In reply to Rick Moen's flatulent wordings,
> Quoting Smelly Pooh (plop at redbrick.dcu.ie):
> > Microkernels were always (in theory) going to be weaker on performance,
> > but since microkernels like QNX could deliver good performance I imagine
> > the problem lies more with Darwin's implementation (or that of Mach)
>> Implementation tends to hold lots of nasty surprises for people who
> chose to believe the predictions of academic computing theory. My point
> was that performance and stability on the level of QNX's has turned out
> to be possible _but difficult_. In the end, it doesn't matter if your
> OS could theoretically be superb _if and when_ you were able to work out
> the extremely thorny threading, semaphore, scheduling (etc.) problems.
> What matters is how soon you can deliver a maintainable, stable codebase
> with perforance in the target region.
For the sake of hypothetical discussion lets say that QNX's microkernel
were to be made available to OS implementers (be it in the form of
licensing or open source). Now you have a guaranteed stable and
efficient microkernel to start from, and you can write priveleged
servers on top of that to provide OS level services. It's what so many
OS developers are doing on top of the Mach microkernel, but everything
I've seen has indicated that Mach is just a poor implementation to begin
with, affecting everything that's been built on top of it.
I guess I'm backing microkernels as an OS concept for 2 reasons
1) software reuse (building on top of an existing microkernel saves the
tedious, bug prone work of writing the low level code of an OS)
2) separable, pluginable OS services
> Have you read this? http://www.jwz.org/doc/worse-is-better.html> Interesting thoughts to contemplate, if you haven't already.
Yeah, I read it about a year ago when it was linked to by
www.joelonsoftware.com, another good source on software design. I'm
very clearly a follower of "the right thing" approach, I even
wrote a scheme interpretter for my 4th year project.
Back to the original topic though, which approach should the college
course on OSs take?
> >> For starters, get written so that they work and perform well.
> > Well that's the advantage/disadvantage of microkernels isn't it?
>> You sound (I figure) like you're thinking theory. I'm thinking
> implementation following theory. But I tend to do that; I'm just a
> lowly sysadmin, not a systems developer.
I'm thinking one good implementation that everything can build on top
of... but no good implementations are available to other OS implementors
that I know of
Maintained by the ILUG website team. The aim of Linux.ie is to
support and help commercial and private users of Linux in Ireland. You can
display ILUG news in your own webpages, read backend
information to find out how. Networking services kindly provided by HEAnet, server kindly donated by
Dell. Linux is a trademark of Linus Torvalds,
used with permission. No penguins were harmed in the production or maintenance
of this highly praised website. Looking for the
Indian Linux Users' Group? Try here. If you've read all this and aren't a lawyer: you should be!