Tanenbaum's Book (out of Book Recomm.)

[Crazed123]

Is it true that Tanenbaum's book ("Design and Implementation") doesn't include things like paging? Or security?

[Colonel Kernel]

Is it true that Tanenbaum's book ("Design and Implementation") doesn't include things like paging? Or security?

The book discusses paging, but Minix doesn't support paging so there's no example. From what I remember, security isn't discussed much at all.

[Candy]

Is it true that Tanenbaum's book ("Design and Implementation") doesn't include things like paging? Or security?

The problem when quoting books by author is that when an author publishes two books you get a problem.

As I've been trying to tell the scientific community for a while now (no success whatsoever), use the damn name of the book and not the author. In terms of OS development, Tanenbaum is half your curriculum as far as I can tell, and the rest mostly releases books based on his books.

[Solar]

I have to second this. Refering to "Tanenbaum's Book" when talking about OS design is like refering to "that Greek dude's teachings" when talking about mathematics (or philosophy)...

[Crazed123]

Note the title (subtitle, actually) of the specific book in parenthesis right next to what you're referring to, telling the reader exactly which of the author's books I'm talking about.

[Candy]

Note the title (subtitle, actually) of the specific book in parenthesis right next to what you're referring to, telling the reader exactly which of the author's books I'm talking about.

It still looks like something you added later on so because you remembered (or something) that he wrote more than just that book. I'm still going to try to convince people that you should just call it Operating Systems - Design and Implementation by Tanenbaum rather than the other way around.

[distantvoices]

or this way: A. Tanenbaum - OS Design & Implementation.

Is also nice and one knows what the heck I talk about.

[Freanan]

I lent Modern Operating Systems by Tannenbaum and it seems pretty good.

But i am not sure wether i should get design and Implementation...
After all it is, as it seems, intended to be some kind of practical tutorial on writing your os.
But it won't be THAT practical if your memory managment is totally different...
Does he implement adress spaces with use of segmentation, then?

[Colonel Kernel]

The problem when quoting books by author is that when an author publishes two books you get a problem.

I didn't have a problem knowing which book Crazed123 was referring to...

I lent Modern Operating Systems by Tannenbaum and it seems pretty good.

I haven't read that one. Does it include Minix source as well, or is it all theory...?

Does he implement adress spaces with use of segmentation, then?

AFAIK, no. Minix has no memory protection whatsoever. I could be wrong though... I just skimmed the MM chapter as soon as I realized that Minix didn't support paging.

[Freanan]

Is it all theory (except some abstract pseudo-c-code to show how it works - but that's pretty throretical too ).

[Xardfir]

"Implementation and Design" is more than just theory. After all Minix does work compared to some of the stuff (including my own) I have seen. It's multi user, multi tasking, supports networking and dare I say has a small amount of applications written for it.
What Andrew Tananebaum pushes through the book is -> here are the basics, if you don't like it change it, improve it, add graphics, more devices.
Whilst I am not pursuing such a project, why not have a go at adding paging and more to Minix?
Linus Torvalds did - the result is Linux.
This book is worth it's weight in gold regardless of the topics it doesn't discuss.

[Colonel Kernel]

"Implementation and Design" is more than just theory. After all Minix does work compared to some of the stuff (including my own) I have seen.

I was asking about "Modern Operating Systems", but I agree about the other book.

Whilst I am not pursuing such a project, why not have a go at adding paging and more to Minix?
Linus Torvalds did - the result is Linux.

And that's why it's a primordial soup of really ugly code. No thanks, I'd rather start fresh.

This book is worth it's weight in gold regardless of the topics it doesn't discuss.

Agreed, even though Minix is a great example of how not to design a microkernel. Sometimes good counter-examples can teach a lot too.

[Freanan]

What Andrew Tananebaum pushes through the book is -> here are the basics, if you don't like it change it, improve it, add graphics, more devices.
Whilst I am not pursuing such a project, why not have a go at adding paging and more to Minix?
Linus Torvalds did - the result is Linux.

That's a good idea. Probably the result would be "better" than linux, as linus sadly changed the kernel design to monolithic...

[Candy]

What Andrew Tananebaum pushes through the book is -> here are the basics, if you don't like it change it, improve it, add graphics, more devices.
Whilst I am not pursuing such a project, why not have a go at adding paging and more to Minix?
Linus Torvalds did - the result is Linux.

No he didn't. He did not use any single byte of Minix code in creating Linux.

That's a good idea. Probably the result would be "better" than linux, as linus sadly changed the kernel design to monolithic...

Why would monolithic always be worse than microkernel?

The use of monolithic in OS kernels does not always imply that it should be a static chunk you can't change without recompile, which is what most people still assume monolithic is.

The division between monolithic and microkernel is that a monolithic kernel runs drivers in kernel space, in the reasoning that you can't prevent the system from crashing if the driver is handling irresponsible, so why bother with the extra layer if it's still going to crash. Microkernels are based on the assumption that you can recover from a device or driver crash.

Good design isn't based on either of these in full. There's a point to be said for the first (if your harddisk driver crashes, you can't reload it from disk), as well as for the second (when your webcam driver crashes and goes on a memory-eating rampage, you can terminate it without any problem).

Why not look beyond the stupidity of cubicle-thinking (straight translation, comes from dividing people into groups such as colored people, white people, chinese people etc.) and design something that's better than each of the possible cubicles into which you could have stuffed it?

Say, actual design which involves looking at all simple options and trying to get a design which has the positive points you like of the designs you're evaluating with the least amount of counterpoints you don't like.

[distantvoices]

re cubicle thinking: Hear Hear!

I daresay that's a good statement.