for allowing Desktop Linux to take hold. I believe Ubuntu Desktop will be the consumer platform that will be the front face of this phenomenon. Everyone knows how slow Windows platforms get over time, well unfortunately MacOS is suffering from the same problem in my opinion.   
+Rodney Dawes has it right. The only bully in here is +Garry Lyndon

He just doesn't like it because this community won't let itself be bullied by people like him, spreading their lies, misinformation, and prejudices. Like most bullies he cries foul when others stand up to him.

Seriously, if he thinks Linux is arcane then he hasn't compared it properly with Linux (ah, but he admitted that... he actually stated he was completely ignorant of the facts and was just quoting 2nd hand opinions).

I've been a Linux user for over 12 years. The number of times I have been told to RTFM when I asked for help I can count on the fingers of one hand (and even then I have usually been given a direct link to the relevant FAQ entry or documentation page which answered my question). Only a sociopath would consider this rude or unhelpful.

I have also personally helped many people who have come to me for help or assistance and I'm happy to because, like most in this community, I know new users are an asset to the ecosystem. In general the open source community is one of the most helpful and accepting around. Of course we can be blunt when required: some people need that.

It should be noted that this thread is largely about development policy and covers some contentious issues that many people here have a lot of stake in. The arguments are "robust"as they should be as we are talking tech religion here ( a bit like trying to get a Java developer and a PERL developer to agree). This isn't bullying; it's discussion... it's something that grown ups do.
+Emmanuele Bassi given your rantings on LWN I'd say you have about ... negative 3 legs to stand on when it comes to talking about E or EFL. Some clue, patience and maybe having spent the minimal effort of asking a question or two may have helped, but these were far beyond you, so you'd rather rant. 
+Alan Cox thanks. :) And we'll stay to our roots. As such I'd say Linux has remained "niche" as far as the desktop goes. It's never "succeeded" (managed to get critical mass in the eyes of the consumer/market for desktops), and as such we need to remember, that being niche means we have to look after ourselves first

You can chase "Joe Average" or "your Grandma" users all you like, but they frankly don't care for the most part. You can't chase these and then alienate our core userbase - the power users who already have come to Linux for many specific reasons.

At least that's where E (and EFL) itself is staying - make power users happy, give them some eyecandy and use a minimum amount of resources to get there. A bonus result of that is that the embedded world takes notice as this is exactly what they need.
+Miguel de Icaza I have to say, if I heard the childish comments you heard at FOSDEM, I'd be just as displeased. While breaking of things (APi's etc.) happens, it shouldn't be something you launch into with glee. I do think that you have a small part of the picture right - breaking API to make it hard for 3rd parties to support your environment hurts you, but there is much more to this.

But I'd say our problems are much more than "API breakage", though that's part of it, and I don't think the kernel or it's devs are to blame here. Over the years at least they have done a stellar job of actually not breaking much (OK I can rant about changes to OSS vs ALSA etc. and /proc vs /sys and stuff moving over, but a lot of this is distros disabling the fallback/compat paths and moving over to new ones totally and thus breaking userspace apps - the kernel kept compat, but then packagers broke it).
considering that Miguel has been praising MS ecosystem for quite a while, left the gnome project, and have been focusing on c#/.net , I'm not surprise he is praising another proprietary software company.
This is his choice, his life, He has all the right to do so.

But for many people the Gnu/Linux its still about freedom. Perhaps he don't cares too much about freedom, and he is ok to be sucking os MS/Apple tits for life. But not for many.
I'm happy with gnu/linux, in matter of fact I'm happy with gnome-shell, I like it a lot!
So for someone, who's once was part of this craft, this kind of post is only spitting on his own plate, and it's quite unfair with other people efforts.

Actually the biggest unfairness in this history its been with the gnome project team, which has been working hard, and now are getting harshly targeted, in the name of someone who's been away from the project for a long time.
+Carsten Haitzler "At least that's where E (and EFL) itself is staying - make power users happy, give them some eyecandy..." — Well, well... There's a thing... As much as I like E (and I really do, been following it since 1999 or so), these days I expect a desktop environment to take care of unifying the look of applications.

I could probably try to live with native E's look that's unlike everything else, if it at least made an effort of loading Qt and GTK+ themes. But it famously doesn't even attempt to. No amount of bling and fancy animations is going to make me forget how ugly fallback Qt and GTK+ themes are.

So while you have a lot of fun doing low-level work on E and EFL, I have a lot of fun not even noticing the desktop environment of my current choice (GNOME3) and just doing my work in software that looks consistent.
John Kim
From a humble (and largely agnostic) Windows developer's perspective:  I don't think the Linux desktop is dead (I'm typing this comment on an Ubuntu VM), but it's definitely at a crossroads and there's ample opportunity lurking.

Windows is becoming a hairy and confusing ecosystem to develop for, developers in my circle have been privately murmuring been wondering why we don't make our break to *nix.  Joel Spolsky ( talks about the Windows ecosystem problems better than I can, and I think it's become worse in the 8 years since he wrote that article.

Given that Canonical makes Linux as easy to administer/use as Windows, if Linux can focus their efforts on just one sane and defacto-standard native API set I don't see why (an inexpensive) Linux desktop couldn't establish itself.
I have to thank Gnome, actually.  Had they continued to develop 2Gnome instead of 3Gnome I would never have switched to XFCE. Kidding aside, the comment about the user interface is telling: had gnome "only" broken their library users it would have been a developer revolt; but the contract between Gnome and the person using it was never stable.  Learn how to do something, then it quits working and you have to find a new way - if it's possible at all.  Even without KDE, or the less-common options, gnome on it's own would have provided three entirely different targets for developers.
John, that article was from 2004 based on his experience from the 90's.    I never understood attacking MS and Windows when they are probably one Windows is probably one of the most prolific, pervasive technologies known to man.  I don't think we need to attack MS.   Windows is still popular among personal computer users.  Who is to say that in a non-MS world that the personal computer would have had as much success.  They were popular 20 years ago and popular today.

Personally, I think that the Linux community (OS and Desktop) could have started out by making Linux pretty.  I am talking about visually as well as from an interface stand-point.  In the 90's and early 2000, your best bet for working with Linux was through command-line.  There was no point trying to get X or Gnome or whatever up and running.  You had to buy special (rare) graphics cards (lone nVidia or ATI cards) or spend days trying to compile X and add that custom fix.  It was pointless. 

And that was in 2000.  Win95 and Win98 were still popular.
+Carsten Haitzler There were more problems, that is for sure.   We had a golden opportunity the year Vista tanked, and we had that chance up to the time Windows 7 shipped.

Anyone digging at what we were up to in those years around our community will see how despite our best intentions, we, the community that cared about Linux on the desktop for consumers failed.

Perhaps someone will write a book about engineering mistakes based on what we were up to from 2006 to 2009: they years we could have gained some market.

Also, I did not blame the kernel developers for their API stability.   +Ingo Molnar also seems to have misunderstood what I said.   What I said is that the attitude of kernel developers towards kernel driver ABI stability was misunderstood by the larger community as an endorsement to "do it right or kill the code".   A hacker macho culture if you will, that bubbled up to the desktop, to system libraries, to system daemons and pretty much every other part of the stack.

And the finished product that was delivered to consumers was not the carefully crafted Linux kernel ABI that still allows Alan's 19-dickety-something program to still run on Linux.   But a fairly complex stack that involved hundreds of maintainers and apps were built against this sophisticated stack, not the kernel with libc.

During that 2006-2009 period, I wish the kernel guys had given the world a stable ABI for kernel drivers, it would have saved users and us a lot of trouble and a lot of anxiety.  I do understand the kernel's guys position on not having a binary ABI for device drivers.   But I continue to think it is a mistake, even if Greg KH has shown the world that he could convince major vendors to open source their drivers.   Greg won, but I think that we paid a high cost for this triumph during this critical period.

We live in 2012, and in 2012 the Linux Desktop is better than ever and it will likely continue to evolve as a fantastic workstation environment, but in my opinion, the momentum to become a consumer OS is gone and hence the title of my original post.

Linux is everywhere, and chances are that a new incarnation with the proper funding and the proper inflection point in the world will make a variation like Chromebooks succeed where the traditional Linux desktop failed.
+Miguel de Icaza You seem to be blaming the kernel's attitude towards internal interfaces, instead of the desktop developers from only paying attention to the kernel's attitude towards internal interfaces, and IGNORING the kernel developer's attitude towards external interfaces used by applications --- and then taking the wrong lesson by focusing on the wrong thing.  May I suggest that this is totally and wholly the GNOME developer's fault?   The fact is that we've always had the attitude of You Will Never Break Applications has been a hard and fast rule for decades, and we have reverted kernel changes when breakage has been discovered.

If the GNOME developers decided to ignore that part of the ethos, like they've ignored the power users who like 2D virtual desktops, and point to focus, and the ability to change fonts without having to use whacky registry hacks, that's their problem, and it is completely unfair to lay that at the feet of the kernel devs.

EDIT: s/click to focus/point to focus/
@Berlin: Well, Joel's blog post is a reaction to the change-over from the 90s to the 2000s and some of his views have panned out spectacularly. A big one: WinFX and the originally-planned Windows Vista never happened in the way it was intended.

In any case, I'm certainly not one to attack Microsoft.  As a user, I particularly don't care what OS I use as long as it's both affordable and not painful to use.  As a developer, I don't care which technology-platform I create apps for as long as the platform:
1). Has a future.
2). Evolves but is stable.
3). Cares about the longevity of both platform knowledge and apps written for the platform (i.e., backwards compatibility).  
(optional) 4). I enjoy developing for the platform.

What I find interesting is that--assuming these posts are any indicator--the Linux desktop seems to be shifting towards what made Windows successful while Windows is losing focus.

There could be some interesting times ahead.
+Miguel de Icaza I do agree that a stable driver ABI for the kernel would have been a massive boon. It hasn't helped in terms of getting 3rd party hardware vendor support which is critical. Also I do agree the kernel attitude was misinterpreted and then applied all through the userspace stack too, but the kernel isn't to blame, nor it's developers.

Reality is that the average consumer just doesn't care. They don't care about openness. From their view, windows is free because it's already there on their PC. They have little reason to want to try out some fringe OS like "Linux". We never gave them really convincing reasons to try it. Going by the number of people who have told me "Enlightenment made me a Linux user because of the eyecandy", I'm pretty convinced that if we gave people more REASON to use it other than it just being "free" or "open" it may have succeeded more. We needed to do much less copying of other OS's and desktops and less making it "familiar" and more "making it unique and attractive".

As such Linux stayed where I guessed it would.. As a niche OS. I love it. Millions of people do, but It hasn't gotten critical mass. There are a vast raft of reasons "maybe why". One BIG reason I'd cite is the dot-com bubble bust. This killed off a lot of funding that was beginning to gain momentum in giving Linux the real support it needed. Sure it survived, but only the areas that were already successful did (server-side). Desktop remained a hobby.
Jeeez... where is +Lennart Poettering when one needs him in a heated blame game?

(Lennart: thanks for your work.  I'm listening to music in my one-half-of-my-liver-and-a-kidney expensive studio audio gear through Ethernet thanks to PulseAudio, on my Beefy Miracle booted with system motherfucking d.  Wouldn't be possible without your work.  Thanks.  See you tomorrow at the journal sprint.)
The principle issue of the original article is in reference to it's title and meaning "Killing the Linix desktop". Since entities like the City of Munich in Germany, many sections of the Brazilian Government and umpteen others are busy implementing Linux Desktops (and Servers), it behooves Miguel De Icaza to clarify the false impresion he gives the general technology reading public since word 'kill' is totally incorrect and midleading if thousands of Linix desktop are thriving as opposed to 'dead'.
All the stupid technical bikering is missing the point.
Interesting thread.  I've used both KDE and Gnome as a user in the past.

Really, what I need is a "stay of my way" desktop manager.  I don't need pop-ups showing up every time I receive an email.

I don't need the app menu to show up when I'm focused on one task.

I've switched to Fedora because of what Ubuntu do with Unity...  I use Gnome shell and like it. But if I find something else that suits more my needs, I will surely make the switch.
+Torgjer Johansson I love C++ still, and am jazzed by C++11.  I wouldn't propose writing a kernel with it, but for application coding and even language implementation it is a workable choice.  Whatever opinion one may have of KDE, its use of C++ does not vastly matter.  IMO it's an advantage, but then I like Perl, so what do I know.  :)
HTTP is good between C# and C++ instead of p/invoke, which would seem better until you need viral markets or paravirtual emulation (with no wait states).
I think this rant fails in quite a number of ways.  First and foremost, have we forgotten all about the controversy that Mono created just a few years ago?  How that started creeping into Gnome and infecting it like a disease?

What we're forgetting most of all is that each and every user has different needs.  That is the beauty of Linux.

I don't know what Miguel has in mind for "success" of Linux on the desktop.  Yes, we occupy somewhere in the vicinity of 1% of all computer users.  We've steadily grown as an "alternative OS since the very conception of Linux.  Not only that, we ARE a niche market.  Linux is designed by and for people who want more control over their computers.

If we go the route of "one distro to rule them all" and remove choice from the desktop, I review that as complete failure.  What drew me to Linux (and FOSS culture in general) is the idea of choice.  I didn't have to look at a Windows or Mac desktop anymore.  I could make my interface function the way I functioned.  When we lose that ability, we lose our credibility.

I'm so happy for you that you prefer Steve Job's vision of computing over your own.  Great for you.  I personally find it counter-productive.

As for not recommending people to Linux except for some sort of ideological reasons, I find this to be completely untrue.  I have personally introduced several people to Linux in an effort to get their older hardware running again (why the FUCK are you slamming backwards compatibility again?!), or to get their newer hardware running more efficiently.

Case in point.  A friend of mine has a netbook.  This poor little thing was crashing incessantly under Windows 7 Starter Edition.  I moved her over to Linux Mint 13 with the Cinnamon desktop.  She has had no prior Linux experience. So far, the only complaint I've heard is that she can't watch Netflix on it.  However, she was so happy that she could change the wallpaper on her desktop.
Here's Linus Torvalds and Alan Cox discussing this very same point in person in front of Lennart Poettering:

ELCE 2011: kernel panel on the importance of users

But he doesn't get it, and GNOME 3 is proof that GNOME people don't get it either:

The absolute worst thing any project can do is break their user experience.
+Torgjer Johansson Of your arguments, I concede only language interop as a meaningful problem for C++ here.  But you're arguing theory when there the facts are right in front of us.  C++ has become widespread enough to work.  KDE lives. "Eppur si muove," as a famous scientist actually never said.

Ubuntu doesn't get to pick "the standard desktop"; AFAICT there isn't one.  Which is the fault, largely, of Miguel and the Gnome project....

PS: Name mangling changes over time are a PITA, but infrequent; and versioned symbols are adequate to cope.  cf libstdc++
I'm one of the various, normal, non-contributing (thus feel free to ignore me) Linux end users that recently switched to a Mac. I've been using Linux regularly for over a decade, but only started using Ubuntu as my primary desktop (or laptop, actually) about 5 years ago, and I loved it. I use Linux (Debian) on my servers, on my phone (Android) and I'd love to use it as my PC, but I got tired of dealing with the problems. It doesn't matter what my particular problems were (they weren't audio, amusingly enough) but my Samsung laptop just became a pain in the ass to use for various reasons and I went and got a MacBook a month ago. Sound familiar? I've heard this story various times before and I didn't think it would happen to me, but it did. Linux on the desktop, by its nature, will always be a bit of a gamble, but usually the benefits outweigh the drawbacks. Until it doesn't. In my case I blame Ubuntu in many ways - Gnome 3 might have been farther along had they not split effort with Unity. This caused more pain for end users (me) in various ways and generally sucked out my joy in using Linux as my desktop OS and sapped confidence that the situation was going to get better any time soon. So I swapped. I can't say I love OS X - trying to get a few open source projects up using MacPorts makes you appreciate apt in lots of ways, not to mention the ever-present spinning beachball. (Also how is it possible in a modern 'nix system for Chrome to crash the whole OS!?...). My point is that once some stability comes to the Linux desktop again, I'll probably switch back. That is assuming this isn't my last PC - maybe the comment above was correct and the future of Linux's GUI is Android?
Why is everyone assuming that the reason why desktop Linux has only 1% is technical? Seems to me that other factors such marketing, business strategy and consumer mentality are way more important then what developers did or did not do.
I would install coLinux with xen as the patch, by the way.
+Mirza Borogovac Why are you assuming what everyone's assuming? :) Technical folks tend to discuss things they understand better. Isn't it fair? :)
+Torgjer Johansson Sure, I understand that Mono is easy to program in.  I understand that it's nice to have a language that's compatible with .NET.

The real controversy wasn't that it existed.  It's that this language was creeping into Gnome as a dependency.  Nothing optional about it.  This piece of possibly litigious software was becoming essential to the Gnome experience.

Code your software in whatever the hell you want.  That's part of the Linux experience.  Just keep your languages that might require expensive licenses (that fly in the face of FOSS philosophy) out of our dependency list for essential software.
Will Hill
The other factor is called anti-competitive practices, +Mirza Borogovac, and those are something +Miguel de Icaza seems to ignore because it's his paymaster that does the worst of it.  Everywhere there's a gnu/linux success, Microsoft sweeps down to crush it in one way or another and they are constantly working to sabotage hardware.  Take gnu/linux on desktops at Walmart back in 2006 , woops that was quick.  They did the same thing to OLPC and netbooks just a couple of years later while Microsoft thrashed with Vista and Vista 7.  Any company that has anything to do with Microsoft is punished harshly for working with gnu/linux.  Companies that have nothing to do with Microsoft are harassed in court with bogus patents or copyright claims.  These things are continuations of the way Microsoft and Apple have always dealt with their competition, and we can go back to look at horror stories from a decade ago, like SCO, Caldera, Correl, Netscape, and on and on.  

Blaming Linus and other developers is a double insult.  It's an insult because those developers shoulder the costs of Microsoft's crimes and it's an insult to think people are too stupid to remember the source of their problems.  Way to blame the victim, +Miguel de Icaza.  

Here are a few counter examples to the assertion that non free software makers can't deal with gnu/linux diversity without an ABI.  Skype.  Remember that?  It worked great with gnu/linux and ran repositories which I know worked well with Debian and derivatives.  That was before Microsoft bought it, now I hear it's broken.  ID Software has always tinkered with gnu/linux and continues to do so but now we have Steam and other good stuff coming along, if things like Play Station were not good enough.  I also work with a radiation therapy treatment planning system that works on Fedora while Windows systems are buggy at best.  Oh yeah, CT/PET machines have also moved to Red Hat.  Rock Stable.  So that's a range of software from frivolous for the masses to demanding for a narrow band of technical users, and it all just works because nothing is hidden with free software.  Non free developers can, if they please, make things work on a few distros and reach almost everyone interested.  Again, this is something that's been going on for more than a decade, as native versions of Word Perfect and Corel Linux showed us back in the late 90s.  Strangely enough, when you step away from x86, you find wild linux success in the form of Android, again without an ABI.  It's the freedom that makes it work.  Loading it up with buggy, non free software has a tendency to make "Linux" suck.  
+Will Hill I didn't see any official announcement in that link. An official link would have to come from Microsoft or Miguel. So now you appear to have linked to 2 conspiracy theories in the same discussion- it's cracking your credibility.
In few years, outside of the dev community, there wont be such a thing as a consumer desktop as we know them. Projects like +CyanogenMod based on Android AOSP have a good shot at becoming the Linux distros/desktops of the future. Important for these projects to learn from past attempts. 
okay, folks.. closing comments on this.  Feel free to discuss on the many postings out there on reddit/slashdot/etc.