Somehow used in more than half of all computers produced today and still falling apart at the seams. Funny, I wonder why. Actually, none of us do
Other urls found in this thread:
B-but how can this be?! After all microkernel is superior technology, right?! Right guys?
Since intel integrated it in the Management engine, literally almost all PCs in the world runs it.
Same for MINIX 3. Classic monolithic *nix, as mediocre as it is, is "just good enough".
"Good enough" for what exactly? DragonflyBSD has already caught up to it in terms of scalability, despite being a much newer and smaller project. I guess Linux is "good enough" if you don't mind blowing 1000 times as much developer man years.
Maybe it made sense to use monolithic kernel in 1991 when the best hardware the average dude could buy was an 68030 or 80386, but that was almost three decades ago. And on top of that there's not a whole lot more security problems that would benefit from being isolated, which a microkernel design is better suited for.
How retarded do you have to be to think that? MINIX is only used since Skylake, so that alone severely limits its numbers.
It's good as a joke, but hardly justifiable as a statement of fact on a Vietnamese wood-carving forum.
Plan 9 failed because BSD and System V were better than Plan 9. The kernel functionality was better. The desktop environments were better. The shells were better. They sucked (and still do) compared to a real mainframe or workstation OS, but Plan 9 was even worse.
I'm interested as to what you think is a "real workstation OS".
I was wondering about this too
Maybe "LITERALLY ALL COMPUTERS run Minix!" is an overstatement, but it is present in enough computers to deserve attention.
He said produced not computers in use.
I'd rather have understandable text that is awkward to read instead of easily readable text that makes no sense.
This is so true. DF also has the added advantage that it's close enough to FreeBSD that porting is relatively easier.
I seem to remember reading something about how including MINIX iniside the Intel ME violates the terms of the BSD license, making a legal battle a reality. Intel derserves to die.
You remember wrong, the whole point of BSD-style licenses is to allow for it to be used in proprietary software.
As Linus said, BSD license is for the code, you don't care, even if it's forked and became better in some way. Obviously better license for smaller programs than the copyleft crap.
Thanks to BSD we have an all powerful OS controlled by Intel running at ring -3. But hey, at least it is "improved", right? Obviously number of installs is what software should be the goal number one. Copyleft would have given us the freedom to rid ourselves of it. Permissively licensed code is worse than proprietary software and only corporate shills and the retarded ancaps support it.
Linus would do everything for his corporate masters btw. Hand that feeds you and all that.
Intel could use QNX or keep the pre skylake system. Also think about the countless examples of GPL infringements that makes Linux as bad as BSD and lawsuits push firms to use even less copyleft software, like Google did with Busybox or spearheading clang support for Linux compiling just for decrease the size of the toolchain.
Minix doesn't even have the advantage of having a noticably different user and development environment (like Plan 9.) It's just another Unix in a world that already has too many Unixes.
The BSD license has only one term: do whatever the fuck you want with it, as long as I get credit.
So theoretically all computers with ME should have a copyright Andrew Tanenbaum notice somewhere.
It doesn't. BSD infringement is also a nice achievement
Microkernels are really awesome in theory. The practice is shit. If there was an interesting microkernel os that was GPL v3 I'd install it on partition and play around with it, but I'd still consider it just a glorified toy kernel.
Do you think that GNU Mach is not good enough for you?
No. It's not interesting.
I'm not sure what you mean by an interesting microkernel. Microkernels are minimalist by definition. This means that microkernels are interesting by the virtue that they don't do more than what is necessary for the job they do.
Which is exactly what any software should do everywhere.
If it has Mach in it it's shit. Even the people who designed it hate it.
what do you think
nobody gives a shit about how "REAL" it is
if it runs like ass and has horrible software to back it up
Microkernels are going to get fucked hard by Meltdown. Switching address spaces on syscalls is bad enough, now imagine that happening orders of magnitude more often as multiple services communicate to do the work of one syscall.
You have this software, it's called Plan 9 OS. It is not a popular way to use your computer.
That's actually a good thing. You're supposed to implement USB drivers as a userspace service.
Can't wait to use Minix3 as a desktop workstation with opensource stable software...
MINIX is supposed to be a "teaching OS" that teaches you how to write an OS. Not a replacement for UNIX. Also, Intel forked MINIX over 10 years ago, so it's fork is probably significantly different.
We will probably know soon, once somebody gets the source, or extracts it from the motherboards that half of the world uses and decompile it
sta.li is just a GNU/Linux distro without the GNU. Instead of GNU it's pure suckless autism. I like suckless but couldn't live without X.
MINIX and MINIX 2 are "teach OSes" (note: they are OSes in which the development is explained in a book from which you can learn, they are still usable and pretty good operating systems). MINIX 3 is a whole new deal. Entirely new kernel, userspace, and development process. Calling it MINIX is only a namesake if anything. MINIX 3 is a solid competitor to UNIX aside from the lack of hardware support (USB being a PITA).
Please stop spreading bullshit. MINIX 3 is not MINIX 2.
No. rc is fucking brilliant, and to ignore that is to do a disservice to yourself and the developers. The problem is, UNIX was just good enough. BSD is irrelevant to the Plan 9 debacle.
stali stands for statically linked, what makes you think it cannot have x and graphical display?
it is also funny that linux users are not able to distribute binary executable files (and sources) that simply run or even can be compiled on any system, because there are cancerous dynamic dependencies with unlimited faggotry, stali addresses this problem. It is also funny and ironic how statically linking everything makes a system that is much smaller and faster.
Sorry, I only remember reading on Tannenbaum's blog that that's what he wanted Minix to be, I don't have a link to it. Most people (myself included) didn't know that there was such a conceptual difference between MINIX 2 and MINIX 3
Tannenbaum mentioned his concern towards Intel chosing an older teaching version of minix instead of their production ready version.
"I certainly hope Intel did thorough security hardening and testing before deploying the chip, since apparently an older version of MINIX was used. Older versions were primarily for education and newer ones were for high availability. Military-grade security was never a goal."
What a cunt. Intel fucking up is good for us though.
wtf are you on about?
I'm still trying to understand why the fuck can't I make static binary out of glibc. Why did you freetards have to let this happen?
because you obviously never tried having statically linked X, let alone any graphical userspace using gtk/qt
yea that shit needs deleting some junk
Anta baka? He said that it was wrong for them to do this and doesn't agree with it, but he foolishly thinks that it being one of the most used OS's on the planet is a good thing. Reminder to trust in Stallman. Run Libreboot or TALOS and only GNU/Linux flavors that respect your freedom.
Intel wouldn't. The tech industry is trying as hard as possible to kill Blackberry.
BSD license basically says : "Please take my code and use it for your proprietary garbage without giving me any compensation or recognition."
But at least, this way Theo the rat can stick it to the (Stall)man
They never conformed to normies.
For a text-centric workflow Plan9 is great, it's also quite good if you're a programmer. Vi and emacs tards hate acme though.
It's worth getting used to just in case, and the whole environment has far fewer lines of code than systemd or the Linux kernel.
Linus raped Andrew so hard
The BSD license appeals, the way I see it, to "anarchists" who see any kind of rules as constrictive, but aren't quite smart enough to realize that their philosophy, a priori free-er than GPL, in reality turns out to be just another iteration of neo-con/ultra-capitalist ideology.
The license is literal shit, at least Stallman had a vision of software and beyond society that went beyond kindergarten "would like rules or rules? I want no rules! yeah, i'm an anarcho/crypto/poly meme."
We're not using MINX today because Tannenbaum spent decades being a retard. The old versions of Minix were all payed proprietary software, and by the time Tannenbaum realized that nobody wants to pay for some hobbyist's kernel, GNU/Linux had already cornered the market in free (both gratis and libre) operating systems. If he'd just made it free from the get-go we'd all be using GNU/Minix right now.
Fixed it for you. Poettering would have crept along anyways, so, in the end, what's the big difference.
I don't know that a microkernel could possess all those linux-specific functionalities that systemdicks depends on.
Allegedly he wanted to make it free, but it wasn't possible for some stupid bureaucratic reasons. So he just sold it for the minimum.
Too bad he couldn't, because the design was simpler/better than Linux, and ran fine on much older/cheaper 286 computers with less memory. So at least it accomplished its instructional purposes well.
Depends, Muen kernel claims they are safe from it.
We thoroughly studied the potential impact of the recent
Spectre/Meltdown speculative execution CPU design issues on the Muen
Separation Kernel. In this mail we would like to share our findings
regarding Meltdown. The analysis of Spectre will follow in a separate
For the technical details of the Meltdown vulnerability the reader is
directed to the associated papers and blog posts .
Meltdown is part of a new attack class which relies on observing side
effects caused by speculative instruction execution by the processor. It
is also referred to as Rogue Data Cache Load (CVE-2017-5754).
For a successful Meltdown attack, three requirements have to be met:
(1) The memory space of the unprivileged attacker contains privileged
memory mappings to which the attacker has no access (U/S bit not
(2) The mappings contain desired information
(3) The attacker can measure the timing effects introduced by its attack
Muen uses VT-x and not ring-0/ring-3 transitions as isolation mechanism
between subjects and the kernel. As VT-x transitions automatically
switch the memory layout between guests and the host, Muen does not use
the User/Supervisor bit in page tables for the enforcement of access
Consequently the precondition (1) for the attack is not met and the Muen
kernel is not vulnerable.
Subjects which internally rely on ring-0/ring-3 transition (e.g. Linux,
Windows) are vulnerable from local attacks unless adequate mitigation is
performed at subject level. E.g. for Linux guests, Kernel Page Table
Isolation (KPTI, formerly KAISER) must be enabled.
Meltdown is defended by our design decision to have a simple
architecture which only utilizes a single isolation mechanism: hardware
virtualization. We prioritized a minimal design over performance
considerations and decided not to use ring-3 in VMX-root mode for native
subjects. Since Meltdown only affects the ring-0/ring-3 isolation
mechanism we were spared from that pit of lava.
The Muen Team
 - meltdownattack.com
No, if you go back and read the Torvalds/Tannenbaum debate, you'll see that Andrew actually defended charging for Minix.
He was bound by a publishing contract.
Ironically, the modest $69 fee for floppy disks and shipping would probably have been considered reasonable by RMS. Since all the source code was available and provided thus, he would or should have vetted this OS as libre software, which the user has full access to source codes for the purposes of learning and modifications.
Where is the firefox port for it?
There's no VisiCalc port either.