Re: NetBSD Next Logical Step: Microkernel...?

[Martin <martin.kelly4000%gmail.com@localhost>]

 Magnus Eriksson <magetoo%fastmail.fm@localhost> wrote:
> I generally like the microkernel idea too.  NetBSD is based on a

> different model though, and I suppose people interested in contributing
> to the project are the ones that think that what is already in place is,
> basically, what a decent operating system should be using; and are
> unlikely to want major revolutionary changes.

> There's always Minix, they've been importing a lot of NetBSD code lately
> it seems.  I've been meaning to try it out more seriously..

Yeah your probably partially right, but i am pretty confident there other reasons too (the ones that my curiosity are bugging me about lol), i mean while NetBSD is open source and anyone can contribute i am also pretty sure there is an executive board of members that overlook the direction, and ultimate goal amongst other things in regard to NetBSD similar to how FreeBSD is run.
Probably the same board that decides what becomes part of the mainstream kernel/ OS and what is just a critical app that doesn't have to be pre-installed but can be added as a package. RUMP anykernel being a decent example of something that is included and not installed via pkgsrc.

Just take for example the abstract of the PDF link Valeriy provided where the author briefly talks about the security and performance differences between microkernels and monolithics. That explains a partial reason but its not as though either of those points are a killer. It just provides a reason to push that boundary which is quite clearly what the RUMP anykernel does, which is awesome.

Minix is cool especially because it so easy to understand how it works, a shame it doesn't have the software support. Although another awesome one for totally different reasons is Plan 9.

Valeriy E. Ushakov <uwe%stderr.spb.ru@localhost> wrote:



> Are you aware of pooka's work on kernel modularity?

Yes i have heard of the Rump any kernel. It is one of the main reasons i raised the topic, thanks for the PDF link it looks really interesting but very long too! (362 pages isn't going to be a short read) :)