▲ 52 points
back
5 comments
FYI: The main project URL listed on the site, http://www.cs.utah.edu/projects/flux/oskit/, results in a "page not found" error.
The correct URL apparently is:
1997? The OSKit is old enough to go out and drink beer now, yikes.
They stopped development on it shortly after this, didn't they? Like within 4 years? It's easy to see why, given changes in the codebases that they pulled from (Linux, NetBSD, FreeBSD) to support newer hardware.
It's been over 15 years but the impression I have is that the complexity and need for "glue" makes OSKit impractical for learning,and the dated nature of the drivers makes it impractical for cherry picking.
I wonder if we have anyone from the osdev community/forums here interested in commenting. It seems like the kind of thing they'd be very disapproving of[1].
Reminds me of genode
Wow, I have fond memories from when we tried porting GNU/Hurd to this. Good times! https://www.gnu.org/software/hurd/microkernel/mach/history.h...
> We built our system, called ML/OS, by porting the Standard ML of New Jersey (SML/NJ) implementation [6] to run on a PC using the OSKit. SML/NJ is a complex, Unix-based system comprising about 144,000 lines of code, in over 1000 source files. The run-time model used by SML/NJ is fairly exotic - for example, the system runs completely without a stack, using instead very aggressive heap-allocation and garbage-collection techniques to manage procedure frames.