For anyone who doesn't know: Minix is not for use as a daily personal desktop. Its primary function since inception has been to provide a small, bare-bones OS for students to study and modify for themselves to learn low-level OS architecture. Anything extra was an after-thought. Comparing its application library or functionality to modern OSes like Windows, OSX, Linux, or x-BSD is a moot point because it's not trying to be like any of those.
This is a great & superb explanation of a new modern operating system architectture!!!. Long live to Minix and just hope with every new version doesn't follow the 'bloating' trend too. Thanx for the information Mr. Tanenbaum.
Did he say half way in that this presentation was recorded in 2009? Its very insightful to see the parts of this research that are more relevant than ever to the refinement of computational science and its role in humanities future, whilst showing lineage to projects that have made head way since then (I'm thinking GUIX, modern init systems like systemD/GNU Shepherd, and research into dependently typed languages, and the idea of type universes, type fields/rings etc )
since in Linux we don't have this architecture of microkernel in Linux we have to use virtualization and containers and orchestrators to get the same kind of recovery semantics or simply just code in Erlang
Linux maybe the shiznit now, but I learned Operating Systems on Minix that was on the back of AST's book. It was simple enough to grok. 10,000 lines of code for the whole thing. Give or take. and you could install it on an 8088 (that's like running a whole OS on a friggin Arduino++ for chrissakes!!!)
I believe this concept will eventually be ironed out as a system of implementable rings that depend on each others axioms like the idea of type fields/type universes (U, U+1 etc) in dependently typed languages, and of course compiled to the CPU'S rings through micro code or w/e. Is RISK V doing something like this already, I wonder?
@montaTheNuts they have to rewrite the wheel for creating the vga out for minix, so they used windows rather than spending days trying to make a vga out driver.
it would be neat to have the message cache keep messages around in case the reincarnation server has to kill a process.. combined with the data store, this could mean that the process is more likely to be able to pick up where it left off. Presumably there would be a mechanism to denote if a message has been responded to (or expects a response) to begin with. The whole thing reminds me of scala's actor (akka) model.
Lol some of you just don't get it. Performance vs. reliability. 1. MINIX is intended for embedded systems, not for you to run your WoW. A software environment in a bus, for example, should never crash. And it doesn't really require more than like 128 MB of memory for this task. But it wouldn't matter, because 2. Memory and processing power nowadays is cheap, especially if we consider an industrial investment. So what?
Unix system V is a real operating system. Likewise so is this. And yes you can actually run WoW. You missed the part where he talks about Win98 being installed on Minix and it was working flawlessly. Likewise so should WoW. Or more specifically Wine. I.e. it's all there.
A lot of theses technologies are in BeOS and Haiku if I've correctly understand ! But if one day Minix 3 can run firefox or another HTML5 capable browser, I think I will try it seriously on real Hardware. This OS seems to feel all the holes I hate on current OS models.
also kernel crashes are not the biggest problem. the DE freezing can force the person to restart, even though the kernel is still sticking. in two the only time I got a kernel crash on my mac was from a crappy FOSS ntfs-3g kext module. go figure.
@roopeshvaddepally Keep in mind that Minix is lacking drivers. That means you cannot run it on any random laptop, until you have ported all the necessary hardware drivers to Minix.
@triplesquarednine You have to think about application, a nuclear reactor can not be out of control for even a second for reboot or handover. That is why he is talking for industrial controls
At 45:00, the Chrome question is little wierd because the Chrome OS use Linux kernel as the OS. Different name for the system but the OS is well know but hided from wide public.
According to another video posted by his school, they have Solary, Linux, and Windows. He teaches distributed computing in the school. Thus, he uses some unix-like systems in his teaching. You may right, it is a irony, but I think the irony is FOSDEM's meeting room provides WINDOWS.
The current Intel lineup of processors run too hot. MacBook's and other laptops eventually heat up in your lap. Then the fans run at full speed, and then I have to turn up the speakers. Slower hardware is better.
While I agree that Linux has become bloated and reliability in GP OS is lacking, the analog to televisions and submarines, these are apples to oranges comparisons. The difference with a GP computer system is that it is not, by definition, an appliance. Instability often occurs from installation of software with bugs, driver incompatibilites, etc. Note Andrews comment on the instability of Windows due to drivers. Appliances like Cisco network devices and Linux-based TVs are more reliable.
It may be right that Linux is bloated and huge. But Microkernels are much too complicated either. Of course drivers has to be written properly but I prefer kernels like Linux or FreeBSD with many working drivers to those ones(Minix), where drivers(graphics driver, network cards driver) doesn't even exist. And I don't no if Minix is as stable as Linux. I never had a kernel panic on Linux. And I'm using it for allmost a year.
You run `./configure' and it always fails. /Always/. You know? What happens? It was looking for, I dunno, Perl 5.2.3.26.9.4b, and if that wasn't there it gives up--even though the application you're trying to install doesn't use Perl at all. ROFLMAO
Systemd is its own world. The problem now is I don't think anyone understands what is in Linux or if bad actors can find a way to crash, hack or backdoor it. Let's not forget about the spectre bug as well. I really think Minix is inevitable.
***** That's because he's American (he's from New York). Not sure about the Jewish part (though that name sounds German to me). Do Jewish Americans sound different from other Americans?
+Nave Wilson He can be American Jew, American Christian, American Muslim, American other; are you hung up on Religions or do you think they turn into a race of people?
***** lol. Learn "how"? That there's less bugs in a project many thousand times smaller than yours, developing at a completely inferior rate, mainly focusing on just stability? MINIX was started in 1987. Linus used MINIX himself, but was disappointed in it. He released Linux in 1991. Tanenbaum critized it, calling it a giant step back to the 70s. A few years later, Linux had a vastly larger user base. Now with Android, it's the most popular kernel in the world. Performance and features top stability. They make things interesting for our species itself. That's evolution. Tanenbaum even said it himself (47:20): | _"Performance is not put on our agenda."_ | Yeah. We could all use a raw X11 based system with a bunch of xterm windows and a tiny/stable micro kernel running it all, or we could just live in the year 2015, and leave that to systems and cases where anybody needs it. If you're looking for a dead stable Linux, then you don't have to keep up with the absolute latest kernel versions. There's already distros like Debian that focus on stability, with years old software, and who only accept bug and security fixes.
Det WTF are you even ranting about? I made this comment a year ago and I don't give a shit about the subject. Linux is just a shit OS used by a few dozen loud mouth open sores clowns like you,
***** comma, and...? Were you going to conclude that angry post how? Like I said, genius, Linux is currently the most popular OS in the world :). A lot of people have loud mouths, but I hope yours only opens up like that in the internet.
only morons define crashes as kernel panics. crashes to a normal person is the unresponsive desktop, where they are forced to restart. andrew has to wake up and walk outdoors into the world after 2000. Kernel crashes are very uncommon. No gradma is willing to sacrifice performance to avoid a single kernel crash every three years.
There are so many more useful implications to this research than just 'we only need one kernel that doesn't crash'. Simplicity is a cornerstone of trust and trustlessness, which our world desperately needs more understanding of
I get a bad impression from this waste of euros. The whole O/S and all mentioned software is truly archaic -- decades behind. The trade-off of performance for some nebulous reliability claims is unacceptable; I don't remember my Linux servers EVER having crashed. Reminds me of a programmer I used to work with whose solution was to use a `for' loop to retry any code that had an error.
For anyone who doesn't know:
Minix is not for use as a daily personal desktop. Its primary function since inception has been to provide a small, bare-bones OS for students to study and modify for themselves to learn low-level OS architecture. Anything extra was an after-thought. Comparing its application library or functionality to modern OSes like Windows, OSX, Linux, or x-BSD is a moot point because it's not trying to be like any of those.
if you have an Intel CPU with ME then you already have MINIX running on your desktop
This changed since they now have NetBSD userland over minix. So maybe minix could become viable as a daily driver if enough people use it.
This is a great & superb explanation of a new modern operating system architectture!!!. Long live to Minix and just hope with every new version doesn't follow the 'bloating' trend too. Thanx for the information Mr. Tanenbaum.
Did he say half way in that this presentation was recorded in 2009? Its very insightful to see the parts of this research that are more relevant than ever to the refinement of computational science and its role in humanities future, whilst showing lineage to projects that have made head way since then (I'm thinking GUIX, modern init systems like systemD/GNU Shepherd, and research into dependently typed languages, and the idea of type universes, type fields/rings etc )
It really is a very interesting presentation!
It's really great for all the os developer as well for com.sc. students.
since in Linux we don't have this architecture of microkernel in Linux we have to use virtualization and containers and orchestrators to get the same kind of recovery semantics or simply just code in Erlang
I thought Andy Tannenbaum was a dutch guy... I always imagined him having this dutch accent... lol ... nice talk.
Linux maybe the shiznit now, but I learned Operating Systems on Minix that was on the back of AST's book. It was simple enough to grok. 10,000 lines of code for the whole thing. Give or take. and you could install it on an 8088 (that's like running a whole OS on a friggin Arduino++ for chrissakes!!!)
also printer drivers don't run in kernel space. i'm amazed no-one from the audience pointed it out to him
This would be even better if we still had Ring 2 and Ring 3, instead of just Rings 0 and 1.
I believe this concept will eventually be ironed out as a system of implementable rings that depend on each others axioms like the idea of type fields/type universes (U, U+1 etc) in dependently typed languages, and of course compiled to the CPU'S rings through micro code or w/e. Is RISK V doing something like this already, I wonder?
so good, lots of inspiring info !
@montaTheNuts they have to rewrite the wheel for creating the vga out for minix, so they used windows rather than spending days trying to make a vga out driver.
it would be neat to have the message cache keep messages around in case the reincarnation server has to kill a process.. combined with the data store, this could mean that the process is more likely to be able to pick up where it left off. Presumably there would be a mechanism to denote if a message has been responded to (or expects a response) to begin with.
The whole thing reminds me of scala's actor (akka) model.
Lol some of you just don't get it. Performance vs. reliability.
1. MINIX is intended for embedded systems, not for you to run your WoW. A software environment in a bus, for example, should never crash. And it doesn't really require more than like 128 MB of memory for this task. But it wouldn't matter, because
2. Memory and processing power nowadays is cheap, especially if we consider an industrial investment.
So what?
Unix system V is a real operating system. Likewise so is this. And yes you can actually run WoW. You missed the part where he talks about Win98 being installed on Minix and it was working flawlessly. Likewise so should WoW. Or more specifically Wine. I.e. it's all there.
Nice description of the joys of GNU "Autohell" (TM) - 38:39
A lot of theses technologies are in BeOS and Haiku if I've correctly understand !
But if one day Minix 3 can run firefox or another HTML5 capable browser, I think I will try it seriously on real Hardware. This OS seems to feel all the holes I hate on current OS models.
also kernel crashes are not the biggest problem. the DE freezing can force the person to restart, even though the kernel is still sticking. in two the only time I got a kernel crash on my mac was from a crappy FOSS ntfs-3g kext module. go figure.
I was a student of you though your book, AST.
really good presentation!
It's nice to know that a sailor taking a massive shit won't trigger world war three....
Microkernels FTW.
respect
@roopeshvaddepally
Keep in mind that Minix is lacking drivers. That means you cannot run it on any random laptop, until you have ported all the necessary hardware drivers to Minix.
@triplesquarednine You have to think about application, a nuclear reactor can not be out of control for even a second for reboot or handover. That is why he is talking for industrial controls
To get perfectly reliable software you need for formally prove the entirety of the code and use, though this seems to come closer.
At 45:00, the Chrome question is little wierd because the Chrome OS use Linux kernel as the OS. Different name for the system but the OS is well know but hided from wide public.
According to another video posted by his school, they have Solary, Linux, and Windows. He teaches distributed computing in the school. Thus, he uses some unix-like systems in his teaching. You may right, it is a irony, but I think the irony is FOSDEM's meeting room provides WINDOWS.
Doors too.
This post was written 11 years ago. Android is actually encroaching on the Windows space rapidly.
I remember the time before HD TV or color phones like it was yesterday.... Hmmm... or Maybe it was yesterday.
~
It's sad that the TV is becoming more like a computer instead...
@catalintgjiu I'm pretty sure they have a good relationship!
My car is like that, usually it wants a new belt, fluid, or a new engine. of course its a FORD "Found On Road Dead" vehicle.
"First On Race Day"
The current Intel lineup of processors run too hot. MacBook's and other laptops eventually heat up in your lap. Then the fans run at full speed, and then I have to turn up the speakers. Slower hardware is better.
2:38 European Union ERC Advanced Grant
so is the arm port finished or even started yet? btw appreciate the talk.
why cant we use an
itu t v44+IPR=921600 v92 modem at the speed the manufacturer designed it to work at?
@Darr1066 Minix is a microkernel whereas Linux is a monolithic kernel. It's sort of apples to oranges.
problem is that no graphical browsers have been ported to MINIX 3
While I agree that Linux has become bloated and reliability in GP OS is lacking, the analog to televisions and submarines, these are apples to oranges comparisons. The difference with a GP computer system is that it is not, by definition, an appliance. Instability often occurs from installation of software with bugs, driver incompatibilites, etc. Note Andrews comment on the instability of Windows due to drivers. Appliances like Cisco network devices and Linux-based TVs are more reliable.
@roopeshvaddepally, I'm pretty sure that doesn't count as irony.
We can tell who thumbed down this vid.
what is the problem with that ?
It may be right that Linux is bloated and huge. But Microkernels are much too complicated either. Of course drivers has to be written properly but I prefer kernels like Linux or FreeBSD with many working drivers to those ones(Minix), where drivers(graphics driver, network cards driver) doesn't even exist. And I don't no if Minix is as stable as Linux. I never had a kernel panic on Linux. And I'm using it for allmost a year.
@roopeshvaddepally Perhaps he needs software that hasn't been developed for Minix yet?
You run `./configure' and it always fails. /Always/. You know? What happens? It was looking for, I dunno, Perl 5.2.3.26.9.4b, and if that wasn't there it gives up--even though the application you're trying to install doesn't use Perl at all.
ROFLMAO
7:40 and what happens in Linux? Systemd got introduced 🤣
Systemd is its own world. The problem now is I don't think anyone understands what is in Linux or if bad actors can find a way to crash, hack or backdoor it. Let's not forget about the spectre bug as well. I really think Minix is inevitable.
a really old version
the only non-fosstard unix developer.
i clapped at 43:35
*****
That's because he's American (he's from New York). Not sure about the Jewish part (though that name sounds German to me). Do Jewish Americans sound different from other Americans?
+Nave Wilson
He can be American Jew, American Christian, American Muslim, American other; are you hung up on Religions or do you think they turn into a race of people?
These fucking configure scripts!
irony: minix 3 creater using windows
Linus Torvalds should learn from this.
I'm sure he has, him and Tenenbaum apparently still friends.
***** lol. Learn "how"? That there's less bugs in a project many thousand times smaller than yours, developing at a completely inferior rate, mainly focusing on just stability? MINIX was started in 1987. Linus used MINIX himself, but was disappointed in it. He released Linux in 1991. Tanenbaum critized it, calling it a giant step back to the 70s. A few years later, Linux had a vastly larger user base. Now with Android, it's the most popular kernel in the world. Performance and features top stability. They make things interesting for our species itself. That's evolution.
Tanenbaum even said it himself (47:20):
| _"Performance is not put on our agenda."_ |
Yeah. We could all use a raw X11 based system with a bunch of xterm windows and a tiny/stable micro kernel running it all, or we could just live in the year 2015, and leave that to systems and cases where anybody needs it. If you're looking for a dead stable Linux, then you don't have to keep up with the absolute latest kernel versions. There's already distros like Debian that focus on stability, with years old software, and who only accept bug and security fixes.
Det WTF are you even ranting about? I made this comment a year ago and I don't give a shit about the subject. Linux is just a shit OS used by a few dozen loud mouth open sores clowns like you,
***** comma, and...? Were you going to conclude that angry post how?
Like I said, genius, Linux is currently the most popular OS in the world :). A lot of people have loud mouths, but I hope yours only opens up like that in the internet.
Det If by the "world" you mean your "warped delusional mind" than I agree. Otherwise STFU asswipe.
hahaha well that says alot :D
zomg you are using windows and microsoft office for the presentation? ahahah your minix can't do that?
No GPL? No thanks.
Best throwing MINIX TV Android box in trash disgusting
118 people appreciated ,Linus didn't :)
only morons define crashes as kernel panics. crashes to a normal person is the unresponsive desktop, where they are forced to restart. andrew has to wake up and walk outdoors into the world after 2000. Kernel crashes are very uncommon. No gradma is willing to sacrifice performance to avoid a single kernel crash every three years.
There are so many more useful implications to this research than just 'we only need one kernel that doesn't crash'. Simplicity is a cornerstone of trust and trustlessness, which our world desperately needs more understanding of
I get a bad impression from this waste of euros. The whole O/S and all mentioned software is truly archaic -- decades behind. The trade-off of performance for some nebulous reliability claims is unacceptable; I don't remember my Linux servers EVER having crashed. Reminds me of a programmer I used to work with whose solution was to use a `for' loop to retry any code that had an error.
freebsd has very reliable kernel. please don't make me laugh. their monolithic piece of s**** would kernel panic when unplugging a usb stick.
But it would reliably panic.
@@mawangdwei3866 rofl!
dos have iceWM or chrome?
web browser? usb 3?