friendly reminder that minix 3 uses netbsd for user land and almost everything is ported including clang, x11, numerous window managers and so on. tanenbaum calls it netbsd with a micro kernel.
it has a different mission now, being a highly reliable operating system.
why don't you give it a try today?
>inb4 muh intel ME
tanenbaum was not informed about that, minix has a permissive license so anyone can use in commercial projects.
Other urls found in this thread:
phoronix.com
minix3.org
nerves-project.org
blog.darknedgy.net
beagleboard.org
raspberrypi.org
youtube.com
genode.org
epics-controls.org
rtems.org
beagleboard.org
phidgets.com
youtube.com
twitter.com
Why though?
I'm actually surprised we haven't seen more adoption in the sever market.
All the Linux drama and "features" noone wants that shit up kernels leave minix pretty attractive.
Except most of those window managers don't work. Nor does any browser more complex than w3m.
One reason it might not be getting adopted is because it lacks a 64-bit x86 version. It's pretty much only good for embedded stuff.
Look to Genode, which has stated its aim to become more user-friendly in 2019, and is already a rock-solid system built on the seL4 microkernel.
phoronix.com
How about some links?
minix3.org
firstly, microkernels are slower than monolithic kernels like linux.
secondly, minix is only 32 bit and the main target is embedded systems. so it can only use 4gb of ram which is not enough even for laptops.
www.minix3.org
yeah, it was ok yesterday, probably will come back soon.
HelenOS is also worth a look.
And they're in FOSDEM again. I'm looking forward to the talks there.
lol an html page is a few KB, your google account may have been hacked, glowing nigger
>microkernels are slower than monolithic kernels
maybe, but certainly not slow enough to turn Google off from using one for their new OS
>It's pretty much only good for embedded stuff.
Since the Minix philosophy is similar to Erlang's, maybe the thing to do is put them together.
"The Nerves platform is a customized, minimal Buildroot-derived Linux that boots directly to the BEAM VM."
nerves-project.org
Instead of a Linux kernel, how about the Minix micro-kernel and a collection of drivers?
>I'm actually surprised we haven't seen more adoption in the sever market.
Probably because it's so fucking slow, not sure if it can handle large amounts of connections well like freebsd or linux either.
>firstly, microkernels are slower than monolithic kernels like linux.
blog.darknedgy.net
What is a good processor/board for a fault tolerant, never stop Minix-Erlang embedded system?
Something supported. That means old hw now (x86, like some random athlon-xp).
Sigh, they do need to release 3.4 already, like, two years ago. But they do have serious manpower issues unfortunately.
The HURD is a deadend, and is more active, which is annoying. I don't know if it's bad or a blessing... it serves as containment project for certain people.
Does Minix 3 support multicore architectures?
Beaglebone or whatever the arm board they support is.
It has SMP, but as AIUI it's broken partially or completely in the current code.
Minix3 does really need a full time developer. Patreon anyone?
I honestly don't know why people haven't gone all in on pkg-src.
>32 bit arch means 32bit address space
lmao
Waiting for this and/or Fuschia seems most likely to succeed for obvious reasons but I'm most interested in seeing people use seL4.
I'm basically in the dark about Minix.
pkgsrc is great. i experimented with it on slackware and it just werked.
The Minix 3 website is down so it's difficult to find the relevant info. This does not inspire confidence in the technology.
Certainly, irrespective of the quality of their design and implementation, they do suck at promoting it.
Support for 64-bit ARM SMP devices (e.g., raspberrypi.org
>see their demo
>think it's awesome
>2010
youtube.com
God damn. I can only imagine how much progress they've made.
They have far more recent demos. Particularly, with dynamic scenarios, as they're trying to make it into a general purpose OS and dogfood it internally.
See: genode.org
Rather than going crazy with hardware support, I'd rather have them focus what little manpower they have in polishing what they have, fixing bugs, making more netbsd software build and run and so on.
The beaglebone is a much more open, robust and well documented platform than the raspberry pi, which is why they picked it.
Rather than EPICS on RTEMS, a distributed control system built in Erlang/Mnesia on Minix might have some attractive characteristics.
eChronos, seL4.
Interesting reference. Thanks.
Can't get the odf or the image to work in vmware. Dang.
You are already running everything in Java. Any pretense of good performance already left the building long ago
Fixed it by manually extracting the ovf and attaching the disks by hand.
>nerves-project.org
>a distributed control system built in Erlang/Mnesia on Minix
From a development perspective, this seems like it might be a good idea for building-automation/smart-home applications. The controls software could be continuously developed, updated and debugged on installed, running hardware. If that hardware were to abide by similar principles as Erlang and Minix - e.g., inexpensive commercial grade hardware that can fail and be replaced while the system is running - that might be very cool for DIY projects.
beagleboard.org
maybe?
This guy's accent is a bit much but...
youtube.com