Minix from Scratch Project Established

decuser writes "The MFS - Minix from Scratch project was established in the wake of the Brown-Tannenbaum controversy. MFS aims to be to the Minix community what LFS is to the Linux community, a recipe for building an alternative OS from 'scratch.'" See the project's website at mfs.sunsite.dk or minixfromscratch.org.

Comments

huh?

[crayz]

Wasn't Minix already built as much "from scratch" as Linux was?

Isn't Minix intentionally incomplete?

[Anonymous Coward]

It is my understanding that Tannenbaum never wanted Minix to be feature complete because it is a teaching OS, not a production OS. He often rejected submitted patches on these very grounds. The idea was to use Minix in a teaching environment. By keeping Minix simple and incomplete, instructors could tailor lab assignments to provide missing features and extensions.

Minix is great for teacher and student, but it shouldn't be the first choice for general purpose operating systems. At one time it didn't even have networking (but maybe that has changed).

Re:Isn't Minix intentionally incomplete?

[The_Dougster]

What makes me wonder about all this good PR is that the Hurd existed as a project before Linux, and it's still alpha code. Why? And why is the Hurd only available for 32-bit x86? Is the hype surrounding microkernel false, or was there some other factor that has slowed down the Hurd despite its microkernel superiority? (And if so, what is that other factor -- human factors among the the Hurd development leads perhaps?)

Personally, having been messing around with gnumach for quite a while, I think that its a tremendously overcomplicated microkernel. Adding device drivers is a nightmare, and programming for it seems to be really convoluted and alien. It was and probaby is really advanced in some respects, but the complexity caused by this has really hindered the Hurd developer's efforts.

PC hardware advanced right beyond gnumach's ability to keep up almost overnight and anybody who is smart or knowledgable enough to make it current is probably 1. either working 70 hour weeks and getting rich or else 2. too busy living the good life from doing 1. above and too burned out to stand the sight of a monitor anymore.

I wish I could magically patch in linux kernel 2.6 drivers into L4 or gnumach but I'm just a hobbyist and I'm just happy to be able to compile and get the stuff to run. We non-CS major's are at somewhat of a disadvantage I think when it comes to kernel hacking.

Re:Isn't Minix intentionally incomplete?

[steveha]

As I understand it, the microkernel advantage is supposed to be that almost your whole system is running as user-level processes. There has to be an inner kernel running as kernel-level, but it is stripped down and basically just schedules tasks. Everything like disk I/O, network, etc. is handled in user space.

So, if you need to debug your network stack, you don't need some special kernel debugger, you can just use gdb. And if you make a mistake in the network stack, it should have limited ability to lock up your whole system, unlike a monolithic kernel (such as Linux) where any error in kernel code potentially could take down the whole kernel.

Yes, the monolithic kernel is simpler: you don't need to build a message to the network subsystem, send it via the message queue, and so on; instead you just make a system call, and you are done. And in fact I believe this simplicity is what made Linus choose the monolithic kernel, despite all the good PR attached to microkernels.

Note that Windows NT started out as a microkernel-ish design, but the overhead of all the message passing led the NT guys to integrate several large chunks of the system (such as video) into a monolithic kernel. Thus a bad video card driver can take down a Windows server, and you cannot run a Windows server completely headless as you can a Linux server.

Linux is now good enough that banks use it to keep money, while the Hurd is still alpha code. Which is why, despite not knowing much about all this, I suspect Linus chose correctly.

steveha

missing the point

[fred fleenblat]

The tempest in a teacup is over whether an OS could be written from scratch by a single person. Making a shared project out of it fails to prove that a single person can do it. Worse yet, Ken Brown's sense of logic will lead him to cry that it's proof that a single person cannot do it.

It's probably a fun project, but it isn't really going to prove anything new that reasonable people don't already know. And it will fail to convince unreasonable people of anything.

Agree. Better places to put in effort

[EmbeddedJanitor]

How about contributing to GNU/Hurd instead? At least Hurd intends to oneday be a real OS and has a microkernel architecture.

I guess though it is also worth noting that while Minix was only designed for teaching OS concepts it has been used for RealWork. The same happened to Pascal. Nobody was ever supposed to write any RealCode in Pascal - it was also intended only as a teaching tool.

What controversy?

[Ohreally_factor]

Ken Brown was full of shit and was quickly debunked by most of those he interviewed and the guy he hired to compare the code. The only controversy exists in his delusional mind and in the minds of other nutjobs. Even MS has distanced themselves from this FUD project, as Ken Brown is so clearly a fool.

Re:Gentoo...

[BrainInAJar]

no it isn't, because the idea of LFS isn't a couple miliseconds faster response time, it's to learn how the OS works

"in the wake of the Brown-Tannenbaum controversy"

[FatTux]

Controversy? I cant see one, all I can see is that Brown said a lot of BS and was rebutted by Tanenbaum. My 2 cents.

educational tool

[noldrin]

It seems like a lot of people are missing the point here. Minix is a great way to learn about how operating systems work. A Minix From Scratch project will make it even more useful for this. This will hopefully be a great education tool for people wanting to get into how oeprating systems work. I really don't see it meaning to be a big competitor with GNU/Linux or *BSD

It is EMPTY controversy, so the project'll fail

[mikelang]

I wonder if Brown can be sued if he lies outright and has no way to prove his claims? The book he writes is not a fiction...

PS My favourite part of Brown's response is:
...United States Patent and Trademark Office, an internationally respected agency...

Minix Instead of Hurd

[RAMMS EIN]

I think choosing Minix over Hurd has everything to do with Minix being a working kernel that has been ported to a number of architectures, whereas the Hurd is struggling to work on even one architecture.