As I was reading /. before going to bed (it happens on a regular basis), I stumbled over this Tanenbaum-Torvalds Microkernel Debate Continue, with a link to Tanenbaums latest installment of the debat (Tanenbaum-Torvalds Debate: Part II) which have been going on for a very long time. To make it short it's about micro (Tanenbaum) vs Monolitic kernels (the thing whic makes your computer tick). The interesting thing is that even though I'm a Linux user (well kind of, as I use Apple OSX on my desktop for stability and usebility), but linux is fine (for me), as long as it stays on a server.
Now Tanenbaum have a point whic Linus might have missed - it's not about functionality it's about stability, would you like to reset you TV everything you switch channel (I have a topbox which more or less requires me to do so - which is why I don't use it much). Linux is very stable compared to some of the things available out there, but it bloated (Try to install any disto with X, and all the wistles and see how far you get on 256mb ram). Not only that but because all the drivers run in kernel space (ring 0) any driver can kill the system if it is written badly (what many are).
Not saying that Minix3 is the holy gral, but from my standpoint the computers we have these days should be fast enough to run systems which are written in such away that no single subsystem can bring them down (ok, ok there are things like disksystems which are vital), but what fun is it to have a system which might stop working simply because of a badly written driver. Imagine that you install a new video driver on your WindowXP or Linux PC, and follow the instructions and reboot, and suddenly you either get a blue screen (WindowsXP, actually it's a trademark), or a text screen which tells you that something is wrong in a way which is non-human understanable. That is no fun... Why is so hard for the people who write operating systems to get it - we (I hope that I'm not alone in this) actually would like to have something which works most of the time.
Some people asked me why I switched to Apple OSX - well it's easy, I got so tired of spending more time getting things to work when I just wanted to read my email, that I decided that the money would be well spend, and they where.
Now the interesting part, is that when I did my associate degree in software design/development (1991) we used a book by Tanenbaum (Structured Computer Organization, ISBN: 0-13-148521-0) which I actually never got around to read the whole of, but the parts I understood is still something I use today. He must have gotten something right, as it still a very usefull book (was updated in 2005).
I've spend an awfull amount of time on Linux systems the last 10 years, and even though I know a lot about them, they still frustrates me. They can be very stable, and very very reliable, but they can also be the oposite, and in most cases noone know why the latter happens. I just think that sometimes Linux is not the best solution the problem....