“How many distributions does the Linux world need? And what exactly is a distribution, as opposed to just an edition of another distribution? Why is it that there are so many developers who feel inclined to start their own project instead of joining another, more established one?”
The answer to the questions above? One word, freedom. Freedom of choice, freedom of code, freedom to redistribute.
Of the 500 distros the article says they do bring up a good point. Move some of that talent to a more focused area. I’d much rather have 20 distros specializing and making a name for themselves than 500 doing nothing but replicating.
Would a replicator really have anything to add to another project? A remastering with different packages might produce a distribution that is a better fit for some out of the box, but it is in no way innovative.
No matter how you slice or dice it, it is the top ten of distributors that fuel the innovation and the rest just paints by numbers. (When speaking about OS design). Nothing wrong with that though, one day one of those painters might decide to get an empty canvas and some real oil paint and come up with a masterpiece.
I know that most people feel that freedom of choice is a good thing, even if it leads to so many distributions. However, not all are even worth effort. For example, I am a student of Boston University, and when I found out they create their own GNU/Linux distribution, I felt compeled to try it out. The idea is a good one, create a distribution that is preconfigured for the on-campus network or off-campus computers. It is based on FC4, and this new distribution makes sense so that students have little to no work in getting their computers up and running. However, I did experience some problems. For starters, my iPod does not work out of the box. College students are some of the most inclined people to buy/use iPods. Therefore, a distribution geared for college students should have this out of the box. The scroll portion of my touchpad did not work either. Recent data shows that laptops are outselling desktops, and college students would (in my opinion) would be more interested in a laptop than a desktop. I encourage people to create a custom distribution for a specific workplace or purpose, but make sure that it is really different. For me, BU Linux is no different than Blag or any other FC variant.
i don’t think a new distribution is the right thing for universities. it’s not their primary goal to create such a complex piece of software.
i’m a student too and i think it was a better idea to provide some good scripts configuring the pc as you have to… like wireless startups and so on.
an uptodate distribution is a hell of work for an university. and gnu is able to be configured to hell.
some userstory for your pleasure:
one guy was struggling with his w32 system to get wlan working, me myself was working on my notebock (no problems). when the guy asked me about how to get this running… i’ve told him just startup this script.
he didn’t know anything about scripts and this funny os. i gave him an ubuntu live cd, per usb we copied the scripts and … wow! it works. (yes i’m happy about his wlan card too… no wrapper needed
)
bit off topic sorry.
if the world has let’s say 5000 (just a random number) active developers in the OSS sector and 500 linux distros that makes 10 developers for evry distro, but if we had only 50 the number per distro would be 100… and as long as management is done right more developers=faster and maybe better development, less bugs, and more features.
let’s say I’m a musician, and I’d like to use linux to do some music, there’s at least 20 distros out there for music making, but none is really finished, none I’d call a great solution to my needs… if they would join up in, say 3 groups maybe things would be different.
Just to add on to the music making distro, 64 Studio is due to release 1.0 in 3 days. Its a pretty neat distro I’ve been playing with over the last week or two, unfortunately I don’t know jack about making music. But its still fun to load up a drum machine, virtual keyboard, or synthesizer and start pushing buttons.
if the world has let’s say 5000 (just a random number) active developers in the OSS sector and 500 linux distros that makes 10 developers for evry distro, but if we had only 50 the number per distro would be 100… and as long as management is done right more developers=faster and maybe better development, less bugs, and more features.
let’s say I’m a musician, and I’d like to use linux to do some music, there’s at least 20 distros out there for music making, but none is really finished, none I’d call a great solution to my needs… if they would join up in, say 3 groups maybe things would be different.
I think there are too many bands putting out too much music now, with most of it being crap or at least making it difficult to find the quality music.
So you as a musician can either become part of the problem and start a band or join one that’s fringe or has no future, simply because you care about enjoying your job, or you can become part of a solution by sacrificing your personal feelings in the matter and joining some pre-configured corporate run band in the name of economizing the music industry and helping end confusion for customers.
Ok, extreme example, but you’re using flawed logic in your argument, and it’s the same flawed logic many others use. There seems to be an assumption that developers are just desperate to work on OSS projects that they’ll take whatever comes along or are told to do regardless of how they feel about it, rather than pursuing their own ideas.
Seriously, the guys that are breaking new ground and trying to things differently in distros like Arch, or Gobo or even Gentoo are not likely to give up the cause and conform with a mainstream distro simply because people think they should. It kind of misses the point of why many people embrace the OSS model in the first place.
I’ll never understand why people think linux needs one desktop, one distro etc. To me choice is and will always be the real power and the market/userbase will ultimately determine what works best without denying choice to others.
well, I don’t think we need one desktop, I think gnome, kde xfce are all different enough to have a “right” to be, I think too that choice is great… but there is a lot of distros that are basically the same thing.
the point is, if I want and have the needed knowledge I can take one of the existing distros, say ubuntu, and make my own based on that… but why not throw my ideas in the pool for ubuntu itself? why not just work on a metapackage for it? it’s a bit like “why are banshee and rhythmbox two separate programs?” joining the the two toghether could make a better app…
There is a flaw in this reasoning. It asumes that developers are uniform interchangeable creatures, that can be lumped and combined in any way a “manager” sees fit.
In reality these are not “coding robots”, they are people. Most of the time people with very outspoken ideas and opinions. You can’t just lump them into centralized development teams and expect that configuration to work. I’m not even thinking about faster, better, increased efficiency or more innovation.
Some people can’t work together. (L. Torvalds and A. Tanenbaum working on the same kernel?) Some people can work together, but still produce nothing of consequence. Some teams are just the magic mix that produces wonders.
Problem with finding the magic makers is that it is a random, darwinistic proces. It takes a lot of albatrosses, to yield one perfect development team. Just lumping different teams together will not yield a faster, better, innovative team.
To answer your question, just look at all the forks of different projects and the reasons why they exists.
Most of the time the reason falls into one of these categories:
– I don’t like your indention style
– I don’t like the patch tracking system in use
– I don’t like foo but prefer bar (this can be initstyles, package format, color of the login prompt, anything really)
Very rarely is it something of actual value. And that’s when magic happens 🙂
…”you can become part of a solution by sacrificing your personal feelings in the matter and joining some pre-configured corporate run band”…
That’s not a solution, that’s contributing to the problem.
I’d mod you up but you’re already at 5.
Throwing more programmers at a problem doesn’t always solve the problem faster or create a better solution.
Read http://en.wikipedia.org/wiki/The_Mythical_Man-Month for details.
Remember that most distros are built from the same set of core applications, and each distro is created to scratch a specific itch. Sometimes it’s something as “trivial” as making a certain window manager the default, but to a person using that distro such things could be a critical element in deciding which distro is the best one for them.
I don’t see any value in limiting distros. Perhaps a certain level of standardization could be attempted in terms of software location, filesystem structure, or perhaps even package formats, but in the end a free software universe will generate as many distros as their are interested parties — and that’s a lot!
Edited 2006-08-28 21:00
Yes, I agree.
There are too many Linux distros out there already!
The question is why? Well, because people are taking a certain distro (as their base) and modify or customize it to their liking. What for? Probably to show that they have their own version of Linux (& hope it becomes popular?). Maybe it makes them feel good about themselves? Or they are just really bored and need something to challenge themselves? Who really knows.
Most would agree that it would be better to have fewer Linux distros! Will this happen? No it won’t.
I think a maximum of somewhere between 50 (to 100) distros would be more than enough for most users out there to choose from.
If you surveyed a large number of Linux users, you would see that they tend to use distros listed in the top 50 (or so) on Distrowatch. So, do we really need the other 450 out there? The simple answer for the majority of Linux users is no.
Do we really need 500 linux distros to choose from? NO WAY. Too little or too much choice is always a bad thing! Think about it.
I’ve found all that really matters are the top 50 distros on Distrowatch. You could even push it up to 100, but anything past that isn’t worth looking at (unless you need a really specialized distro).
Developers of the small Linux distros should join up with a bigger project (ie: something in the top 100). They can work together, to report bugs, improve functionality, provide alternative backgrounds, icons, make suggestions, etc. Help the big distros to become even better.
Unfortunately this won’t happen, because I think the smaller Linux distros are created by individuals or groups, that have something to prove to the world (to satisfy their ego? to prove they can make a Linux distro? Or to make a specialized Linux OS) and want total control over everything (because they feel it is better done their way).
But it doesn’t matter if there are ten developers per distribution or not – since in general the developers create *their* software.
Later the distribution adds it.
Right now there are probably less people writing distribution-specific projects than there are people writing software which just happens to get included.
If you take 10 developers and double them five will write a text editor, five will write a mailserver and ten will write games. It isn’t as though by creating more developers “per distribution” you’ve changed what they’re working on ..
I don’t fret all these choices. Actually, I think it’s great. Right now, I have Sabayon Linux (Gentoo-based distro) on my desktop. The developer(s) did an excellent job of creating an easy-to-install Gentoo systems with XGL, to boot. On my laptop, I have (Open)Suse 10.1.
While I do understand the point of having fewer distributions benefitting from an increased pool of develpers, I am grateful for all the options that exist out there for me as a user.
and this may improve security allso, as the attackers cant just automate their way thru it…
Five years ago there were 10 major distributions. Now five of those ten have gone, but there are still 10 major distributions.
How many distributions does the Linux world need?
As many as there are people willing to start new ones. It’s about a little thing called freedom.
As many as there are people willing to start new ones. It’s about a little thing called freedom.
Your answer is wrong because you misunderstood the question. The question was, “How many distributions does the Linux world need?” and you provided the usual but strange reply concerning freedom. You see, there’s so such thing as a direct dependency between ‘need’ and ‘freedom’, and nowhere in the world does ‘freedom’ magically create ‘needs’.
Likewise, in the Linux world, the freedom to create distributions does not mean that there’s actually need of new ones.
“A handful” is probably a much better answer. Excess harms, as does lack. In a perfect Linux world, we’d have a handful of distributions, leaving enough room for choices and individual preferences and your ‘freedom’, covering most needs and expanding limitations where new needs grow; enough distributions to create a healthy atmosphere of competition and joy of steady improvement, but not so many that focus and strength are wasted on badly reinventing things.
Edited 2006-08-28 21:12
Thought out within the confines of the conventional market where succeeding is an economical demand on monetary grounds. Since most fringe distributions are a passtime of their respective developers, it doesn’t matter if they exist or not. They are a labour of love and don’t answer to the bottom line.
They don’t take from nor add anything to (most of the time) where GNU/Linux really matters. Where GNU/Linux matters, there are still the top ten choices. Where people consciously choose to use something other than the “big ten”, it was beneficial without harming anybody.
Fragmentation is a “Ghost of FUD” flung around for various reasons. Truth is, that GNU/Linux is GNU/Linux is GNU/Linux. The similarities are far greater than any perceived differences. For proprietary binaries, there is always static linking. For FL/OSS it is just a recompile.
The biggest problem GNU/Linux faces is Old-Think.
It is the nature of a free system. The system will consistently compartmentalize and fragment, but in doing so it grows. Competition is embedded in the system. But the coolest thing is that the competition is inherently passive, because of the limitations created by open-sourcing. So it has the benefits of competition, it has the wide variety of choices/opportunities/avenues, while at the same time, all of those different distros can benefit from learning from what all the others are doing. So everyone is always involved, no one is left out (because if they are, they start their own distro) and we all benefit!
Why there are so many Linux distros, and why there are so many open source apps that do the same thing, a bit differently. Maybe one of the best ones is that it just scratches a developer’s itch to do it their way.
I can’t lay claim to using every different Linux distro, although I’ve used quite a few of them. Some of them were very good, some not so good. Right now I’m mainly using Debian, Ubuntu and Slackware, but I have used Red Hat, SuSE, Mandrake, Libranet, Gentoo and a few others I can no longer remember the name of. Each one had it’s strengths and weaknesses. Perhaps my favorite is Slackware, with Debian (and Debian-derrived distros) coming in a close second.
As for why don’t developer’s throw their support behind an existing project, instead of striking out on their own? Well, my answer to that is perhaps none of the existing applications do things the way the developer wants. I’m in the middle of writing my own Amateur Radio logging program for Linux and open source. Yes, there are several good ones out already, but not a single one of them does things the way I want to do them. I’m not writing this program for you or anyone else out there, I’m writing it for me. I will release it under the GNU GPL, when I feel like it’s ready, but if no one else uses it I won’t lose a wink of sleep. This is being done for me. It scratches an itch I have. Understand?
I don’t mean to harp on this subject, but frankly I’m sick and tired of listening to people whine about how many distros of this there are, and how many apps of that there are, and that things would be better if the developers would get together and do this or that. Bull. Freedom IS the spirit behind open source. It’s more than just the freedom to look at and change the source code to someone else’s program, it’s the freedom to use any distro you wish, or any application you wish, and if there aren’t any out there that work the way you want, then it’s also the freedom to create your own distro or app, and have it work the way YOU want it to. And it’s the freedom to not have to worry about someone else telling you you can’t.
Admittedly going off at a bit of a tangent here (I suppose I could be replying to a number of posts in this thread) but just to apply the frequently used car analogy to the last paragraph – is it true to some extent to say we should distinguish betwen the kit car and and the custom car “model”?
The kit car you may assemble yourself, based on standard parts that are known to work, according to a standard plan that is clear. The builder of a kit car may have a number of reasons for not buying a “standard” production-model car, but the kit car system works because the parts relate to the plan and the plan to the parts – it’s standardised in that way. There may be different types of kit car model for different types of motoring. You can even buy kit cars that are assembled by other folk.
The custom car model allows an owner to radically alter the appearance of a production-line car (or even a kit-car, come to that). It can be as excessive or as modest a change as the owner feels, but basically after all the changes, the car still runs because literally under the hood it’s a standard car – you may still turn the key in the ignition and it works.
Sometimes I get the feeling that there’s confusion in the Linux world between the kit car and the custom car – the custom car is taken to be the basis for “another distro” when it should be the kit car, which is what in my very humble understanding a true distribution should consist of.
For all the geeks, hackers, and savvy folk out there, no-one whining about an apparent loss of overview or potentially unnecessary dissipation of effort is going to stop you creating your own custom jalopy – you have the technique, technology and can rebuild – you are not going to lose out.
On the other hand, the ‘ordinary user’ would probably in my view win a great deal of confidence if, in being able to chose an alternative to a standard production-model OS (say, Windows), he or she could at least recognise it as a car, expect it to work according to the plan, and apply their ordinary driving skills to it.
just my ‘bear-of-little-brain’ view…
Edited 2006-08-29 12:35
Isn’t it true that if you remove the gnome package from Ubuntu and install the KDE pacakge you will get Kubuntu? If this is true, how in the world can it be considered a different distro? When was the last time someone was able to do an apt-get install redhat on a Debian system? I think their logic is fuzzy.
Those people that think, “O what a waste of manpower, all these gifted developers working on all these different distros,” thou shouldst not worry.
a) Some geniuses have the type of egos that don’t function well in a bigger team. Just fork their code, it’s legal.
b) Many people don’t really know how to program good stuff, but they do know how to make their own (live)distro and tweak KDE a bit. With all due respect, these people, and their distros are as lazy as I am, so no great code wasted on that either.
c) A lot of distros simply vanish as quickly as they have come. No tax money is wasted on it. Not much environment pollution. And you don’t have to check out DistroWatch if you don’t want to.
d) The few distributions that do matter are already the ones people look at all the time.
Rivalry, competition, different accents, it’s what made (will make?) Linux a great success. It keeps people from getting lazy. You think Suse aren’t doing their best to take Ubuntu’s market share back? You think Fedora/RH won’t try to come up with a great Aiglx/Compiz-ready desktop, rivaling SLED? You think people aren’t embarrassed if their distro is so much buggier than another?
People think that mergers and takeovers and big players are a recipe for success, because they believe in an economic model that is based on short term revenues, firing people, market share, etc. Linux should be about long term software quality and happy users, not about world domination.
The day there is one unified Linux distro is the day Linux dies.
How much does the average distro developer actually do? Really, they just compile and package other projects’ works. The exceptions being custom package management stuff, like portage.
I use gentoo, so I’ll use portage as my example. Obviously gentoo developers are sending patches back to various projects.
I just don’t buy the “if there were fewer distros, there’d be fewer, more refined distros” argument. It’s not the distributions themselves that are providing Music Creation applications, as mentioned in an earlier post, but the specific Music Creation projects.
How much does the average distro developer actually do? Really, they just compile and package other projects’ works. The exceptions being custom package management stuff, like portage.
Back porting of security patches to a given version springs to mind… And in theory, regression testing lots of various bits of software together.
I use SUSE 10.1 and that’s the only distro that exists as far as I am concerned until I am back on the market. I am sure it’s the same for Mandriva, Ubuntu and other long time users.
Linux as an OS, whilst still the best for my machine has stopped to get me excited. It works magnitudes better than XP for me, for reasons (XP/my box faults) that I don’t have the time to investigate.
What I keep an eye on now is applications. kde-apps.org and gnome files is a lot more interesting to me that distrowatch. I recently found out that there were now many apps to graphically master video DVDs. That’s what I am interested in.
However, the variety of Linux distros is like a gigantic market study, if you forgive the relatively small market size. The success of this or that distro shows to the others what the way forward could be: knoppix has shown the interest of live CDs, pclinuxOS has shown that it’s an excellent install technique, now Ubuntu shows that initial simplicity is an asset. PCLinuxOS , Linspire, automatix and others show that out of the box multimedia capacity is a desirable feature for most desktop users.
But technically, the fact that everything can be done with any of the ‘polyvalent’ OSes out there, be it Linux, OSX or even XP shows that distro proliferation it’s no technical necessity.
To the distro editors out there, I am more game for supported, niche-purpose only repositories for SUSE than for a new distro. And yes, if QA and support is good, I would consider paying a subscription.
Edited 2006-08-28 23:04
Cant resist .. comment :
The choice is great.
It means there is a distribution for every taste out there.
But there are more users with a certain taste/requirements than others.
There is tons of choice but oviously the work across distros will often be the same – fixing bugs in features , new stuff etc.
The number of big groups of certain taste are not that big compared to the 400 or so distros .
Distrowatch has 18 distro categories + BSD + Specialist and still there is a lot of choice in each category.
Close to all of these 400 distros are duplicating efforts as they will be targeting one of these 18 .
I love the choice but I wish some projects would merge to finally create the e.g. proffessional audio editing/recording environment etc.
& so furfill certain industry requirements .
There are six cluster distros on distrowatch.
A Christian version of Ubuntu.
A distro purely to show off amarok.
There is “Oralux” a distro for blind/half-blind people – this effort could be merged into mainstream distros.
There is so much duplicated effort which could be combined & speed up the progress of linux in all the various areas it can operate in .
Just IMO
EDIT : Yeah .. its of course also an ego thing – who wants to give up their distro for the common good ?
Just like politicians want to reduce taxes but dont want pay cuts for themselves .
EDIT : Focused development seems to work great for Apple-OSX – they end up with cool new features each new release. Their development is not so widespread as open-source or seemingly chaotic as Microsoft’s .
Edited 2006-08-29 01:12
Not enough. I hope in two years there are over 700.
…Notepad clones does Windows need?
…FTP clients does Windows need?
…virus scanners does Windows need?
…archive applications does Windows need?
Obviously all these developers “must” join forces to create the one true Notepad/FTP client/whatever since all this choice and work on what you like stuff is just nonsense.
Well done!
I’m not using Linux because this chaotic distro world. I think it’s good for users to choose the distro they want, but many can’t rely on incompatibilities or different application installation methods between them.
That’s why I’m using FreeBSD (yes, call it, monolithic, centralized, etc). Many things of course are harder to achieve in FBSD than LINUX (big hardware support) but the planned development model of BSDs are my way of thinking. I like the traditional UNIX way.
Anyway, Linux is a great thing but many efforts should be centralized or put on the road in ONE way.
“inclined to start their own”
Why ? Because they can. This is one of the beauties and strengths of FOSS. That’s all folks.
The number of diffrent distros that is still maintained will drop for sure. Keeping a dist up to date is a huge effort.
Can’t mod that last bit up enough brother. Takes about two years for that reality to settle in so I would hope there’d be enough new distros born to balance out the drop off keeping that number fairly constant and the new idea pool full.
Edited 2006-08-29 14:54
There is no “Linux World”.
Instead, there are thousands of individual Linux users, each with their own set of individual needs.
Many of those users are technically inclined, and if a distribution doesn’t exist which already does what they want, their natural tendency is to create one that does.
It’s the same process we see in the open source community in general.
Does it create a certain level of redundancy? Perhaps, but it also creates a very healthy software ecosystem in which the superior implementations tend to rise to the top over time.
This is better, in my opinion, than the proprietary software market where the folks with money and/or marketshare tend to dominate regardless of the ability of their software to perform in a satisfying way.
It would be great if desktop became a layer on top of a basic, solid, stable, linux base, just like ubuntu-desktop/kubuntu/christian ubuntu/xubuntu on top of ubuntu base install.
so you can use your XXX desktop knowing for sure that repositories would be always updated and everything underneath is OK.
maybe the distros will take the mepis way but a step further, providing just a new desktop to ubuntu, so will be choices for everyone, but sharing a common base.
a new developer that wants to create a new user experience will not reinvent the weel, but just focus on what he/she wants.
sorry for my english.
How many distributions does the Linux world need?
One, it’s called Debian.
😉 Nice one.
… because it reflects, well, old fashioned thinking. Linux is not the product of a single company (it’s not Apple, it’s not Microsoft – thank God), but a product of the community. And it is not an operating system, but a kernel. And when it comes to the distributors you’ll have as many choices as there are developers.
The good news is: you can’t buy Linux. That’s why I love this kernel. I read in Linus Torvald’s book that Steve Jobs wanted to assemble the Linux developers behind Apple’s Mach kernel. Another one who has not a clue about Linux at all …