Slashdot Mirror
Too Many Linux Distros Make For Open Source Mess
AlexGr writes "Remember the 1980s worries about how the "forking" of Unix could hurt that operating system's chances for adoption? That was nothing compared to the mess we've got today with Linux, where upwards of 300 distributions vie for the attention of computer users seeking an alternative to Windows."
Must be a really slow news day to bring back this ancient argument.
are actually in use though? Ubuntu, Fedora, SUSE, Redhat, Gentoo, Slackware, Debian? There are many distros, but most are specialized forks. Most people would use one of the listed ones.
kill all the fucking niggers
You don't need to know all 300 distros to make a good choice. It is pretty clear which distros are mainstream and which ones are not. If you are looking for a general-purpose replacement for general-purpose Windows, you can go with Ubuntu, Suse, Redhat, Debian or Mandriva. Almost only if you're "hardcore", you will dive into special-purpose distros such as business card/feather linux, freesco, etc. That is from a user perspective. From developers perspective there is such a thing as LSB.
Visit http://ringbreak.dnd.utwente.nl/~mrjb/growingbettersoftware to download your free copy of the book
I have always kinda thought that this was at least one of the reasons why linux adoption is low among the 'mild computer user' crowd. It isn't easy to explain to them either, since there isn't a corollary in the "windows world" where nearly all of those users reside (with good reason).
maybe with this recent gathering of support behind ubuntu there is the potential for more of a standard-bearer in the linux world, at least in the eyes of those who only use windows/osx.
Beware the Jubjub bird, and shun the frumious Bandersnatch.
Forking of software development projects has interesting consequences,sometimes good, sometimes bad, sometimes neither. Having more than onedistribution... I'm not sure that "forking" is even the right word toapply to that.
Bruce
Bruce Perens.
http://bbspot.com/News/2000/4/linux_distros.html
Regards,
John
Falling You - beautiful
Otherwise, no. of Linux distros would soon exceed no. of Linux users!
Do something!
This was and maybe still is a valid point (although diversity isn't that troublesome for businesses), but now Debian-based distributions and especially *Ubuntu got extremely popular, and are in the way to become the defacto standard for Linux, whereas other distributions will remain domain-specific. For example, if you have a business and want tech service and all that, you may want to try SuSE or Red Hat. And if you are a ricer, you may want to try Gentoo :p .
Fortunately, natural selection and evolution of distros made one very popular, which means more packages and less compiling for the general public. This is what Linux needs. The fact there are many other distros for more specific or purist purposes is alright - it doesn't affect Linux' adoption because if you're concerned about popularity you get *Ubuntu.
I was about to say 13256278887989457651018865901401704640, but it appears this number is private property.
A distro is not a fork. It is not a fork if the patches flow upstream.
...
I know there are exceptions to this rule (iceweasel, icedove) but in general, all distros contribute back to the same pool.
The only issue here is consumer choice, not wasted developer power (unlike real forks). And the Novell fiasco shows the problems
with having a single "one true way" distro - even if it is a community project (in which case its death comes from group
think and dragging its feet on decisions).
A distro, 'taint a fork
Quidquid latine dictum sit, altum videtur
After all...
This. Is. SLASHDOT!Also, there are to many brands (not to speak of models!) of cars. My prediction is that cars will TOTALY fail within a year or two and the horse+carridge will make a glorious return!
Remember the 1980s worries about how the "forking" of Unix could hurt that operating system's chances for adoption?
Yes, I remember. All of us can see now how "forking" hurt Linux's adoption. Not. Besides, wouldn't hurt to try figuring out what the difference between forks and distros are before next time.
I am putting myself to the fullest possible use, which is all I can think that any conscious entity can ever hope to do.
The striking thing about all of the distros I've seen is that barring incidental things like packaging systems, KDE or Gnome, etc. they are largely the same. The biggest change I've seen of late is an huge increase in quality of the free-as-in-freedom distros.
But why would you want to invest a large %age of your time making something that well, is already done reasonably well by somebody else.
What would be nice is if the smaller distros start to take a role of really experimenting and breaking the rules.
OLPC is an example of what I'm talking about. They work from requirements, think outside the box and have come up with something truly amazing, something new.
So those slaving away on their boutique distro that looks like the rest, please, find something better to do, like really innovating. That's the only way to make your distro a break-out success anyway.
It's kind of like US presidential candidates. The field starts out pretty wide but you know early on most of them don't have a chance. The fringe candidates should at least make themselves useful, speak the truth and stir things up.
-- John.
The car analogy is a good one too. There are now far fewer platforms than there are models, e.g. in Europe VW has the Polo, Golf, A4, A5 and A6 platforms that are used by a wide range of models spread over several brand names (SEAT, Skoda, AUDI, VW). Ubuntu can be seen as using exactly the same approach, with Kubuntu, Edubuntu and Ubuntu as brands but based on a small number of real platform variants. You can argue that the Linux world is actually more visibly attuned to the consumer market, while Windows is more like Communism - the State of Gates decides what the factories will make, and the end users put up with what they are given.
Pining for the fjords
Too many Linux distros make for Open Source mess
Isn't that the same as suggesting too many different brands of cellular telephone make for a communications mess?
"Oh dear me, there are far too many different cell phones! How do I choose? What do I do? Oh, damn it, I'll just send letters instead."
I think not.
Linux/Open Source/Anti Microsoft News
Most people don't know a thing about the throng of Linux based distros. It's more an insider joke. You're mild computer user knows one or two at best. If they know more they've been digging around and no longer fit the category.
The truth is that the diversity is great. I don't want to see 1000's of distros pushed mainstream per-se, but there is often a reason for the variety. It suits someone anyway.
What I would like to see is more collaboration. Why is Redhat/Fedora building the cludgy system-config* and Suse sticking with YAST while Mandrake (who seems to be losing favor but has committed all their development to the GPL) created DrakeConfig, which actually almost worked.
Quack, quack.
When there were many UNIXes, the problem was that software written on one would not work on the other. Linux has maintained almost complete binary compatibility for applications for ages (I guess a.out binaries could now be considered "not compatible"). All that is needed is to install a compatibility library. This means that essentially all of those different distributions are equivalent to one single UNIX version.
People really don't remember their history any more. There wasn't even really source level compatibility from UNIX to UNIX. There were two completely different operating systems (BSD and SystemV) both used as the basis for the different incompatible UNIXes. If you used, for example the "ps" command, the arguments would be different from one to the other. This meant that even shell scripts weren't portable. Claiming that the different Linux distributions are like different UNIXes is crazy when you compare the differences between SunOS4 and SunOS5 (also known as Solaris) which are bigger than the differences between RedHat 6 and Gentoo 2007. Damn youngsters.
Because the average person that has even heard of Linux only knows of one distro: Ubuntu.
Us Linux users are not mindless cattle to stampede the shops and get the latest and greatest distro there.
Instead, we are gentlemen of leisure; our systems are updated via network as soon as the new packages hit the server - we have no need to wait for them to be burnt onto CDs, packaged in pretty boxes, delivered to stores and sold at premium price, while we risk our lives in the stampede.
Then again, when you wait for a new version of your OS for five years or more, it is understandable that you want to upgrade immediately; you have tested your patience long enough. We, on the other hand, live upgrading what we choose, when we choose; our patience is never tried, never tested, never gone.
Oh, yes. I nearly forgot. If we really really want the CDs with Linux on them and can't afford to download the ISO, we simply order a bunch from Canonical and have them delivered to our doorstep. And we chuckle when they arrive, for we imagine you standing in line or stampeding the stores to get the bestest and latest, while we sip our drinks and surf the net while our systems upgrade.
Keep your mad rushes. We don't need them, we don't want them.
Ignore this signature. By order.
I work with QA in a team that produces traditional closed-source software for Linux. The thing is, thanks do the fact that there are so many Linux distribution, our software quality automatically increases. This is how it works: we, of course, need to test on as many distros as possible. Naturally, we focus on the distros that customers use. But basically, we just shove in as many different Linux variants as possible into our testing systems (given our hardware constraints), and each night test the latest nightly builds on some 30+ different distribution/version/architecture combinations. This might seem like a lot of work, but it turns out we can find the most obscure bugs thanks to testing on such a diverse set of platforms. And in the end, this gives us an advantage in that it forces us to produce code that works well on pretty much all different kinds of Linux configurations out there. Usually, since the more specialised distributions tend to be based on one of the mainstream ones, we automatically cover most of them too. If a big customer starts using a customized Linux distribution, we're likely to add that to our automatic testing system, too, but usually the big names are enough.
So while it may seem a hassle to test on a vast number of platform, it really makes you think about code robustness and quality in a different way. Of course, there is a long way to go in certain areas, not to mention universal third-party package management and desktop integration, but we're slowly getting there, too.
It wasn't ever true. Linux distros were never like the great Unix fragmentation mess.
/opt (e.g., SuSE) which others install in /usr. And about 90% which just download RedHat's RPM's and put their own name and logo on it, so basically they don't even really count as different distros.
What we have now are maybe 10% distros which pack a _slightly_ different mix of the same tools, or just different default tools, or sometimes just and maybe have a slightly different config tool. Or maybe they'll install one tool in
Either way, from an end-user point of view, whop-de-do, you run the same tools, with the same options and the same interface. That's especially important because for an end-user the OS doesn't even really matter. The computer is just a tool, and the OS is... well, I think Joe Average isn't even sure what the OS is, he just knows he has to have one to run the important part: the apps. What matters is what you can run on that computer. (See the endless "but it doesn't run MS Office" and "but I can't play the latest games on it" arguments.)
Even if one distro skipped a tool you want, you know, there's nothing to stop you to download it yourself.
The Unix fragmentation was a whole different issue. Each of the major vendors actually worked hard to lock their customers in. Unix got fuc^H^H^H forked so hard, it wasn't compatible even at source level any more.
As I always remind people, people want interoperability and open standards when they're the underdog, and they want free access to the top dog's customers. When they're on top, even on a niche, they don't want that any more. Then they want walled gardens and penned captive customers that they can milk and shear regularly. Then they want you to think, "damn, if I get a mainframe to replace these aging Sun servers we have, we'll have to change all this mountain of source code, and for some we don't even have the devs any more and for some, well, we thought we're smart if we get it cheaper without sources... oh well, better buy the next servers from Sun too." And the difference in parameters and effects for the supplied tools, meant you got to retrain all your admins and rewrite your scripts too.
When you're at the top of your own niche, it's all about trade barriers. You want to make it as hard as possible for a competitor to steal your customers. (And unsurprisingly, IBM for example was not only on the receiving end of an antitrust trial long before MS, but also the word FUD was originally used about IBM's practices.)
So, anyway, that's what they did there: each took their own fork of Unix and ran in their own direction with it, as far from everyone else as they could and could afford to. AIX and Solarix, for example, weren't just different distros, they were almost different operating systems. "Portability" was only a buzzword everyone used only in marketing, but tried to keep it to a minimum otherwise. It meant little more than that they all had a C compiler (but even then with subtle "improvements" of their own), and they had to have the same standard C library (but again, each felt free to make their own subtle "improvements" to it.)
What I'm getting at is: in a way the plethora of distros is even a good thing in that aspect. Noone is that secure at the top, or even king of the hill at all. (Not to mention they're all underdogs in the shadow of the 800 pound gorilla called Microsoft.) Noone is in a position to fork their version of Linux and try to lock customers in it.
Lock-in doesn't work when you're the underdog. The same fence that keeps your customers from escaping, also keeps you from reaching everyone else's customers. So noone does it when they have 10% of the market. At that point, you want open standards.
And with the current Linux market structure, we're pretty safe and secure that everyone will want open standards for the next decade straight. Unless MS manages to implode, anyway.
A polar bear is a cartesian bear after a coordinate transform.
When we saw the Unix fragmentation, we saw a bunch of different flavors that ran each on some kind of proprietary hardware (or a bit less proprietary, when it used an industry standard bus like VME) that were actively marketed by the makers of such hardware and were deliberately incompatible with each other in order to provide some measure of lock-in and differentiation on a largely common software platform.
If we ignore the vanity Linuxes (the ones someone did to claim they made one) and the specific-purpose ones (router-on-a-floppy, rescue, media-box) and the opportunistic ones ("let's nail some OEM deal to make some cash" kind) we are left with only a handful of very serious vendors pitching what amounts to be the same product plus some limited bells and whistles, that run on mostly any computer you happen to have, and making money out of supporting it rather than selling you disks (or tapes, if we account for those ancient times) and servers/workstations.
The difference is that I could not run the same binaries on my DG Aviion systems and on my IBM AIX boxes. I can install a Red Hat package on my Ubuntu notebook any time I feel like it (I usually don't)
http://www.dieblinkenlights.com
Earth has 1,250,000 species of animal. This is obviously a bad thing, and we should limit this to just 1 or 2 for the greater good!
Yes some Linux distros are a bit pointless, a fair few are redundant and some serve a niche that doesn't exist. But we actually need a large number of distros suited for different environments and in each niche the needs to be some competition to ensure quality.
A small list of niches off the top of my head:
Ideological (Debian)
Source based (Gentoo)
Business Server (RHEL, SUSE Server)
Business Desktop (RedHat, SUSE Desktop)
Home (Ubuntu, Linspire)
LiveCD (Knoppix, Morphix)
Router (LEAF, FREESCO)
Specialist (Musix, GNUstep)
Localised (Red Flag, this is really a whole extra dimension with server/desktop distros etc. needed for each local).
And that doesn't take into accounts preferences like Gnome/KDE, architecture, stable/bleading edge, security/easy of use etc, all of which can effect distro choice in any of the categories above.
300 distros against the mighty empire of Microsoft. This is insanity! No, this is Linux!
"Precisely how many distros there are is probably unknown"
How is that important, and who really cares?
It's a lot better than being roped into something you have no
way out of. At least with Open Source, you have options to
do things differently if they're not working.
"Ubuntu, which is clearly the flavor of the month "
Debian has been a favorite for a long time. It may be one
of the oldest distros. Ubuntu is merely icing on top of a
Debian based system. If you remove all the init 5 stuff, you
basically have a command line Debian system ready to be anything
you want it to be. As well as a robust update system and all
the great free stuff that makes linux so great.
Other distros follow this same paradigm. Centos, Fedora, Red Hat.
The underpinnings (since you are in an arcane mood) are the same,
It's the name that changes.
"Ah, so Linux is like a religion."
If you mean Linux is based on on the idea of something that works and
has a large following of people that understand it's advantages,
then yes.
"It is indeed true that the kernel hasn't forked in any significant way"
Other than XFree86, I haven't had any other forks impact me in the least.
And the xorg fork was a necessity. I think forking is good to the extent
that it drives people to come up with new ideas. The duplicating effort
argument I dont agree with. If we hadn't re-invented the wheel at least
once, we'd still be riding on round stones.
"There's no other way to put it: Linux is a forking mess."
And not under the control of a forking monopoly. Just because you find
duplicated effort in many different distros doesn't mean that's automatically
bad. You need to understand that people need to experiment. Distrowatch
is evidence of the experimenting people are doing. You should be glad
these people are putting alternatives out there for you. When you go to
write your column in Vista someday and you DRM key says your running a
pirated version and shuts you out, you'll consider it Linux again.
"So I'll grant readers that, if there's anything amiss with my argument"
Oh, there's plenty amiss. I think you got up on the wrong side of the
bed this morning. Everyone has bad days, it sounds like this is your's.
boycott slashdot February 10th - 17th check out: altSlashdot.org
The thing about all the different cars, cellphones, etc out there is that they still work in a similar way to each other.
The cellphones that don't have buttons laid out in familiar ways (eg the Nokia that had all the buttons in a circle like a dial, etc) never become mainstream, even if they may be better than the others.
Similarly, if you got into a car and instead of the ignition key there was a touchscreen on the dashboard, and the gears were shortcut keys built into the back of the steering wheel, then even tho this may be more efficient than the mechanical interface we're used to, it would be difficult to catch on.
In short:
- Any mac user can navigate their way around any other Mac desktop with ease.
- Any Windows user can navigate their way around any other Windows desktop with ease.
- The boon and curse of Linux is how configurable the interface is, and hence how different 2 desktops can be from each other.
(Unless you're the girl from Jurassic Park who can recognise "Unix" from a 3D file explorer).
The problem is not actual complexity, but rather perceived complexity. Joe User decides he doesn't want to pay $600 to upgrade to Vista, so he decides to look at Linux. What does he see: Red Hat, SuSE/Novell, Ubuntu, Kubuntu, Debian, Gentoo, Fedora, Slackware, etc.
What should he pick? Which is right for him? If they are all Linux then what is the differences? Is one as good as the next?
The problem is not that there is choice, but that there is too much choice. Most average users would rather just fork over the money and get Vista rather than spend hours, days, or even weeks trying to figure out what distribution of Linux to get, then installing it, then learning how to do actually use it.
What you and so many other people forget is that people are willing pay for familiarity and ease of use rather than accept strange, confusing and a learning curve for free.
There is no "-1 offended" or "-1 you don't agree with me" mod options for a reason.
Isn't anyone concerned that the forking of our current Time-Space-Continuum Thingy into the Multiverse will lessen the likelyhood of our stream being selected as the Earth Prime? Someone, please, stop all this forking.
Starting right now, everyone, stop making decisions.
O.K., now, don't think about, just stop...o.k. now, wait, no, now...on second thought....NOW.
Politics is the art of looking for trouble, finding it everywhere, diagnosing it incorrectly and applying the wrong fix.