next »

[khurt]

I've been looking into Open Source operating systems this morning, since I'm really only familiar with unix based systems such as Linux and Solaris. I also tried out Ubuntu on an old laptop not so long ago and was very impressed.

There are quite a lot of open source systems available, please advise me on some of the better ones. Thanks a lot.

No one appears to have answered your original question.  Everyone got carried away with promoting their favorite Linux distribution.  Linux is not an operating system is a kernel for an operating system.  That leaves you with two choices, OpenSolaris and BSD.  I have very little experience with BSD ( unless you consider OS X to be BSD ) and OpenSolaris sounds too corporate.

There an extensive list of Open Sources operating systems on Wikipedia.

http://en.wikipedia.org/wiki/Comparison_of_open_source_operating_systems

http://en.wikipedia.org/wiki/Comparison_of_open_source_operating_systems

[Square Bottle]

No one appears to have answered your original question.  Everyone got carried away with promoting their favorite Linux distribution.

Lol. Did you just not read really try to read the thread, or did you misunderstand that when the original poster asked for recommendations and people said what their favorites were, they were answering his question? I and some others actually recommended several based on level of computer experience and the suggested usage. So take your finger off the trigger, please.

And then wait a second! You go on to basically say that the only choices are OpenSolaris and BSD since Linux is just a kernel? Hehehe, I'm sorry, but there are plenty of fully featured operating systems that use the Linux kernel, and quite a few were specifically recommended. "Linux is a kernel" is no reason to disqualify Linux operating systems from being listed as options.

Lolcano.

[khurt]

Lol. Did you just not read really try to read the thread, or did you misunderstand that when the original poster asked for recommendations and people said what their favorites were, they were answering his question? I and some others actually recommended several based on level of computer experience and the suggested usage. So take your finger off the trigger, please.

And then wait a second! You go on to basically say that the only choices are OpenSolaris and BSD since Linux is just a kernel? Hehehe, I'm sorry, but there are plenty of fully featured operating systems that use the Linux kernel, and quite a few were specifically recommended. "Linux is a kernel" is no reason to disqualify Linux operating systems from being listed as options.

Lolcano.

I'm an idiot.  Accept my apologies.  A coworker "educated" me to the fact that while Linux is only a kernel that Ubuntu Linux or Slackware or Gentoo is an OS.  I agree with that.  My favorite open source OS is Red Hat Linux.

[NotSure]

Could someone explain  the differences and similiarites between BSDs and Linux distributions?  Do both make use of the GNU library for basic OS functionality?  I've always heard about Berkley but never actually given any of them a try.

Even though Square Bottle gave a good reply, it does miss one point. The future.

There is one difference between BSD and Linux. BSD is a Microkernel. Linux is a Macrokernel. In use they are not so different, but one is designed as a small safe OS that runs much of the less critical processes outside the Kernel, and the other is a huge fortress that runs everything behind a wall.

This sounds boring, but as time advances, macrokernels become bloated, slow and unresponsive. An example is Windows. Again, this may not have any effect on the user, but as the code becomes larger, it becomes harder to debug, and progress slows.

Personally I have no preference in OS, but I lean toward any that are microkernels. Solaris, BSD, HURD.

[Square Bottle]

Could someone explain  the differences and similiarites between BSDs and Linux distributions?  Do both make use of the GNU library for basic OS functionality?  I've always heard about Berkley but never actually given any of them a try.

Even though Square Bottle gave a good reply, it does miss one point. The future.

There is one difference between BSD and Linux. BSD is a Microkernel. Linux is a Macrokernel. In use they are not so different, but one is designed as a small safe OS that runs much of the less critical processes outside the Kernel, and the other is a huge fortress that runs everything behind a wall.

This sounds boring, but as time advances, macrokernels become bloated, slow and unresponsive. An example is Windows. Again, this may not have any effect on the user, but as the code becomes larger, it becomes harder to debug, and progress slows.

Personally I have no preference in OS, but I lean toward any that are microkernels. Solaris, BSD, HURD.

Ahh, the old microkernel versus macrokernel debate! Alright, since you've thrown in your two cents in favor of microkernels, allow me to offer a rebuttal.

The less places a system can break, the faster development can occur. This became very apparent when GNU Hurd was being developed. I heartily recommend the documentary "Revolution OS," in which Richard M Stallman himself concedes that his goal of developing a microkerel system ended up taking forever specifically because there were so many points of failure that needed to be addressed that debugging was practically impossible, and thus, the Linux kernel ended up being used to complete the "GNU package," so to speak (which is why so many people insist on calling it GNU/Linux, but that's a whole different issue).

So would a microkernel be better than a macrokernel? Yes, under ideal circumstances, I'd agree. However, the system that allows for the easiest and fastest development is the system that will get the most developers and will also be able to incorporate new ideas and features much faster. Due to having so many more possible points of failure since all the different components are separated to such a greater extent and need to be able to talk to each other and so on and so forth, microkernel development takes a huge hit in this area. If development for microkernels worked at the same rate as macrokernels, there'd be no issue, but the fact of the matter is that we now know that it's a huge issue for microkernels. This is actually partly why I'm so impressed with Mac OS X, which uses Darwin, a microkernel, and yet still is able to incorporate new features with each major release, which happens at a decent pace.

But yeah. So the basic reason for why I prefer macrokernels is that I believe that the system that can offer the fastest and easiest development is the system that will get all the cool abilities sooner. This has been the case with Linux, and is a very big part of why Linux is so competitive even with the steady, orchestrated, impressive releases from the BSD community and the well-funded private endeavors from Cupertino and Redmond.

At any rate, this is really about as nerdy of an debate as it gets (I've actually joked about this exact debate and used it as a humorous example of something that will almost certainly bore people in spite of the fact that I personally find it very interesting). The average user or even the power user will never, ever, ever need to know or care about microkernels versus macrokernels. Heh, especially me as a graphic designer.

But you asked, and NotSure is right. I should probably have at least noted macrokernels versus microkernels when you asked about differences between Linux and BSD. But again, unless you have the carreer goal of being an extremely low level programmer or something, then this will not affect your usage of either.

P.S. I didn't realize Solaris was a microkernel. That explains a lot about their development over the last several years, actually. Thanks for the tidbit.

[NotSure]

Do not get me wrong. I prefer microkernel, but I realise that the success of Linux to this point, including bringing in an army of programmers is the fact that it is easier to add to.

I just point out that almost every problem Windows is having today will be part of the Linux future in ten ro fifteen years. The only difference is in security. Security will be more difficult for Linux in the future, but not to the same degree that Windows has. Microsoft shot itself in the foot time and again by opening ports so it could attempt things like taking control of the net. It failed in their plans, and made their OS insecure with their efforts.

Each has their strengths and weaknesses. Macrokernels normally hit the ground running, and microkernels normally go the distance. This is why I refer to Linux as GNU-Linux, I see the FOSS platform being GNU... who cares what the kernel is, because we all are probably going to have to switch kerenel eventually.

[Square Bottle]

I disagree.

The problems that Windows faces today are because of a terrible foundation, lack or respect for industry standards and recommendations from hardware vendors, and closed source methodology. Linux has none of these problems, which is why it overcomes whatever problems it faces in a matter of months pretty much regardless of complexity.

Security will not be more difficult for Linux in the future. A lot of people seem to think that Linux is only secure because no hackers out there have given it their attention, but nothing could be farther from the truth. Linux handles system calls totally different. While Windows has all of its security at the user level (which is the same level where the viruses it's trying to protect itself from are), Linux is secured at all abstraction layers, which it makes very good use of.

Also, since so many web servers are Linux based, real hackers really do give Linux quite a bit of scrutiny. There are more script kiddies targeting windows, but they're not the ones developing anything or hunting for holes anyway. The real hackers are. And the real hackers out there can tell you why Windows is not safe, and why Linux is. Feel free to go over to www.binrev.com and ask for yourself. In fact, the guy attributed with defining what it means to be a hacker, Eric S Raymond, is a prominent Linux advocate and developer.

So when you say that Linux will encounter all the problems Windows has faced, I just disagree. The Linux community has already pointed and giggled at those problems, and typically uses them as examples of why people should use Linux, or why the open source development model is so much better. And to sum it all up, they're right.

But back to macrokernels versus microkernels.

Macrokernels can "go the distance" just as much as microkernels, and I'd actually argue that they "go the distance" better. Because it's so much easier to develop and debug, whenever anybody gets an idea for a new feature or a better way of doing something, they can pump it out pretty quickly, and then move on to another idea, and another idea, and another idea. By the time our hypothetical team of microkernel developers has gotten the first idea working with their system, the macrokernel team has finished with that feature and many others. So in terms of "going the distance," the macrokernel team gets significantly further ahead in much less time. On top of that, since it's easier to develop for, they will attract more developers at a faster rate, too, which will further increase the speed of development and put the macrokernel team even further on the microkernel team's horizon, and they'll just keep getting faster, too.

Again, if both microkernels and macrokernels could be developed at the same speed and difficulty, then microkernels would be better. But they simply aren't, and as such, I feel that macrokernels are better in practice.

However, I think Apple deserves a lot of respect because they use a microkernel and have a pretty competitive development rate. I think that Mac OS X is ahead of Vista or XP, and Microsoft's products use macrokernels after all. But this just goes to show that there are other factors that will impact the quality of a kernel. We can talk about those factors too, but we're isolating the microkernel versus macrokernel factor for this debate, heh.

[NotSure]

Then we disagree. But do you know what you are disagreeing to?

I do not follow with the "obscurity equals security" theory at all. In fact I can prove it incorrect by bringing up far more obscure platforms that were riddled with virii. Please do not assume that I believe in this "Windows Fanboy" theory.

I know that Linux can only get larger. More code. More legacy code. Over time the management will become a nightmare. With that much code, something will break sooner or later, and with the administrators of the Kernal stretched just trying to make new things work, security checks will lapse.

Microsoft did not stick to industry standards. This is true. They did this to break other peoples work for their own profit. I doubt that this would have lessened security. Security was deliberately sabotaged by Microsoft on more than one occasion. Everything else comes from handling a huge OS with so much legacy code that it now takes four years to beta test it properly.

You may find that your preconceptions about microkernels are not completely correct. If you look up the work of Carl Sassenrath, you may find that one person can put together a microkernel (well close enough for the time) in a couple of years if he has the talent. His work was not changed for over a decade. Only higher levels of the OS were changed. How many man hours has Linux had? It seems pretty certain that Linux has been successful because of all the man hours. What if HURD had those man hours instead? Hypothetical I know, but when Linux is compared to "Executive", the hypothetical Linux team you mentioned are bogged down trying to make it all work.

You seem to be comparing only HURD or only BSD to Linux. You are not comparing the successful Linux with a successful microkernel like Mach (there are more mach users than linux) or Executive. You are also not comparing the real man hours put into these kernels. The total man hours make a mockery of the claim that a macrokernal is and always will be faster to develop.

[Square Bottle]

I'm glad to hear that you aren't relying on the "security through obscurity" argument, but there are a couple incorrect facts and unfair conclusions that you've made.

First off, you talk about how Linux will get bigger over time as a reason why it will fail and why microkernels won't. This is unfair because microkernels will also grow overtime. They are not called "micro" because they're smaller, but because they are divided up into smaller parts.

In your next paragraph, you talk about Microsoft a little bit. First, you claimed that deviating from industry standards (and presumably hardware vendor recommendations) does not lessen security. In fact, it does. Many standards are made to ensure secure operation, and are the result of people noticing that doing things in certain ways produces holes. In the same paragraph, you go on to talk about how Microsoft struggles with handling a large operating system with lots of legacy code, which is irrelevant. Operating systems are always much larger than kernels, which are quite tiny by any comparison. The fact that Microsoft severely bloated and mismanaged the whole operating system speaks only of their production model anyway, not that I think their kernel is good. At any rate, Linux operating systems are considerably more trim, and have been around for a sufficiently long time that we would have already seen it get crippled by too much legacy code as you claim it will be.

Next, you talk about a single developer. I agree that there are fabulous developers out there that can make all the difference, but if we took Carl Sassenrath and applied him to a macrokernel's development, he'd get things done quicker too. I already talked about Richard M Stallman and his original GNU team too, who I'd also consider to be quite outstanding, and even they admitted that trying to use a microkernel was the reason why Linux's development occurred at such a faster rate. And actually, let's talk about that a little bit more. You want to talk about man hours? Compare Linus Torvalds to the entire GNU team then. In spite of having a considerable head start on Linus, he was able to get a functional kernel designed all by himself way, way, way before them. And then when he showed the world what he'd been fiddling around with, it really took off. Since then, development has only been getting faster and fater, with more developers contributing now than ever. So while say that Linux should be slowing down anytime now, well, it isn't.

Then you go on to try to say that I'm cherrypicking comparisons (which is amusing because you provide multiple possibilities of what I could be talking about, and yet leave out the only microkernel I specifically mentioned, which would be Darwin). Everything I've said so far has been pretty general though to the concept of microkernels though. And in any case, I actually directly compared it to the most successful microkernel operating system of all time, Mac OS X. I don't know where you got that there are more Mach and Executive users than Linux users, but there are definitely more Darwin users, so actually, I did talk about a successful microkernel project at length, and I even made note that I was impressed with how competitive Apple's development team was. However, when it comes to releasing early and releasing often, the fact of the matter is that no group or person has been able to come anywhere near the Linux project.

Last be not least, you say that I'm not comparing man hours. Er, I don't think so, but okay. If I'm not comparing man hours, then you aren't comparing results. Development happens much faster for macrokernels than microkernels, and that's all there is to it.

[NotSure]

Let us compare notes. I admit from the outset that there is a place for macrokernels and that people should use any kernel that suits them. I point out that there is a place for microkernels as well. You refuse to admit that there is a place for microkernels, and then you try to claim that microkernels are slow to develop and are hard to debug. I then state the obvious and point out how Linux (I did not even bring Windows man hours into this argument) has had more man hours of development than any microkernel, and that one successful microkernel was created by one man in two years. Then you claim that Linus (who admits he cannot code at a proficient level) designed Linux on his own.

Linus could design the kernel on his own, but he needed a large team of programmers working for 15 years to make something of it. Carl Sassenrath did the design and the programming on his own.

Now I am not claiming that Carl Sassenrath could not create a macrokernel faster than Linux. He almost definatly could if he wanted. Carl Sassenrath does not believe in macrokernels, so it is doubtful he would ever try. His achievement on one microkernel is only used to prove that microkernels are not "faulty by design" (paraphrasing you). If all microkernels were bogged down with technical problems, delaying the release date, then Carl Sassenrath should have taken far longer than the 15 years the dozens-hundreds of programmers took to make Linux a usable OS for the general population.

Debugging a macrokernel is much harder than debugging a microkernel. Ask any programmer if he prefers debugging a modular system or one huge chunk of code. Linux can currently be crippled by one bad device driver, or have a security issue because of one device driver. A microkernel does not have this problem at all. A driver can do whatever it wants outside the kernel, the kernel is protected from the random behavior of drivers. Which does bring up another interesting point. Linux, as a GPL macrokernel is self sabotaging when it comes to drivers. If Linux ws a microkernel, then the drivers would not need to be GPL to be built into the kernel. Hardware developers would be able to create closed source drivers with no GPL problems. Drivers are not the only problem Linux has and is facing, but it is a problem that would have never happened if it were a microkernel. Yes, all OS kernels grow (excluding Executive that was never changed), but how much do they grow, and how large are they to start with? Would you prefer to debug a 20000 line OS that had grown 50% in 10 years, or a 500000 line OS that had grown 50% in 10 years? I bet there will not be a lot of volunteers to debug Linux in a decade.

Now, can you just accept that one is not inherently better than the other? Early development speed is VERY important in the success of an OS. Linux has succeeded because of this advantage. Long term, the evidence would favor the microkernel. The great thing about all of this, is that the programs are running on GNU, and GNU can run on a microkernel and a macrokernel. So we can all use Linux today when the sun is shining on Linux, and we can all switch to HURD or similar in ten years time when the sun is shining on their camp. There is no need for animosity.

[Square Bottle]

Easy there, chief. You're making it sound like I never said anything good about microkernels when in fact I'm a big fan of Mac OS X and FreeBSD, and have declared such. This debate is about which type of kernel we think is better though, and I'm going with macrokernels.

Also, I tried to note before that actually, GNU Hurd had a huge headstart on Linus Torvalds, who _did_ start out all on his own while going to school in Finland. And it didn't take 15 years for something to come of Linux. Obviously it wasn't for the average person for a while, but he wasn't exactly the only person interested in his project for fifteen years.

Please do not paraphrase me as saying microkernels are faulty by design. I never said that they are inherently faulty, nor did I imply such in any way, so that's kind of a grossly inaccurate thing to imagine me as saying. I even pointed out a few microkernels that I liked, and even specifically said that if development occurred at the same rate and at the same difficulty, they'd be better.

Carl Sassenrath is a great programmer. This does not change the fact that microkernels take longer to develop than macrokernels. You can go ahead and ask Carl Sassenrath if he thought what he did was easy, if you like. I can wait, heh. Or, you could watch the documentary "Revolution OS." Please do if you're interested, as this stuff definitely comes up (and it's very watchable, too).

Debugging macrokernels is certainly easier than debugging microkernels. The whole reason why microkernels are harder to develop for (again, go and see "Revolution OS" because even Richard M Stallman admits that this is why GNU Hurd lost out to Linux) is because thee modules are not quite so independent of each other. See, the problem they faced (and one more time, you can watch the documentary yourself, chief) is that if any one of the separate parts of the microkernel didn't work, the whole thing would go kaput, and if they all worked independently but didn't talk to each other right, they'd still go kaput, and it was hard to tell which was the case and really, really hard to find out where the problem was, especially if it was more than one thing at once, as it often was... Etc etc. You're thinking of modular in terms of independent parts. Microkernel has separate parts, but they all need to work together, so it's not the same kind of modular at all as when you think of extensions for Firefox, for example.

Next, you talk about licensing. This is not relevant to whether macrokernels are better than microkernels or vice versa, so I will not go into detail here except to say that I agree with the Linux community the closed source Linux kernel modules would be immoral, impractical, and thankfully illegal.

I'd also like to note that I'd like to take you up on your bet that there will "not be a lot of volunteers to debug Linux in a decade."

Can I just accept that one is not inherently better than the other? Well, if I remember correctly, I jumped in to defend macrokernels from you! But the debate is about which concept is better, and I think that the macrokernel concept is better than the microkernel concept because of how development is easier and faster. There is no animosity at all, least of all toward you, my friend! I consider this a scholarly debate, and I'm enjoying it quite a bit. I myself use Linux when I have a choice, but I'll also be very happily using Mac OS X when I head off to art school. But yeah man, I'm enjoying myself, and I hope you're enjoying yourself too. Where else can we debate the merits of macrokernels and microkernels, really?

Edit: Here is the trailer (well, first eight minutes, actually) for Revolution OS. Assuming you're interested. And I hope you are. Because. It's. Good.

<a href="http://www.ifilm.com/efp?flvbaseclip=2421280" target="_blank">http://www.ifilm.com/efp?flvbaseclip=2421280</a>

[lm8]

Since I didn't see Reactos mentioned in this thread, I just thought I'd add it.  It's still pretty much in the works.  However, if they keep advancing in the area of binary compatibility with Windows programs, it should eventually become a very nice Open Source option for Windows users.  I've also read that the Reactos team is sharing code with the Wine team.

Since the thread covered some versions of Linux, I'll mention Absolute Linux for older computers with less resources (though it works fine on newer ones as well).  Also, there's DeLi Linux if you need something for an even older machine.