Page 1 of 1

Exokernel architecture

Posted: Fri Jun 05, 2009 7:51 am
by quanganht
I found just a little number of documents on this kernel architecture out there. Any suggestions?
Note: I've visited MIT's XOK and Nemesis, and Googled everywhere.

EDIT: And what do you think about constructing a new architechture ?

Re: Exokernel architecture

Posted: Fri Jun 05, 2009 8:18 am
by Solar
quanganht wrote:And what do you think about constructing a new architechture ?
You asked. 8)

Personally, I think most if not all of the things required to build excellent operating systems, be they desktop, server, or embedded systems, are already available and well-documented.

Including excellent real-world examples on how not to do it, like Linux or Windows.

What the world is waiting for isn't an all-new architecture with its very own set of fledgeling problems, but a system that does away with all the problems "modern" OS'es have from decades of learning-by-doing and legacy compatibility. A system build solid from the start.

Being able to roll back a software installation without artifacts remaining. A solid solution for DLL and dependency hell. A system that doesn't give one malicious piece of software full system access if it only flashes the right credential. A responsive and consistent GUI without surprises or style breaks. And so on.

As I said, you asked. That's what I think about constructing a whole new kernel architecture. ;)

Re: Exokernel architecture

Posted: Fri Jun 05, 2009 8:46 am
by quanganht
Well, it's easy to say but hard to do anything real. Even those called "real-world examples" is just applied for one or some situations only, not ideal for a really NEW thing

Re: Exokernel architecture

Posted: Fri Jun 05, 2009 5:27 pm
by mathematician
quanganht wrote:Well, it's easy to say but hard to do anything real. Even those called "real-world examples" is just applied for one or some situations only, not ideal for a really NEW thing
What do you mean by a "really" new thing, and what advantages would a new architecture have? Whilst coming up with a really new thing isn't impossible, it isn't eay either; otherwise somebody would already have done it. You would need to find a problem which was crying out for solution. Where can you find a significant number of people saying, "If only......"?

Re: Exokernel architecture

Posted: Sun Jun 07, 2009 8:46 pm
by steveklabnik
Exokernels are awesome, we're doing one. Right now, actually. Didn't you make a post to our list a long, long time ago or something? I feel like we may have talked about this before.

What kind of things are you looking to know? I thought those original papers were pretty good.

Re: Exokernel architecture

Posted: Sun Jun 07, 2009 10:06 pm
by Wilkie
Part of our defunct documentation, but valid for your interests:

http://code.google.com/p/xomb/wiki/ExokernelInfo

Although, anybody else with good info that perhaps I haven't read, it would still be nice.