[clug] Kernels are easy, ls is the hard part!
paul.wayper at anu.edu.au
Tue Sep 18 01:41:45 GMT 2007
steve jenkin wrote:
> As for 'microkernels', they've been promising big things since 1985.
> Which is why Microsoft chose to license Mach as the base for Windows
> NT... Not just NeXT.
> Perhaps they'll morph into VM hypervisors.
Interestingly, I had one guy come up to me at SFD, point at a Ubuntu CD,
and say "I don't know why you're giving that stuff away, that's as slow
as. You know what beats it hands down?"
Fearing a Microsoft Zealot, I said, "What?"
We got talking and he admitted that he had a couple of servers running
Ubuntu and Red Hat at home and work, and he seemed to hate Microsoft
even more than I do, and overall he was quite friendly. But it's always
amusing to see people bag microkernels because we're used to monolithic
kernels. Read up on the QNX article on Wikipedia - they seem to solve
the message passing problem that Linux kernel advocates cite as the
problem with Minix. (It was amusing to listen to some of the people
coming out of Andy Tanenbaum's keynote at LCA 2007, muttering "geez, as
if you'd really kill the disk driver ten times a second - that can't be
good for performance". None so blind and all that.)
Since QNX just open sourced their Neutrino kernel
(http://www.qnx.com/news/pr_2471_1.html), I think this might be a good
time to broaden our horizons.
More information about the linux