> > areas... but please, don't say "NetBSD has better approach to hardware,
> > than Linux", because it's _not_ true.
>
>
> 	I cannot agree with you. Have you yet run NetBSD and Linux with a
> bad HD? In most cases Linux will just lock, panic or reboot (*this* is

yes i did. and got exactly this.

> true, I've seen it many times). NetBSD behaves the way it has to, keeping
> the system up when possible, even when something goes wrong with the disk.
> I have a co-located server (an old p133) that runs NetBSD and has one (one
> of three) deffective HD. It trows some errors but the machine is running
> fine. It's a 400km trip, that's why it's not fixed yet.
>
> 	Regarding hardware, Linux may support more IDE controllers than
> NetBSD, but you just cannot say it'll work better than netbsd's. Also, you
> cannot compare raw transfer rates with transfer rates obtained by a
> benckmark that uses also the filesystem when measuring.

the amount of supported hardware depends of how popular system is and how
manufacturers cooperate.

there are lot of binary-only linux drivers, lot of "normal" but i can't
see any correlation between this and quality.

>
> 	And for old crap hardware, my machine is a Dual Pentium III with a
> VIA686B controller (I know it's still crap, but now that old), and Linux
> will not set drives to run at DMA mode by default (even windows does).
> Kernel 2.4.19, and yes, I know how to compile Linux kernels.

for kernels 2.2 there were Hedrix IDE patches that did it partially
automatically, but they were not done for 2.4