With Linux traditionally coming in many, many flavours, a common call among some Linux fans – but mostly among people who actually do not use Linux – is to standardise all the various distributions, and work from a single “one-distribution-to-rule-them-all”. In a recent interview, Linus Tovalds discarded the idea, stating that he thinks “it’s something absolutely required!”
There are probably hundreds of Linux distributions to choose from, and this allegedly poses potential users with a choice they are unable to make. To make matters worse, distributions are usually incompatible with one another, leading to a situation where you need to hunt down applications specifically packaged for your distributions, or to situations where a how-to applies to one distribution, but not another.
Linus Torvalds is pretty clear on the subject. Let me copy his reply verbatim:
I think multiple distributions aren’t just a good thing, I think it’s something absolutely required! We have hundreds of distros, and a lot of them are really for niche markets. And you need that – simply because different markets simply have different requirements, and no single distro will take care of them all.Of course, people then often say “well, do you need multiple distros for the same market” when they think about the normal desktop market and just look at the whole issue of having openSUSE/Fedora/Ubuntu all in that same space. But it really isn’t that different – you still have the distributions looking at and concentrating on specific issues, and you do want the competition – and letting the markets decide which issues are the ones that really dominate.
In addition, having multiple players just keeps everybody honest, and allows you to compare them. It may all look a bit messy and complex, but I’d much rather have a multi-party system over a single-party one. Even if it’s more complicated.
I am inclined to agree with Torvalds for a number of reasons. First, the market for desktop Linux distributions has more or less settled on a relatively small number of distributions, such as Fedora, openSUSE, and Ubuntu. Sure, some of these have derivatives, but they are all compatible with one another, so you can’t really claim that Xubuntu is a separate distribution from Ubuntu, for instance.
An additional problem with settling on one single distribution is the issue of desktop environment, probably the most important piece of software in a desktop Linux distribution. KDE, GNOME, and Xfce all serve different types of people, and their respective applications tend to those different types of people as well. In other words, you would already need a number of different distributions to cater to these very different groups of people. A KDE user generally doesn’t like using GNOME, and vice versa.
The competition argument Linus puts forth is also not to be underestimated. It often happens that a single distributor takes it upon themselves to work on a specific element of the Linux desktop that needs work, and then this work gets adopted by other distributions. This is a direct result of different distributions targeting different user groups, and thus having different priorities. If you standardise on a single distribution, this dynamic is killed off, seriously hindering innovation.
However, that is not to say that at a lower level, the Linux world wouldn’t benefit from more standardisation. A few areas that come to my mind is installation locations, a stricter adherence file system layout specifications, and the obvious hot iron package management. Especially that last one usually triggers heartfelt debates, but it is undeniable that a lot of effort is wasted on duplicate efforts, because all major distributions spend a lot of time packaging the same software over and over again in different package formats.
When I first started using Linux (somewhere at the start of this century), I used to be an advocate of having a single, unified user-oriented Linux distribution. However, my years and years of experience in the Linux world has taught me that Free software developers thrive in a world of competition and freedom of choice. Imposing Windows-esque and Apple-esque policies on Free software developers is not only pointless, it would only work to the detriment of the Linux desktop.
I agree with you on the whole, having one (or few) would be pointless, in addition to impossible given that the GPL, and most Free Software licenses, allows it.
Today the short answer, for the most, is Ubuntu. But that is always subject to change, which is a good thing.
Yes, it keeps Canonical honest and on their toes. They have to innovate, integrate and polish like the best.
If they would stop people would run to the distro de jour without hesitation.
Linux users are Linux users for a reason, not because of inertia (windows) or faith/marketing (OSX). (OK OK .. It sounds like flamebait .. but it isn’t.)
And Linux users are using it only because they are blind zealots?
See how stupid my comment sounds. That’s how stupid your comment sounded. This is not Slashdot, you can do better.
Being a blind zealot is a reason too.
I stand by my word.
Most Windows users keep using Windows because that is what they know.
Most of todays Mac users started using Mac because of Apples superior marketing or the Apple cult.
What about those who tried X, but still went back to Y?
(By try I mean used it for at least a couple of months and gotten to know it well enough for their use.)
Are they only going to Y because it’s what they know?
…because clever marketing and fanboys got them back to Y?
…because they are blind zealots who tried something else to they could say “I tried it and and it Y is better!”
Edited 2009-02-05 17:42 UTC
Yep and as someone who doesn’t particularly like ubuntu (nothing wrong with it and i do still respect it), I’d still be happy if software writers targetted ubuntu. Other linux vendors could still easily make sure any ubuntu software ran fine on their distro – and indeed 99% of software written for ubuntu WILL run unmodified on most other mainstream linux distros.
We have a standard base, standard libraries, standard filesystem structure (within reason) – so writing linux apps is already pretty easy. I think we’re heading in the right direction.
It doesn’t please everyone, but if you’re really into Linux then you probably like the way things are going more or less.
..and it wasn’t all that long ago that Ubuntu burst on the scene with the same comments saying “oh no…not another one!”
Survival of the fittest. If a distro comes along that people connect with, it will find its’ niche. If not, it will wither and die. As it should.
Ubuntu works best with most hardware and is very straightforward. I’ve been converting Windows users for months – those people without specific needs like AutoCAD’s and Photoshop’s – and they’re loving the experience.
I tried others and just failed. That’s the beauty of Ubuntu. It actually works!
There are a lot for sure! Check out this list.
http://www.lampwebsitedesign.com/index.php?linux
…is choice.
And choice here doesn’t mean I have to select every single small program or even edit the code to fit my needs
Because of what? Because many others do it and are allowed to share their work. By that, they add real value to the software itself.
It is important to continue with this work, and it is far from being as redundant as it may seem from an outside perspective. First of all, the “one fits all” product doesn’t exist. It is fine that there are mainstream products, but also important that the niches are filled as well. Second, redundancy based on competition is how market economy works. And the cases where market economy doesn’t work well — like creating monopolies — are very well prevented by the free software principles (licenses).
Edited 2009-02-04 15:26 UTC
I think no one argue that there should be only one distribution…but that doesn’t mean it should be 1000 of them either.
I am one of those that don’t care about the amount of distros, but think they should focus more on standardization across distros. LSB is a good initiative, but it should be more adapted. Furthermore, it should cover more and have some sort of carrot for those distros that fully implement it; like browsers work hard to achieve acid-test compliance, distros (at least those who are not a specialized niche product) should work hard to be as compliant as possible to LSB.
…and the #1 focus of LSB should be a _required_ package handling framework in which all the distros can build their own special package handling on top of. My suggestion would be to include perfect support for multiple versions of same libraries
jesus, not this absurd discussion again.
It is not POSSIBLE to restrict the number of distributions since the license allows people to create their own. Why not also suggest that we standardize on movies and music? It’s an outrage that just anyone can start their own band or just buy a camera and go out and shoot movies. There are too many choices! We need to restrict the choices for the good of everyone.
You know, just like how they did in the old eastern bloc back in the day.
Edited 2009-02-04 15:34 UTC
So you agree with the interviewee, article and rest of the comments then?
Not that you don’t make a valid point, I just found your comment to sound like a response to an article claiming that all Linux based OS should be unified rather than responding to an article and comments that universally support distro individuality and free market competition.
No, i’m talking about the endless, moronic discussions that this news item will spawn.
There are too many distros and they are all the same
There are too many distros and they are all different
There’s just too many, I cant chose
etc etc
true, it’s sadly inevitable that someone is going to show up breaching the word of the “unified one true OS”. I think your comment just came across as premtively heavy handed after the rest of the comments agreeing with the interviewee and article.
Here’s both us hoping it doesn’t follow the usual downward spiral though.
Sadly, we’ve been proven wrong.
Yup. This is my giant smile at being proven wrong on this one though.
I have been using Linux since 1993, back before it even had a TCP/IP stack built int, when I found it while looking for a free way to do C development at home. I would personally be happy if every computer in the world ran Linux.
BUT EVEN I AM SICK OF IT.
The fact there are endless distros is perfectly fine, and even welcome by me. There is something for every need out there.
BUT, the fact that every single one uses a different layout of the file system, puts configuration settings in different places, uses a different packaging system (and makes it impossible to use packages across distros), its TOTALLY FRICKING INSANE AT THIS POINT.
It almost makes me wish PC-BSD would have gotten their licensing act together sooner so it would have had a chance to be what Linux is today.
There are lots of BSD distros, but they pretty much all use the same system configuration scheme, can easily install packages from one to the other etc.
The only hope is that we are now down to pretty much only TWO MAJOR types of Linux distros. Those based on the redhat/fedora way using RPM, /etc/sysconfig etc… and those based on the DEBIAN way.
We should just pick one and give it up. Either debian needs to get a clue and go with RPM etc or Redhat needs to move to using debian style stuff.
Until this happens, Linux is going to be about 1/100th as popular as it could be given a chance.
False. And if you have used Linux since 1993, surely you know that.
And this is what most people get wrong when they whinge about the length of the various lists of Linux distros. The vast majority of distros fall into just a few “families”, and are, by and large, variations on a theme. Furthermore, there are really only a few of all those distros that a significant number of people care about. It’s not as though software providers have to target “Gelugpa Linux”. If the Gelugpa maintainers want Gelugpa to be supported by ISVs they will ensure compatibility with a distro that the ISVs care about.
One thing I find very strange about these discussions is that there seem to be 2 major schools of complainers. Those who complain because all the distros are supposedly different, and those who complain that all the distros are really just the same.
On the topic of “choice is good, choice is bad”, I would walk a middle path between the two. Lack of choice is bad. Total chaos is bad. Having a limited number of major families of choices, with further refinements available, to those who look, within those families are what is good. To a great extent, that is exactly what we have. Though one could reasonably quibble over the details of what numbers would be optimal.
Edited 2009-02-04 16:31 UTC
But the tarball is your universal package. Heck, you can toss a tarball on a BSD and have minimal issues if any getting it installed.
I think competition in the package management space is as important as it is between window managers or complete distributions. Aptitude/Apt-get/.deb is great but urpmi/.rpm is also fantastic (not to mention “rpm” done right).
If you want a single universal package standard, there is always LTS and you’ll also get hardware specific builds along with the latest build versions and a truly customized software stack.
Agreed, but I suspect that is already happening to some extent. Essentially, it’s the deb system for home users and the rpm system (Red Hat or based on mostly) for enterprise. In the home sphere, Ubuntu + Mint + Debian must far, far outweigh OpenSuSE, probably the most popular rpm-based distro for home users. And if OpenSuSE catches a Novell cold, quite possible in the present economic climate, then the rift will become even clearer.
I don’t see Debian “getting a clue” partly because the project isn’t set up like that. It’s there to provide a free and open OS regardless of the pressures pushing IT one way or the other. Distros based on Debian, like Ubuntu, can then use this huge base to push in whatever direction they wish to. But Debian doesn’t and can’t afford to be seen to either, imho.
Besides, Ubuntu clearly does have a clue. The catch, imho, is a chronic lack of resources (such as money): they have to do what they do on about 1/1000th of the resources of a major corporation, so it’s hardly surprising there are rough edges. Chicken and egg of course: rough edges = relatively few users compared to Mac or Windows = small resources = rough edges, etc.
As for Red Hat, they are sitting pretty in the Linux world I’d say. If you want free and rough edges, you have Fedora. If you want free and no or few rough edges for the server, you have CentOS. And if you want the second but with paid support, you have Red Hat proper. Each one pushes users up the chain to the other.
Even so, a couple of weeks recently with Fedora 10 was a surprise. It has a very poor range of configuration tools compared to Ubuntu or SuSE, ihmo, and very little effort has been made to steer users by way of details in readmes, example files, helper scripts and the like. If Fedora 10 is the best distro for new or inexperienced users, I’ll eat my hat.
So one of the things holding Linux back isn’t just lack of resources, imho, it’s a reluctance by the distros to stop trying to be all things to all men and instead be much clearer about exactly who their distro is for. It’s fine for there to be thousands of different distros, for sure. But if they are all run by geeks unable or unwilling to think themselves into the head of a non-geek user – someone without the time or inclination to poke around under the bonnet, and someone who is going to dump you if you try to make them – then Linux will continue to be about 1/100th as popular as it could be.
What Linux really needs is a charismatic, natural-born popularizer. Which could just as easily be a gadget or game as a person.
Edited 2009-02-04 20:26 UTC
It doesn’t help anything if enterprises use RPM and endusers use DEB. As long as the distros themselves use different versions of libraries we are still screwed, and there will be dependencyhell between distributions. That is why library-versioning is so important in a package manager.
You are really asking for the argument not to be settled by force majeure. That is, by 2-3 distros becoming so popular they are taken to be the norm and anyone who runs a distro that deviates from them is told “tough luck”. Of course even that concentration of powers wouldn’t solve the problem but it would be closer to a “one-library solution”.
OTOH, there’s every indication that this argument will be settled by force majeure. Partly because, so far, no one has come up with a compelling answer as to why it shouldn’t be.
I am sick of purists and busybodies telling other people what to do. The law does not forbid and license allows it, so people can and WILL build new distributions as they see fit, no matter what somebody else might think about it.
…it’s about having a reference-distribution like the BSDs have
My understanding is that BSDs differ by kernel where Linux distros differ by user space. The BSDs may draw from the same package library but the kernels can be very different.
It’s kind of ironic that he wants multiple distributions, but “people who argue for splitting desktop kernels from server kernels are total morons, and only show that they don’t know what the hell they are talking about.” http://kerneltrap.org/node/14019
Well, there are certainly different compile options that would benefit a desktop vs a server… that is perfectly fine, because the usage of a desktop vs a server is entirely different…
The whole is still the same codebase though, so its sort of irrelevant what you choose to compile into your kernel.
Again, one size doesn’t fit all, so choices are provided.
Not really. The post you refer to simply notes that at a kernel level, there’s little difference between server and desktop, and thus no real benefit in splitting the kernel into server and desktop branches.
In contrast, this article is about the layers where meaningful differences *do* exist – whether a distro is optimised for administering servers, or if it’s a Gnome-based or KDE-based desktop, or something else. Quite different subjects, and no irony at all.
The reference is LSB…
There is no reference BSD though, there is either FreeBSD, NetBSD, OpenBSD, or Darwin. You then pick and choose from there…
Difference being, the BSD’s can’t even decide on the kernel, let alone the rest of the system!
but thats the point
the BSDs differe from top to bottom
you don’t have 7000 variants of freebsd, you have 1 freebsd
and the next bsd has not just a patchwork-freebsd kernel but its own one (be it netbsd or openbsd)
i belive that a lot would be won if linus would make a base-distribution every other linux distribution must be compatible with
Well, actually that is plain wrong… there are around 6 distros of FreeBSD that I can name right now…..
Either way though, you really think its better because no one cares about it? That seems strange.
The problem with distros is that none of them want to be real distributions. It is great to have this open source eco-system. The problem is at some point you have to come out with a finalized product.
So far, every linux distro I’ve tried doesn’t want to commit to this finalized product. I use Ubuntu and I still don’t know what is the ‘right’ way to connect my wireless. I end up using wifi-radar. Whereas in Windows, there is one network manager. Sure you can install other tools to manage it. I actually think it was a slip up for Windows to not support wifi quick enough so that other tools like odessy client were allowed to proliferate. Nonetheless, for the average home user in windows, there is now one way to manage your connections.
This little issue expands when you throw in all the other choices (gnome/KDE… nvidia-settings, various applets, … synaptic, package manager). The end result is you feel like a distribution is just a bunch of packages thrown together for you manage.
This is one the reasons why Ubuntu is as successful as it is… it’s default configuration is pretty good… not great… but pretty good. Consider OSX… built on a nix kernel, but it makes the choices and delivers you one final product.
This is really where linux distros are lacking. None of them are final distributions.
Um, no. At least, not in my (quite non-trivial but still anecdotal) experience.
Every Windows PC I have seen in the past 2 years (several dozen) has the Windows network manager and the vendor’s “easy to use” network manager both installed and active, and in every case trying to sort out which is controlling what has been a real pain. In one memorable instance with my Dell D630 corporate laptop, the help desk went through 3 technicions before we got a stable connection to an n-type router. It’s a freaking nightmare.
On Ubuntu, on the same Dell D630 notebook as the nightmare above, but running Ubuntu 8.10 Live from a CD, I click the network icon in the upper right and select my wireless network, and I’m done. It works perfectly out of the box… er, CD. Same experience on about a dozen laptops and desktops I’ve tried – and one notable failure, on my daughter’s laptop, where the wireless wouldn’t configure on Windows Vista *or* Linux, but what the heck.
Maybe if Linux held an almost 90% market share like Windows, hardware vendors would write their own competing “easier” network managers, and Linux users would live the nightmare, too. But as it is, the problem you describe has been a Windows problem in my experience, NOT a Linux problem.
(One anecdote proves nothing, but my anecdote’s as good as yours, etc. etc. etc. 🙂
One thing I’ve learned in my Windows adventures is to never, NEVER use the junk that comes preinstalled from the OEM. I always uninstall everything that comes preinstalled unless it’s something I was planning to install anyway (which it usually isn’t). Those “easier to use” network managers just muck everything up and are worse than both Linux’s and Windows’ managers, or at least worse than Ubuntu’s, which I used and had a lot of trouble with at first, but then figured it out later on.
I’m not sure what you mean entirely. FOSS uses an ever evolving model versus a hard set “finished” version release model. There is ongoing development and that is reflected by ongoing availability of updated packages.
When I stop seeing package updates for a program or distro, the first thing I do is go looking to see if it’s gone stale. No more updates for Mandriva in 2006 meant it was time to look at rolling over to the 2007.0 repositories. No more updates apearing for my groupware server program mean discovering that developers where moving to something else that had the temporary lead in the ongoing darwin like code evolution.
Closed source by need of marketing image requires big wizbang version releases to get that profit spike again to fund the next cycle of development and support centre jockeys. When WinXP sales slowed down below the optimum supply chain flow point, they started hyping up Longhorn. Since Vista’s rate of sales is not up to the optimum level, win7 is in the news and being fast-tracked out the door. The retail model can’t support ongoing evolution because it’s all about getting that maximum retail price point in a big sales push.
Now, things like network management do get into the relm of distributions. It’s up to the distro maintainer to smooth over the assembled parts in whatever release cycle they choose (I like rolling distros myself though I primarily use Mandriva). Backtrack is not a general use distribution and the netowrk management tools reflect the expected higher level of knowledge (I can connect wifi by command line but F’d if I can use the GUI tools provided for it). Mandriva’s draketools are more simple than osX or Windows to use and I’ll reach for my liveCD in a heartbeat when I need solid hardware support and networking. ubuntu.. well, I really should toss the latest on a VM just to be familiar otherwise I can’t really comment on it with anything more than hearsay.
In the end though, if the distro is not providing the tools you need, grab a few liveCD or cut some VMs and look around for how other’s work. While Ubuntu is a great introductory distro and very popular, you can get better hardware support and control from other distributions.
The products are things like RHEL and SLES… the products certainly exist – but no one controls them, so everyone can release these products.
For me, I don’t need a product, infact, I demand a rolling release based packaging system because releases are plain annoying. If the software is released upstream, I want it in my distro… I don’t want to wait a year for my distro to declare the whole as stable before I can access it.
The power of Linux is that both those that demand products and those that hate distro releases can be satisfied without either affecting the other.
Actually, I think they’re pretty similar in that regard – all major distros now seem to be moving towards NetworkManager as a standard. Much like, as you say, the Windows standard tools are replacing all the 3rd-party apps from Intel and Dell and all the other chipset manufacturers and assemblers.
The way I see it, we don’t have “distributions” anymore. We have operating systems which share a common base. Debian is an os. OpenSUSE is an os, as is Red Hat/Fedora, Slackware, Arch, etc. Those operating systems have derivatives (Ubuntu is a derivative of Debian). They are compatible enough on a source level, though not always at the binary level, though different enough that they can reasonably be called their own operating systems–different folder structures, different defaults, different desktops and system configuration tools, different kernel patches, and of course different package management facilities. I think we need to get away from calling them distributions of Linux. They are GNU/Linux-based operating systems, and when moving from one to the other you are moving from one operating system to another and may have to deal with all the issues that can arise from such a change.
Standardization would be nice, and I for one would love to see it happen. Trouble is, everyone has ideas of how things should or shouldn’t be done, so we have different operating systems to fill that need. Eventually, I think, one Linux-based os will take off and be the favored target of ISVs. Which and when? Who knows, and I don’t see it happening for a while yet, at least for desktops, as certain elements just aren’t polished–ahem, dare I mention audio? It’s these unpolished elements that really need the work, and if one os leads the way others generally follow.
Regardless of what anyone thinks, there will always be many systems based on Linux. Why? Because there can be. It’s that simple.
I’ve been refering to specific distribution as seporate OS by name for a few years now. For me, it started with trying to avoid the confusion of “Linux” meaning everything remotely related rather than just the OS kernel. New users come from Windows or osX with this idea that it’s all “Linux” and suddenly they’re overwelmed with the choice and no way to differentiate.
Debian != Ubuntu != Red Hat != Mandriva != Gentoo; they are all truly individual entities that happen to use the same commodity parts for assembly. A problem with Ubuntu or Debian is rarely universally a problem with all other Linux based OS. They integrate between each other well and respect industry standards like protocols and file formats but they are as different as Hyandi and GM or Boeing and Shwitzer.
As a result, it is very rare that I’ll say “Linux” rather than “Linux based OS” or the brand name unless I’m specifically meaning the kernel itself.
I agree. And to extend, none of the GNU/Linux-based operating systems are intended for generic desktop user who can be expected to use variety of software from different vendors. They are aimed towards specialised system builders.
OS vendor can not be expected to package and distribute every piece of software and a application vendor can not possibly support all operating systems/distributions.
I do not consoder myself as a generic user but today I wanted to try out some new software – just felt like that. First, my distributor hadn’t packaged this application and its developer didn’t provide a package for my system. So I should have grabbed the source but then again, there are dependencies. I gave up as I did not want to add loads of stuff I never use after uninstalling the program I wanted to try.
Therefore, I can’t use Linux because it doesn’t do what I need. Apologists say that they perform the task using software XYZ (see paragraph above).
As long as there are gazillion distributions the developers can not choose to develop for. And from the end user’s perspective, if the OS vendor of choice ceases development for whatever reasons (a new is baby born or a billionaire ran out of cash) he is pretty much screwed. Oh and PC vendors do not start distributing GNU/Linux because in that case they would end up supporting every bloody piece of application software their customers might want to use.
Linux itself is a moving target and there has to be someone between it and end user to straighten things up by providing some sort of stability. See previous paragraph. If such company are nice enough to the customers they might also end up maintaining their own kernel and there is no more Linux but some derivate.
If there was a coherent API for end user applications and a mechanism to keep ~ and /usr/local or a substitute to it fully functional across distributions then the user would be reasonably happy. Oh and everything users install after installing the core system (which provides the APIs)should be in /usr/local. Otherwise it is implied that OS vendor manages the system via repositories and there is no freedom other than as in beer at the alternative cost of switching distros every now and then as required under given application software requirements.
Some users try to assail the GNU/Linux market with the Windows cookie cutter. It doesn’t work that way. You’ll never see one entity like MSFT, who controls what GNU/Linux is or isn’t.
The GNU/Linux market is actually level and competitive. We see lot’s of different OSes with a common core. They do share a lot of similarities, but none of them are the same.
To zoom in on your problem. My OS provider doesn’t package the software I want and I didn’t want to build it myself, so I can’t use < insert specific OS here >.
Suppose you are using Windows and you want to use Garage Band, which incidentally is only packaged for OS X, what are your options? Either you give up on Garage Band or you get a machine with OS X.
The same applies to the GNU/Linux situation. For clarity, lets assume it’s between Ubuntu and Fedora. You want to use application X, which is packaged for Fedora, but you use Ubuntu. Here we have two different OSes (Ubuntu =! Fedora, and never will be) and one supports what you want and the other doesn’t. Simply answered, your recourse is to either give up on application X or get Fedora as well.
The more complicated answer entails a lot of options and I’ll name a few, but I reckon you wouldn’t want to hear them, less act on them. One could ask the application X developer if he is willing to package for other OSes also. One could install a Virtual Machine on one OS and run the other OS with application X in the VM. One could ask the distributor or a third party if they are willing to package app. X. One could try the program Alien and try and convert the package for system Y and convert it to system Z. Options enough, but like I said, one has to be willing to go down these routes.
This is simply false… and there is work to ensure this is even more false going on…
Every distro uses the same upstream brances, all source code can be compiled without patches on another distro… if you only deal with binary packages, it can seem more separate, but that is the flaw of dealing with only binaries.
Since when do all Linux systems use the same upstream source? Each one applies their own patches to almost everything, even systems like Slackware and Arch patch a little. Some source will not always compile on every Linux-based system, due to different library version/compile options/file locations. They are largely source-compatible. They are not 100% so.
Your comment about the BSDs “not deciding on a kernel” is trolling if I’ve ever seen it, and I won’t bother to argue with a troll, as it’s rather like arguing with a religious zealot, i.e. it will make no difference.
No they don’t, many even have defined goals to not differ from upstream… Debian derivs are bad for this.
Either way though, it is still the same upstream, there just might be some backports and the like…
I think sometimes people way over think these things. For me (and only me, personally) I think the options are pretty clearcut.
If you want Consumer options, take Ubuntu (or Fedora/OpenSuse).
Enterprise: Red Hat / SLE
If you are interested in setting up a home server, than I doubt actually having a choice is going to be bad for you, and you’ll probably enjoy tweaking it anyways.
If someone you want to show Linux is thinks it’s too complicated, or isn’t a tech enthusiast, then don’t give them a choice, just install one that YOU can support. For me, it’s Ubuntu via Wubi. Why? Because I can install it without them getting worried, they can uninstall it if they don’t like it, and since I use Ubuntu I can help with any problems.
Then, all the other little choices won’t matter except to the people who want them and need them. Most people using Linux are NOT going to need a Forensic specialized Linux distribution. So don’t worry about it, let the people who do need it worry about it.
I think that’s the key right there. Don’t complicate a simple thing; select from the obvious short list of distros. For me, that’s Mandriva on desktop and Debian on servers.
If your going servers for business it’s Red Hat or Suse where support contracts and fees are desirable. if your going it alone the choice is Debian, OpenBSD, NetBSD or FreeBSD and your all set. (I like Debian because it’s supported by HP as a company policy; if it can’t be supported inhouse anymore then you just call them up and ask about service contracts)
Yeah, it’s simple for you, but to somebody looking from the outside in, it is a massive clusterf**k.
You guys can continue to preach that having 9 million distros is a godo thing, and you can continue to have the majority of computer users not running Linux.
While it’s true that those in the ‘9 million distros or bust!’ camp might be right, you have to ask yourself how important it is to be right vs whether or not you want more people to try out Linux and not be scared away by the number if distros that are out there.
Well, here’s how I look at it. There are usually two ways someone is going to start Linux: either they are introduced or they are out looking for one by themselves.
If someone is introduced, then I stand by my earlier assertion that the person introducing them need to keep it simple by using what they are used to and can support. For example, Dell does not offer 200+ distributions: they offer ONE preconfigured Linux operating system, and that works fine. The same should be done with a person introducing another person. Again, what I do with Ubuntu + Wubi.
On the other hand, if someone is out actively and seriously looking for something such as an alternative operating system for a computer, then they are either ALREADY looking for choice and wanting to explore, or are looking for some thing supported (and sometimes free) because they are having problems with their current OS. If they are looking because they are having problems with their current, all you have to do is just spend a little time reading.
Every person who uses Linux NOW has had to make a choice. It took me a whole 20 minutes, looking at only the top few that were free (Fedora/OpenSuse/Ubuntu), and having chosen by their websites that detail their operating systems. If I didn’t like it, I knew I could switch. I ended up liking it. Now, if someone is actively seeking a serious replacement, but doesn’t want to do the research on which one, they can just order an OEM Linux machine.
So, if someone is “actively looking”, if even after complaining about not wanting to look at any of the top distros, or don’t want to get an OEM machine like they would with Windows or MacOSX, then they are not really actively seeking something new, they are just complaining about what they have, and want just what they have without the problems. And that, of course, is not what Linux is.
I would agree with you that anybody who is introduced to Linux by someone else probably has their distro already chosen.
But why do you assume that somebody interested in Linux means that they are having problems with their current OS? Another big reason may be because all of the ‘Linux is the best thing since sliced bread’ articles that keep popping up all over the place. That being the case, there ought to be one distro that you could refer people to, that was universally agreed upon as the best place to start. But as it stands now, if you ask which would be the best to start with, that’s going to start a flame war between a bunch of distro zealots.
When talking about distros, I’ve been reading this website for years, and have even tried Linux in the past. But even I wouldn’t have a clue which one to start with now days. I was always partial to Slackware myself, just because it ran faster than the others. But I am also interested to see how the more ‘newbie friendly’ distros have evolved. Of course, I’m more of a power user, so I would have to try them all
Edited 2009-02-04 22:08 UTC
Of course, I’m more of a power user, so I would have to try them all
Simple starting point: http://www.virtualbox.org
Get the gratis but closed source VM. Better feature set. Try all the interesting Distro’s and pick the one you like.
Agreed, for some, a new OS choice may have nothing to do with the old OS being broken somehow.
I’m an OS geek so one of my motivations was the fact that it was something different. At the time there was win98, winNT4 and this Slackware disk a friend gave me so I gave it a go; found myself back on NT4 within a week.
.. but the second time around I tried Red Hat’s ftp install and haven’t build a single boot system for myself since.
Mandrake replaced Red Hat when they dropped .mp3 support and all the other interesting things that made it appealing to a highschool kid.
When liveCD started showing up I was downloading them as fast as I could find new liveCD distros just to check them out. This worked great until every distro started providing a liveCD. Now I have a few powerful liveCD specialized distros, Debian for servers and still Mandriva for the desktop.
I also have a few BSD VMs, a plan9 VM, ReactOS on my short list for a new VM.. Dos/win3.11 on a VM since having found my license and diskettes again. If you want to talk about boot times.. wow does Dos load quick on a quad core. I don’t think I could get it booted faster even if I did run it off a ramdisk..
but then.. seeing what different OS can do is party of my particular computer illness so I’m not suggesting everyone start a lab with a terminal for each one.
I always laugh at this… when I first entered the Linux community, it really wasn’t that complicated. Its not like enough people haven’t tried to make the decision making process easier anyways…
We don’t want one size fits all… that is what other OS’s have tried and failed at.
Linux is about fitting the system for you.
How does one manage to choose a breakfast cereal or flavour of icecream? Dare I say; choose a model of car?
The people who make a point of complicating the choice are usually detractors trying to feel better by putting something else down. Really, you can choose in five questions or less and since the average user who would have an issue choosing is unlikely to be installing there own system anyhow, it becomes up to the store rep or system tech.
For those that want to give a go alone:
Ubuntu
Mandriva
PCLinuxOS
Suse
Pick one or all and give them a try. Booting a liveCD on one’s system or for someone who is looking at systems takes five minutes per option. The first two will probably provide a selection though the third is well liked and the fourth give the business folks a brand name to feel good about.
The only reason it’s easy for me is because I took some time to look at the options. I do the same for any purchase or selection though. I wouldn’t toss any old cogs on my mountain bike. I don’t grab a shirt off the first rack in the store and buy it without trying it on or looking at the other available items.
What is the point of these Linux distributions when all you face after the login is KDE4 or GNOME2?
As far as I can tell, I still can’t find an application in KickOff, and I cannot lasso files in listview in Nautilus.
The Linux distro galore should be left aside by the adoption of an accepted and fully user-friendly desktop UI. (They do have Windows-classic to teach them this lesson.)
So I guess you haven’t tried any of the distros based on E17 (Enlightenment) or XFCE.
If all that fails, run Gentoo and examine /usr/portage/x11-wm and /usr/portage/*-base. With Gentoo you can pick your own combination! Although I have to admit, its E17 support is not the greatest.
The point of all the other distros is choice. People want their systems to behave the way they intended, not how a distro provider wanted – afterall, a distro is just someone elses decision for how the system should behave.
The great thing with Linux is none of what others decide need to be what you use. Want to customize your build flags? Go for it. Want that annoying resource intensive feature to be removed entirely? Do it. Anything you can think of can be customized…
Or you can just use what a major distro provides, and live with the consequences.
Also, somewhat offtopic, but the “one desktop fits all” again will never happen…
There is work over at freedesktop.org to reuse as much code as possible, but just as with anything else you can’t please everyone with just one product.
The specifics you stated though, perhaps try a different application launcher in KDE4 (I recommend raptor or the task oriented UI of krunner2), or file bugs on nautilus?
If people don’t know the issues, its sort of hard to fix them…
I vote Crux! Hail to obscurity! A linux distro for Sado-masochists!
This is a vote, right?
But seriously, It’s a distro on the right direction. Someone the big guys should look at.
*cough* archlinux *cough*
Created because CRUX is annoying 8)
apart from the standard pantheon there are distros that can give you good insight into building of the system:
http://www.linuxfromscratch.org/
The trouble with all those distros is that they don’t pose a problem for current Linux fans, but they do for SWITCHERS.
Switchers are very confused by the vast number of Linux distributions.
The result: a missed chance for “Linux on everybody’s desktop”.
The way things are going now, Linux will grow only very slowly.
Of course there is Ubuntu and one or two others, that are currently best for a greater public. Maybe it is the MARKETING for switchers that is lacking. Not kidding here: what about a centralized info site on linux that can empower those switchers to make the right choice(s?:).
The flaw in your argument is quite easy to pick out… despite all the distros, there is still only one upstream per project…
If new comers want a distro to try, they undoubtedly will use Ubuntu until they get a clue, and that’s fine – it makes the choice very easy for them. As people become more aware of the situation, they’ll want something that better accomidates what THEY want out of the system… hence all the distros…
What people are forgetting on this discussion is that there is still things like LFS which allow for even more customization. To say that there is pretty much only DEB and RPM these days is ignorant.
Use whatever you’re familiar with until it starts to annoy you – if it never does, you never need to dive into all the different distros available.
A sensible switcher will decide to use the distro his friends use, that way he’s got easy access to support.
Four years ago, I gave up windows for linux, and used some version 95% of the time for 3 years, except when forced to use osx or windows at work. Partially because all I had to install on my desktop was windows 2000, I also liked the open philosophy of collective effort in linux, also the reputation of linux being more stable and secure. I started with Knoppix, then Mepis, then Debian, then Ubuntu (hated Gnome, Kubuntu was too buggy), then PCLinuxOS (for a while), then bounced between Mandriva and OpenSuse. All this was a great learning experience, but quite a headache. I would have been happy to stick with the first I installed and learn it. As you can see however, and as I’ve read, many intermediate new linux users distro-hop, because they all have issues and shortcomings. Strangely, they also duplicate alot of effort, doing things only slightly differently, and not always better. Package selection varied alot also. I would read of a new program I wanted to try, and then find there wasn’t a binary for what I was using, so I tried learning to build, huge waste of time that often didn’t work. This shouldn’t be necessary. Some cross platform software I found had a linux version, but it lacked a gui or was an older version. I was always suprised that many free windows programs started in linux, but now have a easier to use, more stable, recent, and feature-ful version now in windows. From my experience, thre is far more free stable easy to use software for windows than for linux. Linux has great guts cobled together, but lack of unified systems, usually poor gui interface, the exception for me was kde and konq, both far better desktop for me than windows. The control center of suse and mandriva were also both superior to that in windows, in ease of use. So after giving linux what I think was a good try, I gave back into windows xp, as it came preinstalled on my laptop, and freed up more of my time to just do things, work and play, without endlessly tweaking and searching to get the programs I wanted and a stable environment.
I would like to see linux succeed, but am not a programmer, and can’t steer it directly. What is the goal of linux? To offer a free alternative to windows and osx? If so, it will not grow. It will continue to be a niche for hardcore computer users, servers, and devices. Ubuntu grew due a rich benefactor, and great marketing, cool logo, free cds. Desktop linux overall, still won’t grow much at all until there is a better reason than free, as most people get windows with their computer. More secure, sure, but you can buy osx computer for that, or learn to secure windows. Linux is being created by evolution by natural selection, with a little mutation, with takes 1000s of years, we need some genetic engineering. My suggestion is for the major desktop and enterprise linux companies to unite their effort, not necessarily merge, at least not yet. (No one would force all the 1000’s of smaller community distros to do this, they are fine in their niches.) Get rid of redundancies, agree on more standards to allow one binary program installation basis, a system config and hardware setup management structure, and a stable base for software developers and hardware makers to aim at. If this happened, the amount of free software would blossom. For the “year of the linux desktop” to actually happen, which to me means consistent growth and adoption of new users and existing windows users, it must offer them something they are missing with windows, something they would actually take interest and the time to learn, and even pay for, same for manufacturers feeling good pre-loading it. By eliminating redundancies and establishing a stable base, getting to the point windows and osx already are at, linux could focus on making a complete os that is a uniquely useful alternative.
“…a lot of effort is wasted on duplicate efforts, because all major distributions spend a lot of time packaging the same software over and over again in different package formats…”
How do you know that these people would have actually contributed in any other way to the Linux ecosystem if they weren’t working on a specific package management system? Whats to say that they haven’t added to the Linux ecosystem because people enjoy that package management system more?
I hate that people think we need ISV’s and thus have to cater to them… the ISV’s should just provide their software in a basic format like tar.gz, and let the distros sort out getting it to the users…
Its not wasted effort when you have that many more people fixing and filing bugs with actual knowledge of the system. It helps upstream when some of this work is done downstream. If there was only one distro, and one package management system, it would just limit the amount of eyes looking at the code, and it would hinder the adoption of Linux because the resulting system would cater to even less people.
It is purely ignorant though to say that if they weren’t doing A then they could be working on B, what if B simply doesn’t interest them? If it weren’t for A, there would just be one less developer with a vested interest in Linux. This isn’t a corporate proprietary system, you aren’t forced to work on anything by an employer. If you dislike the stuff say RedHat is forcing you to work on, nothing stops you from quitting or working on something else when you get home. People will scratch the itches they feel need scratching, and everyone benefits when they do.
There is no such thing as the best operating system out there as each of them have there own ups and downs. You talk about Windows and Window’s fan boys get angry. You talk about OSX and OSX fan boys get angry. You talk about Linux and Linux fan boys get angry. I hold no loyalty to no OS because they all have some crack head at the top who thinks they know it all. I definitely don’t, but one thing that I’ve learned is that when someone offers constructive criticism you listen. You don’t go flaming people or getting angry because your feelings get hurt. To all you Linux folks out there, don’t be like Microsoft or Apple because the only thing that they are doing is cheating themselves in the long run. I like competition because competition is good. It brings out the best in us all if we allow it to. Linux has a problem whether you believe it or not. If it did not have a problem, then it would have gained much more users. You all cannot deny that fact, and if you all choose to deny that fact and not face the truth then you all are being ignorant. For God’s sake just grow up. Can’t you all see that there are people out there who want Linux to succeed. I am not saying that everything that others say is right, but all of it cannot be wrong either. Everything that you all are saying are not right, but it all can’t be wrong either. Linux has a problem and whatever that problem is it needs to be solved. This is 2009, Microsoft is moving forward, Apple is moving forward, and its time for Linux to do the same. Move forward in a way that will make Linux more adoptable. If 2010 rolls around, and these kind of flames wars are still going, GOD help you people. I would like to see Linux take off in such a way that have never been seen before. Why is it that all the good OSes always seem to have some sort of problem taking flight. Linux needs to spread it wings far and wide, take off with a powerful thrust, and soar through the sky. Its time you show Microsoft and Apple what you all are capable of. If some say that Linux is not competing against MS or OSX, nonsense. You guys need to compete, and do it like your life depends on it. If Gates and Jobs could kill off Linux like they did to BeOS they would. You guys know they would. Find out what is holding Linux back and fix it, just fix the damn thing.
Find out what is holding Linux back and fix it, just fix the damn thing.
It’s called a predatory monopoly, with a knack of using money and anti-competitive practices to shoehorn everybody into using their products.
It’s complacent users who don’t give a hoot, as long as they don’t have to actively participate and have someone to blame when trivial stuff trips up.
It’s pundits on all sides prophesizing in contradictory ways that Linux needs to do this and that to succeed in the arbitrary and undecided upon goal of dominating the desktop market.
Then again is the goal of catering to users who don’t give a hoot, with the intention of dominating a market averse of empowerment and true innovation really the best course for GNU/Linux to go?
The way it is going now, slowly but surely, attracting people who do give hoot and innovating more and more with each iteration looks pretty good to me.
Yes, because having a single unified standard would deal Linux a mortal wound. Time has shown us how having a universally applicable, universally expectable set of libraries, controls, interface schemes, shortcuts, and applications has so negatively impacted the usability and marketability of Macintosh and Windows over the years.
Yeah…because as we all know, Apple and MS don’t break compatibility or library sets, controls, interface schemes or applications at all.
And how many variants of Vista are there again? 6? Of the one product? What a dubious strategy…choice.
As another person said already – not this shit again.
Let me tell you, if you think that by having one distribution that’ll rule them all will some how bring in the punters when it comes to software availability, driver/hardware support and end users – you’re deluding yourselves. Basically that is the core underlying argument for trying to consolidate Linux under a single banner – that if it were done, all would be great; it would be the land of milk and honey with vendors supporting it by way of software and hardware support.
Vendors who want to support Linux will support it regardless of all the excuses – because this call for consolidation is nothing sort of a shonky excuse by certain vendors who fail supporting Linux. I’ve seen the same excuses and believe me, bending over backwards to accommodate these vendors will only find that once you’ve done that – there will be yet another excuse as to why they can’t support Linux, then another then another then another. No matter how much you change Linux to suit them – the goal posts are always going to end up being moved by those who are adamant in their refusal to support Linux.
IMHO the Linux world would be better off slowly plodding alone creating open source replacements for proprietary software and putting pressure on vendors to open their specifications than trying to change Linux in a hope of winning vendor approval. Most of these vendors are in bed with Microsoft and any capitulation to them is going to be met with disappointment.
I’m supportive of Linux having multiple distros, HOWEVER, I believe it’s time to ditch X and go a much slimmer way of doing all things windowing. It would be nice to see a standardized GUI as part of a loadable module for Linux. It would also be nice to be able to do everything inside that GUI without editing archaic config files. This is one of the big reasons Windows is so successful. Configuration a web server is a matter of right clicking, hitting properties, and changing certain items. Of course that would never happen…
You might consider that there are many other people out there who have thought this a good idea, some of which have even tried doing it. Most of them never get anywhere, because it’s not an easy thing to do, and honestly, it’s really not all that useful.
X11 is here now, and it works. Even if you successfully write a replacement, how do you expect to get people to use it? Your new desktop isn’t much use if people can’t run anything under it, so you’re going to have to get applications from somewhere. Will you write some sort of X11 compatibility layer? Or maybe port existing toolkits like Gtk+ and Qt? But that’s hardly going to give you a standardized GUI, if you’re just running all of the existing non-standardized software. So congratulations, I guess you’ll be writing your own web browser and mail clients, your own photo-editing tools, your own media player, and all that.
Sounds like a lot of work, doesn’t it? What was the benefit, again?
you might also consider Kristian Høgsberg’s Wayland project
you have forgotten the most likely and common cause, lack of time and developer manpower…
it’s been here for over a decade, and its overall architecture hasn’t changed much from the time it’s beeen conceived – which was a very different time from today, from the point of view of functional requirements for interactive gui systems..
interestingly, high end unix systems that focus on interactive high performance graphics, did / do not use X, plain or at all (), and sometimes have it as a legacy compatibility option – there must be a reason…
really, does it?
simple, by writing a replacement that doesnt enforce new application level APIs
actually the best choice is to port toolkits to a “common denominator” graphics library like opengl (even better if leveraging openvg for primitive rendering and antialiasing), and this has already been done – AND to redesign the graphics subsystem around a streamlined optimized component for the local, composited UI, plus an *optional* service that would support legacy X11 server side stuff for applications needing it, while being layered upon the previous component)
the point is not (not primarily) standardization (assuming you mean consistency in applications’ look and feel), because standardization would still be attained by using the same toolkit ( or maybe a common theme library will be possible, now that code sharing between qt and gtk+ will be easier thanks to the lgpl licensing of qt…)
the point is having a more compact and efficient system that can better exploit the capabilities of today’s hardware
what’s required to perform the aforementioned is the ability to render the application’s widgets and work area to opengl surfaces – this requires the toolkit being ported, not writing an entirely new application
and makes your guess sound like fud…
less than one may expect, and some of it has already been done
optimization and bloatware removal, anyone?
From the discussion it seems like we have only two choices, merging into one distro or having many small.
What about the idea of one distro simply becoming popular, big and a lot used and becoming the “standard Linux”.
Actually, I wouldn’t be surprised if that happens if/when Linux gets mainstream. Whether that is going to be Ubuntu, Fedora/Red Hat or SLED (or something else) I don’t know.
What do you think?
Windows SUCKA… Linux RULAH…
Choice, choice, choice… blah blah blah
There comes a time that if you have one more thing to choose from, the confusion starts… this is not very good as a corporative solution. Anyone heard of xpde?
Linus is quite right – one of the biggest strengths of Linux is that there are so many distributions. But the problem is that there is a very low level of compatibility between these distros. That fragments the market badly and severely limits the adoption of Linux by people who care less about configuring their systems than just using them.
Microsoft has the reverse of this problem. They try and make their products into a one size fits all solution, and it just doesn’t work. There are always people who’s needs are not quite met by Mirosoft software, but who are too small of a market for MS to care about. OSS allows even these tiny markets to be catered to.
LSB is a nice start, but it’s only a start. Until software developers can target “Linux” as a platform instead of so many distros, Linux will unfortunately never gain the adoption levels of Windows or even Mac. It won’t even really “be” a platform in the eyes of developers, commercial or otherwise.
Having one big distro muscle out the others is a very dumb way of gaining compatibility, and you would lose a lot of what makes Linux cool. A better way would be to get all these guys to agree on some basic principles regarding underlying filesystems and storage locations. Perhaps having a basic test app would help, where if that app runs without modification then your OS can be certified? Similar to the Acid test website as someone else said?
As it is there are endless battles about trivial things and nobody ever wins, so no progress is made.
he’d be out of work if he wouldn’t
http://oswar.wipetheworldsass.com/
Now why do people feel a need to kiss Microsoft’s arse with a blog like that? It’s kind of like watching the Super Bowl and then declaring, “Yeah, I’ve been a Steelers fan for years even though I don’t live anywhere near Pittsburgh.”
Yeah, there’s too many different cars on the market as well. Remember kids, choice is bad.