Slashdot Mirror
The Linux Identity Crisis
Jayze Calrtini writes "From an article from ZDNet:"If you've been following the current rift in the Linux community between Linus Torvalds and his minions squaring off against Con Kolivas and the mainstream Linux fanatics, you probably know that it's getting quite heated.
You also probably know that these two entirely different ideas could create three possible paths Linux can take for the future: stay geeky and appeal to the advanced tech guru in all of us; go mainstream and leave the advanced functionality and reliable kernel behind to compete with Microsoft and Apple; or face a "civil war" that could lead to total Linux annihilation."
I vote for total annihilation.
I mean, with Vista, who cares about Linux anymore?
The opposite of progress is congress
Two outcomes... Linux gets better or Linux dies. Either outcome is acceptible and should be to any other OSS "believer" as well. Survival of the fittest and all... even if the fittest isn't Linux.
Sounds like another storm in a tea cup. The linux world has had more flame wars than not, and will continue to do so as long as it exists. It's one of the characteristics of a democratic system that people have arguments. The "total annihilation of the linux world" is a load of incendiary exaggeration. Typical slashdot "editorialism", I guess...
Daniel
Carpe Diem
Come on over to *BSD. We're the 'big tent' OS. Room for everyone.
Don't like the direction the kernel is going? Branch the kernel and call it MyBSD. Whatever, no one is going
to get pissed.
Linux folks take themselves WAY too seriously, and besides, *BSD has a 'cool' factor with the chicks that
Linux will never have. You should see the honeys flock to me when I sport my FreeBSD tshirt.
Come on in to BSD, boys, the water is fine.
desktop improvements do not need a revamped kernel. I really don't know where this idea came from.
Both gnome and kde have their irritating features and this - IMHO - is where the problem is.
The article--no, make that rant--has nothing to do with the debate between Linus and Con. The author somehow thinks that this technical debate about the kernel's workings has something to do with "Linux" desktop usability. The author clearly does not understand that there is a difference between the Linux kernel, the thousands of programs that comprise a Linux distribution, and the distributors who glue all this stuff together. He says Linux shouldn't "go mainstream" (here I guess he means distributions) and ignores the fact that Ubuntu can go mainstream while Gentoo can stay geeky.
Total waste of time; prevalence of this crap on Digg is why I stopped reading it, and now Slashdot isn't too far behind it seems.
Penny - plain text accounting
init 11 - for when you need that edge.
One part of me likes the first two ideas. I mean, there could eventually be a Windows killer distro out there. And at the rate things are going, Ubuntu seems to be the likely candidate. On the other hand, Linux does have a place with hardcore geeks out there who like to tinker and tune the kernel.
A second part of me is wondering why we all can't get along. Linux isn't going to be annihilated. Even if Torvalds were to walk out in front of a bus tomorrow, development of the Linux kernel will not cease entirely. Businesses have too much riding on Linux for it to fail. I could be wrong; but I highly doubt the doom sayer's claims.
The game.
What really pisses me off as far as Colivas camp is concerned is that they equate 3D games smoothness to desktop performance and keep on quacking about "desktop linux performance". Their stuff has nothing to do with it.
It is just one tiny facet of desktop linux. Further to this, in order to demonstrate any of the performance you have to throw in two big unknowns - a binary only driver and a card without a fully disclosed and known specification.
Self-serving benchmarks for 3D game on local machines should not be used to claim superiority in all desktop linux tasks period. In fact they should not be considered at all at least until something comes out of the recent ATI and Intel spec disclosures. When non-binary 3D accelerated drivers become widely available there will be a point to start benchmarking towards 3D performance and smoothness. Until then this is a complete waste of everyone's time.
Baker's Law: Misery no longer loves company. Nowadays it insists on it
http://www.sigsegv.cx/
Why can't it keep the nerdy, hackable kernel and go mainstream at the same time? I though that was the reason why we have different distributions; obviously not everyone's going to be happy with Gentoo, luckily the casual user has Ubuntu and Linspire, and us network admins have our server distros. Do these people really have this George Lucas kind of power over the things they have released to the public, or is the community in the driver's seat enough to keep it working for everyone? I feel like it certainly leans more to the latter, although I guess I'm pretty far removed from the development process.
lkml has always had robust arguments bounced about. This is not new, but new people are reading it all the time and sometimes it hits the mainstream. TFA is mainly not about lkml flamewars, but about a review by Walt Mossberg which might be important to a certain readership in the USA. He isn't very important to readers in the rest of the world. I read the review. It was fairly balanced, he found good points and areas for improvement. The fact that he reviewed it at all is more significant than any findings or conclusions he made. I am amazed at the number of meta-articles about this one review that I have seen. Journalists - do your own flipping review. Don't write articles reflecting on someone else's reflections.
Maybe the real question is, how to get Linux
developers to play the game professionally.
Do we really need more incomplete, undocumented,
fail-disable, unverified software? The issue
of Linux success is more a question of when will
Linux software become polished, real end user
value? Why do I spend so much time hacking
around fixing scripts that should have been done
right before they were posted? Why am I re-writing
resolv.conf after re-boot to replace the incorrect
(gateway address, not nameserver address) mismanagement
in some layered, undocumented fork from network?
Com'on guys, the field's 100 yards. No touchdown until
the job's finished. We don't need another 'final coding
left to end user' version of anything.
Kevin O'Kane http://www.cs.uni.edu/~okane/
LINUX IS THE KERNEL.
Ubuntu is a distro comprising of a linux kernel and userland tools/libraries. Why would going the "ubuntu" route would involve any changes in the kernel is beyond me. Ubuntu is nothing more than a well engineered collection of userspace tools that makes the PC useful, it relies on the Linux kernel to manage the system.
In short, you can appeal to the "mainstream" [also known as the dumbification of society] and yet keep a technically impressive kernel behind the scenes.
Someday, I'll have a real sig.
I don't know what he is basing this crap on, like that Linus thinks Linux shouldn't go mainstream. Linus works for the Linux foundation that "promotes, protects and standardizes Linux by providing unified resources and services needed for open source to successfully compete with closed platforms."
Next article, please.
I'm sorry if I haven't offended anyone
go mainstream and leave the advanced functionality and reliable kernel behind to compete with Microsoft and Apple
Would help if the author knew what the trade-off was. Servers are simple. They maximise throughput fairly. Then there's desktops, which are supposed to remain responsive to mouse and keyboard and audio events even under high load. The latter is more complex. It is the one with the "advanced functionality", and it loses reliability in the process.
There are geeky people in both camps. Geeks who want a server, and geeks who want a desktop.
The geeks who want a desktop want advanced functionality at the expense of reliability, and since the entire hypothesis of the article falls over in the first paragraph, I'm not sure why I bothered to continue reading
Then it continues with crap like If we want unstable systems, we can buy a Windows box.
NOBODY, not even windows users WANT an unstable system! I want a good opensource system that will run reliably and efficiently on my desktop. By the same logic I could say "if we wanted a reliable server, we could just use BSD".
Con Kolivas wrote some nice patches. I'm still yet to see if the CFS is as good.
I wrote my first program at the age of six, and I still can't work out how this website works.
Exactly. It is tricky but nearly always possible, through good design, to create a system that works for different skill levels of users. It can be easy to use, easy to start learning, easy to install, with functionality that is easy to discover, and still be highly reliable, customizable, and efficient for people who use it all day every day.
Light cup, beer drink, thin so chain, neck turtle fat, man I won't say it again
This looks like vaporous hype designed to try and make linux look unstable. Didn't Con Kolivas say last july he's leaving linux kernal development?
How did this make the
"We are all geniuses when we dream"
- E.M. Cioran
Maybe. Maybe not. It depends.
Either way, it doesn't matter and we win. If the kernel doesn't fork, then probably some kind of compromise has been reached that brings the best of both worlds. If the kernel does fork, we get two independent projects, perhaps each geared at different requirements.
This has happened before. Firefox started as a fork of Mozilla Seamonkey. The needs of embedded developers have spawned small Linux kernels like ELKS. Ximian started as a GNOME fork that eventually was merged back in. Then there's egcs vs. gcc, and so forth...the list goes on and on.
In the end, the community wins. We get better code, and in some cases, we get new projects that meet specialized needs.
My blog
We're all doomed! Doomed I say!
Am I just jaded or does this seem a wee overdramatic? Total destruction of Linux? Civil war? Yeah.
Unbreakable toys can be used to break other toys.
I think I go for the fourth outcome:
Really good make-up sex between the parties.
If you mod this up, your slashdot background will turn into a beautiful sunset!
That's quite a leap to make without giving any evidence at all. The article first mentions Con Kolivas' spat with Linus as if that is some kind of indicator of Linux's future when it means very little. It makes the assumption that CK's scheduler was more techinically advanced than Ingo Molnar's scheduler. That isn't the case. I don't think the author understands the reasons behind Linux choosing CFS over SD. It was more about maintainability than anything else. It was a decision that took into account long term issues instead of just short term emotions people had for CK and his scheduler.
When did this become a Republican/Democrat issue? Maybe I'm showing my bias here but how in the hell is the "liberal" wing in Linux all about making money? Isn't that the domain of Republicans? If you think that Linux really is split into a liberal wing and conservative wing the comparison would make more sense if the roles were reversed. Conservatives want this to be based about money and the free market. Conservatives would rather have corporations like HP choosing the direction of Linux based on their needs. Liberals are more worried about their rights with the software and abuses taking place by the corporations.
Even without taking the phoney political comparisons into consideration this article is an anti-Linux fluff piece with no meat at all. There is no critical thinking involved at all. It's purely an opinion without any facts to back it up. I wish garbage like this would stop showing up on Slashdot.
Time makes more converts than reason
I actually agree with Con's assessment that Linus' refusal to accept these performance enhancements shows that the desktop is not a priority in the core Linux kernel, just as embedded devices are not. What I don't understand is why there's so much controversy over creating a kernel variant to address this. It's been done before, and these variants seem to coexist more or less peacefully with the core. You have uClinux handling embedded devices, while SELinux has a following among the security community, RTLinux does realtime stuff, and so on. Why should a "DeskLinux" with Con's performance enhancements be any different?
Just another article spreading FUD by making it appear that some internal rift will cause the downfall of Linux.
This whole thing scheduler issue and Con thing regarding focus on the desktop is rather funny.
This is linux we are talking about here, don't like the direction feel free to change it. If no
one will listen patch your own kernel and call it my ultimate desktop edition. It certainly would
not be the first time a focused distro has been developed.
Bottom line, there is no rift in the community somebody cried because there scheduler got beat out. I assume this is because it did not make the cut for some reason, however if I wanted to run Con's scheduler I would just patch my kernel and run it.
Got Code?
There is nothing to see here. These sorts of infights are not only common, they're necessary, due to the very nature of the project. Competition means survival of the fittest, and these fights are the best method for weeding out the strongest code solutions from the ho-hums. Best we fight amongst ourselves, for the world itself wants to crush us in it's fight for mediocrity! But the moment an external force tries to pick on onef us, we unite into one gigantic geeky mass. We can pick on each other, because we're family, even the BSD guys. But nobody else has that right!
Karma Whoring for Fun and Profit.
I don't see the point. Every problems needs a specific solution and there's enough room for both solutions.
The article confuses Linus Torvalds' Linux (just a kernel) with distribution.
No matter what Linus thinks, there are still out there very geeks oriented distro like Gentoo and Slackware with "let the user configure himself everything" in one end of the specturm and Ubuntu, complete with its "means 'I can't install Debian' in african dialects" types of joke.
The TFA is just a meaningless rant.
For me the two outcomes are without linux dying, because each variant is fittest for some specific usage pattern (geek vs. joe 6pack). And thus both outcome may happen simultaneously.
"Sufficiently advanced satire is indistinguishable from reality." - [Tips: 1DrYakQDKCQ6y52z6QbnkxHXAocMZJE61o ]
The article is utter crap.
It confuses Linux (the kernel) and the CK/CFS spat with the various distributions of GNU/Linux, Gnome and KDE and their usability issues for non-techie types.
There is no risk of a "civil war" and one, certainly, would not bring total annihilation. At most, there would be the threat of a fork and some distros offering a CK patched version of the mainline kernel. I would like to be able to start up my machine with a choice of schedulers or, better yet, as someone pointed out, starting my servers assigning different schedulers to different processors according to their workload.
But all of this has nothing to do with how grannies use their Linux boxes.
http://www.dieblinkenlights.com
What in the Kernel (the only place really that Torvalds has anything to do with Linux) makes it impossible for Linux to enter the mainstream... Maybe some slight license wranglings about attaching proprietory drivers to the kernel (this seems to be becoming less of an issue anyway). Some of the current(?) scheduling stuff might be relevant too, but these are _very_ minor.
I do not consider myself a nerd of geek. I use Linux because it works for me, because I avoid vendor lock in, because it is easier to admin and secure.
The FUD machines are still running at full speed and spewing loads of irrelevant lies, damned lies, statistics, and general crap. It's done because it is rather effective on the uninformed masses of managers who have little depth of knowledge and simply want "safety".
Seriously, the Linux kernel is in no danger of imploding any time soon. The community is rather strong and resilient. Really, the big difference is that the development process is visible, as opposed to proprietary software houses where these conversations are inside the walls of the company. The debates we're hearing about are a normal part of development and will eventually lead to a solution that works for everyone.
Desktop Linux vs. "Server Linux" is a total non-issue at the kernel level. The userland tools and interfaces are far more important, and really the only real roadblock right now is a few hardware manufacturers' active resistance to working with free software. This isn't so much a conspiracy to lock out certain operating systems, it's just a way to manage their obselecence cycles to ensure future sales. After all, if customers can keep using that printer until it actually wears out then quarterly profits will see no replacement sales bump when the next Windows release comes out.
This resistance is starting to fray around the edges, and we can see the evidence in AMD/ATI's starting to open up chip specs and Dell's entry into the desktop Linux market. It's beginning to become a non-viable business model to actively impede interoperability with open source software.
Mainstream a Linux desktop, and by mainstreaming, I mean make it commercial. Make it so Joe Notageek, and his grandmother, can install it with less clicks than it takes to install Windows. Provide apps for it.
This is a vastly overblown issue. Normal people don't install OSes. Normal people don't even understand what an OS is. They buy computers, not OSes.
This is the biggest difference between Joe Average, and geeks. To a geek, a computer is a collection of (mostly replaceable) components. To Joe Average, it's an appliance like his microwave, iPod or DVD player. How many people do you know who upgrade the coil in their microwave ?
You should check out www.groklaw.net. There has been a lengthy article just recently about the latest anti-Linux FUD campaign. Now that SCO is bankrupt and nobody believes anymore that there is any Unix code copied into Linux illegally, they had to come up with something new. The new campaign is: Linux is self destructing! Sources are the usual suspects, like ZDNet in this case.
However, if you think about it, there are several thousand Linux developers, and with that many developers, occasional arguments are unavoidable. The same arguments happen within Microsoft software development, except that you don't read about them on some kernel development newsgroup, and the press doesn't pick up on it.
The article is from ZDNet. The author probably stumbled upon kerneltrap for the first time and thought, "OMG! There's a real *war* happening here! This is news!" -- not realizing that the "war" was business-as-usual.
Another thing the author doesn't seem to realize is that Linux code (the kernel) is forking all the time. It may be support for real-time embedded or support for MMU-less processors, etc. The point is, people experiment, discover something interesting (fork), then try to get the interesting part back into the mainline tree. Happens a lot. Let the code fork in a big way? It will later merge and improve, yet again.
I recommend to anyone covering geek news: Be a lurker for longer than ten minutes and try harder to understand what you're writing about. From the article: "Much like Republicans and Democrats, Linux is dominated by two factions with entirely different ideas." In psychology I think that's called "projection".
I find it amazing that people that can't understand the different between source and compiled code come talking about the linux kernel and how it should be split, forked, etc. Maybe we should lock Linus an Kolivas in a dark room, each one armed with a knife, and let them decide in the good old fashioned way what's best for the kernel. This is a slightly better solution then forking the kernel.
First, Kolivas is free to create a kernel for him, just setup his GIT server and he's done. That's what *free* software means. And *ANY* distro is free to use his kernel.
Second, what's all the fuss about the scheduler ? Damm, it works FINE, Linux's problem is NOT the scheduler. It's the lack of some basic features, like MP3 playing, AVI playing, etc. Yeah, I KNOW that this is because of commercial rights and such, but the average user doesn't and doesn't care. Computers are supposed to work out-of-the-box, if it doesn't then it's broken.
Third, what's the point of forking the kernel ? Just compile it with the right options and you're set. The source code can contain dozens of different schedulers, you use the one best for you. Discutions like this **ARE ** FUD and I think Linus must find all this very amusing, because it's a buch of people wasting energy in a non-issue.
http://www.groklaw.net/article.php?story=20070921112733615
Really, this supposed infighting doesn't exist, and having these articles on slashdot just helps us be part of Microsoft's mouthpiece. Even if there was a lot of infighting among the kernel developers (there isn't, by the way - not in the sense of a civil war causing total annihilation), all you'd get is a fork and people would move in that direction. I believe that all these articles about Con Kolivas's scheduler are part of this FUD machine and are blown way out of proportion.
For the curious wanting to understand a bit better about Linus's tree not being the be-all and end-all, check out this gentoo kernel page that talks about some other branches and unofficial trees.
... that really does not exist?
Like the BSD/GPL licensing issue that was used in a failed attempt to create a problem that did not really exist.
Matt Dillion of Dragonfly BSD clairified it... There really was no issue or concern...
Whats this gotta go this way or that way crap now?
There is no spoon....feeding..... there is forking for the masses...
So fork the fool wants to creat a problem that really doesn't exist...
I fail to see the any point at all in your entire posting. This article seems to try to make something out of a dispute amongst a bunch of very talented but very stubborn geeks over an algorithm that's inside a kernel that's inside most Free Software operating systems. This specific dispute has ABSOLUTELY NOTHING to do with the appeal of Linux-based OSes to the end user. Sure, perhaps one scheduling algorithm works better in high-performance clusters and servers, whilst the other is more "well-rounded" and thus more consistent for desktop use. So what? What does that mean to "Joe NotaGeek" anyways? 3 FPS faster animation on a 3D shooter game? OO.o opens 200 ms faster?
.deb vs. .rpm...now THOSE are issues MUCH more visible to the end user. The outcome of THOSE kind of issues can affect whether I can install a package, whether it runs, whether it breaks if my OS is upgraded and whatnot. This scheduler debate is an academic debate--it's not like they're making wholesale changes to the kernel APIs. The moves from kernel 2.0 to 2.2 to 2.4 to 2.6 were far more significant. The move from a.out to ELF was huge. While the content of your post might be something that should be discussed it has nothing to do with the debate over the scheduler.
Sorry, I fail to see the connection between the "great scheduler debate" and "the future path of Linux-based OSes as we know them". Plugging in another scheduler is not an issue anyone but kernel geeks will get emotional about. Period. "Libre" free graphics drivers, GPL versions, design changes in GNOME and KDE, uptake of standards like LSB for distribution-neutral packaging,
It's like arguing over whether Ford could make the Fusion more appealing to the biggest market segment of car buyers by employing an Intel 8049 or a Siemens 80535 micro-controller to run the fuel injection module in the ECU. For 99.99 percent or more of potential buyers they WON'T EVEN CARE and wouldn't notice the difference. For the two dozen people in the world who DO care they'll have their little spat and move on, and perhaps both alternatives will exist in the market for a long time...and only those two dozen people will notice. It is the same thing with the kernel. Even if the scheduler was to fork and some distros use one and others use the other, neither fork is going to break compatibility with the other and consumers won't notice and won't care. It's all GPL anyways so they'll look at each other's stuff and make sure they both remain compatible
at the API level, and we won't have the big mess the commercial UNIXes had in decades past.