On November 23 a new relase of Minix sneaked out. To make things a bit more interesting it looks like there may well be more releases to come: “Version 2.0.4 has been slapped together in a hurry to have a fixed point to start a new Minix project at. The code should be OK, but most new things haven’t been tested. (So there’s already a fix out.) The documentation is alas far behind all the changes.” A Bochs image is also available (in the i386/DOSMINIX.ZIP, use the minix.bxrc to start Bochs).
New Release of Minix?
Submitted by DanO 2004-01-14 OS News 27 Comments
I thought that this OS had died out. Interesting.
Is this Tanenbaum working on Minix or did somone take it over?
Kees J. Bot (firstname.lastname@example.org) is the only name I could find on both the site and in the chnages.txt
I think it got released under the bsd license. The original license was sort of restrictive, so now you can fork and so forth.
At least i have something to run on a 286.
how this compares to Linux 1.2 let alone 2.6? 😛 But seriously thats cool, I have my doubts hardware support is decent at all – will it run on newer hardware? If not Virtual PC here it comes. I’ve long wondered what it felt like…hmmm. I too am curious if Tanenbaum is working on it, is it still microkernel based?
I took a quick look through comp.os.minix, it appears that Andy Tannenbaum and one of his students are doing this new project. Apparently they are going to be putting more kernel tasks into userspace and making the OS more modular.
Kees J. Bot said in one post:
“It’s to try to minimize the kernel by turning all the tasks into separate little processes like MM and FS. I have my doubts that this can be done with any kind of efficiency, but we’ll see.”
Later he said:
“The project is the idea of Andy and one of his students. I’m not personally involved (yet?). I’m not even convinced that it’s such a good idea, because of the huge overhead that lifting the tasks out of the kernel will cause in message passing.”
The above link is a famous usernet post from Jan of 92 from Andy Tanenbaum. For the few of you that have not yet read it anyway.
PS. Title is Andy’s, not mine.
Yeah, so? Tanenbaum was right back then. His remarks were right for the Linux of 1992, not of today (Linux has evolved since then), so there was no reason to link this at all.
I love reading very old stuff like that. The comparisons between then and now are fasinating. He derides linux for its portability and talks about RISC processors replacing x86, yet not only has Linux become the paragon of portability, but x86 has remained the lead CPU architecture in desktops and has even been moving in on low to mid-range servers like crazy.
My endless rants on the Minix newsgroup may actually have finally paid off in some way… not sure. Of course, since Kees J. Bot now hates me and won’t talk to me and no one else on that group will either, it’s nice to see a site like this can help me keep tabs on that “old fart OS”.
I don’t feel Linux is obsolete, hence “PS. Title is Andy’s, not mine.”
I added the Link because that conversation is historic and a very interesting read. I’ve taken no side in the arguement but I do toss my 2 cents into the microkernel camp.
Well, over here at the Free University where Tanenbaum lectures we still use Minix for practical assignments. Great fun for taking your first steps into hacking away at an OS. Having said that, Minix does have some oddities we love to whine about, but I do recommend everyone interested in an OS’s internals to get his books; they’re well-written, in-depth and with characteristic witty remarks that will keep you reading.
And no, I’m not getting a higher grade because I’m posting this
Minix and Minix-VMD are BSD licenced since 04/2000
since the version 2.0.3 it doesn’t run on XT
Minix have been
Minix is much smaller than Linux and it might be interesting to port it to PDA and other small devices (like phones, watch, …)
“Yeah, so? Tanenbaum was right back then. His remarks were right for the Linux of 1992, not of today (Linux has evolved since then), so there was no reason to link this at all.”
His remarks where *not* right for Linux at the time, and they are not right, now.
Just read the bloody posting before commenting.
If the remarks where “right” at the time, then they would still be right, since they STILL hold. But that’s not the case.
Andy was going on from an academic point of view. He lost.
Minix is of very little practical use in the modern world, with competition like Linux, the BSDs and various Unix clones.
To even begin to meet the performance of its rivals would take a complete redesign of the entire kernel. (for the tech-heads who know a little about the OS: this is to eliminate the very crippling FS bottleneck, preventing any useful form of multitasking) And why would you bother? If you’re going to do all that work, you’d be better off redesigning the entire OS and coming up with something innovative that might stand a chance of competition.
As it is, Minix is nothing more than a curiosity for the academic world.
Minix never died, it has been very much alive as a teaching tool, and to some extent as a real working OS. As a working OS it is, of course, not really a competitor to Linux, but it is usable on old hardware that is not able to support Linux. There have been experiments with handhelds and other odd platforms. And some people just like the idea of using a system that they can potentially understand completely.
The official Minix site remains Andy Tanenbaum’s page:
Some time ago Andy and I agreed that my own sites would host
the official Minix FAQ. The FAQ is maintained as a set of
web pages at http://minix1.bio.umass.edu/hints.html and is mirrored at http://minix1.hampshire.edu/hints.html
My minix1.* sites run on standard Minix, as a demonstration and test of its usefulness as a real OS.
There are likely to be more interim releases of Minix in the next year or two, leading up to a new major release with the
publication of a third edition of the Operating Systems Design and Implementation text. I am currently working on this but can not say when publication will occur.
>His remarks where *not* right for Linux at the time, and they are not right, now.
>Just read the bloody posting before commenting.
I’m not so sure *you* actually read the postings.
[…], suffice it to say that among the people who actually design operating systems, the debate is essentially over. Microkernels have won.”
“True, linux is monolithic, and I agree that microkernels are nicer. With a less argumentative subject, I’d probably have agreed with most of what you said. From a theoretical (and aesthetical) standpoint linux looses.”
“MINIX was designed to be reasonably portable, and has been ported from the Intel line to the 680×0 (Atari, Amiga, Macintosh), SPARC, and NS32016. LINUX is tied fairly closely to the 80×86. Not the way to go.”
“The very /idea/ of an operating system is to use the hardware features, and hide them behind a layer of high-level calls. That is exactly what linux does: it just uses a bigger subset of the 386 features than other kernels seem to do. Of course this makes the kernel proper unportable, but it also makes for a /much/ simpler design.”
In other words, Linus agrees Linux (in 1992) was unportable.
So, Linus himself agrees with the two main issues Andy made. If you say Andy’s remoarks were /not/ right for the
time, then you’re disagreeing with Linus himself.
>If the remarks where “right” at the time, then they would still be right, since they STILL hold. But that’s not the case.
Eh? Linux was not very portable in 1992, but now it is.
Andy’s remark was true then, but it’s not true now.
Regarding CISC vs RISC: most modern CISC chips have a RISC
core, using a decoder to turn each CISC instruction into
one or more RISC instructions. So, in a way, Andy was
right about RISCs performance advantage over CISC.
This line from Andy still cracks me up:
“As an aside, I congratulate you for being able to write a POSIX-conformant system without having the POSIX standard in front of you. I find it difficult enough after studying the standard at great length.”
The comments in the old thread are also somewhat applicable to linux of today. Part of it was the old microkernel vs monolithic kernel. Also debates on cpu needed (minimum 386), licensing, etc. Linux ended up working in the end.
Tanenbaum’s defense later was that his was an academic OS. This is sort of right, it was meant for teaching concepts. I’ve heard a true academic OS is not really for the masses anyway. The good ideas though trickle down into the mainstream OS eventually. The bad, er, impractical ideas fall by the wayside.
As i understand it, Minix, at least in its original form, is simple enough for one to grasp most all of it. Would be a good tool … reading Tanenbaum’s book while tinkering with it.
“As it is, Minix is nothing more than a curiosity for the academic world.”
People just don’t seem to realise that Minix *was* an
academic OS. That was the point of it – it was a
functional unix clone that was simple enough for students
to understand, not a high-performance system.
Its goals were entirely different to those of Linux.
See “Operating Systems: Design and Implementation” for more
on the history of Minuix.
Please stop comparing a real world os to an academic one. dont complain when every new os is released that it cant compete with windows linux or whatever.
1) the goals might be different
2) you might be entirely wrong. remember when linux was first released?. would u have guessed it would come this far?
HA HA HA….just goes to show thatbeing an expert does not make you a fortune teller 🙂
he was right abotu RISC….x86 is now RISC based as are every other Chip on the market….he was right at the time about Linux, but now, his comment about Linux portability is just wrong…..he was super wrong about micro v monolithic kernels…..I would say that neither won, they both merged….Linux is a monolithic kernel with modularity, Windows is a micro kernel that morphed into a monolithic kernel, really, OS X is the only main stream OS that is a true micro kernel based system.
I guess SCO found IP in Minix as well forcing them to rewrite from scratch… LOL
Not even OS X can be labeled as a microkernel OS anymore, the fact that it runs a BSD cuasi-kernel in user space make the lines blurry. But there is QNX, and lately the L4 family as examples of good modern implementations. The theoretical battle between mono & micro is far from over and I guess it will never be. The two lines of thought have many good points on its side to stand up and even coexist or merge in some cases.
“Not even OS X can be labeled as a microkernel OS anymore, the fact that it runs a BSD cuasi-kernel in user space make the lines blurry.”
The xnu BSD server runs entirely within the kernel, actually, which makes it even less of a microkernel.
The Minix/Linux competition is obsolete.
Minix is a full operating system running very well on low capacity computer. It is not build to use all the performance of a modern pc.
I prefer to use Minix on 486 and for pentium and up, i use Linux or *BSD.
I don’t see any future for Minix for the modern pc and i didn’t heard about any team that would adapt Minix to use all the hardware and performance of a modern desktop.
Minix should have a future on smaller system – PDA, watch, microcontrol, …
regardless of the linux/minix arguments
If it wasn’t for the fact that minix could be transported from Uni to my digs on a couple of floppies, and that it ran on the ropey but free 286 I dragged off a skip so I could learn the advances of ls,du,dd and vi (and deleting things in /etc breaks stuff).
I very much doubt I’d be typing this on OpenOffice/Mandrake/free P2 laptop I dragged off a skip. (somthings stay the same)
I am glad there are people that keep a part of history alive. I think its a worthwhile endeavor. I think i am going to check it out when i get some hardware to run it on.