4 ms·

Flux OSKit: Reusable Components for OS Implementation (1997)

52 points by deepaksurti 8y ago

adamnemecek 8y ago
Reminds me of genode https://genode.org https://genode.org https://en.m.wikipedia.org/wiki/Genode https://en.m.wikipedia.org/wiki/Genode
snaky 8y ago
> 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.
rnd0 8y ago
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]. [1]https://wiki.osdev.org/Duct_von_Tape https://wiki.osdev.org/Duct_von_Tape
troglobit 8y ago
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.html https://www.gnu.org/software/hurd/microkernel/mach/history.h...
peter_d_sherman 8y ago
FYI: The main project URL listed on the site, http://www.cs.utah.edu/projects/flux/oskit/ http://www.cs.utah.edu/projects/flux/oskit/, results in a "page not found" error. The correct URL apparently is: http://www.cs.utah.edu/flux/oskit/ http://www.cs.utah.edu/flux/oskit/

4 ms·

Flux OSKit: Reusable Components for OS Implementation (1997)

52 points by deepaksurti 8y ago

adamnemecek 8y ago
Reminds me of genode https://genode.org https://genode.org https://en.m.wikipedia.org/wiki/Genode https://en.m.wikipedia.org/wiki/Genode
snaky 8y ago
> 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.
rnd0 8y ago
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]. [1]https://wiki.osdev.org/Duct_von_Tape https://wiki.osdev.org/Duct_von_Tape
troglobit 8y ago
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.html https://www.gnu.org/software/hurd/microkernel/mach/history.h...
peter_d_sherman 8y ago
FYI: The main project URL listed on the site, http://www.cs.utah.edu/projects/flux/oskit/ http://www.cs.utah.edu/projects/flux/oskit/, results in a "page not found" error. The correct URL apparently is: http://www.cs.utah.edu/flux/oskit/ http://www.cs.utah.edu/flux/oskit/