KernelTrap reports: “Roland McGrath recently announced version 1.3 of the GNU Mach kernel, offering several bug fixes. He also noted, “We are no longer actively developing version 1.x of GNU Mach. We plan to make only necessary bug fixes or trivial enhancements in the 1.x line, and make further 1.x releases only as necessary for those purposes.” A new 2.x version of GNU Mach is under development, utilizing the University of Utah’s OSKit.”
…when the L4 jump ??
… should be about that OSKit itself! Why have I never heard of this? It looks like it is trying to be the lex and yacc for OS Writers ๐
See, we’re not all hicks and Mormon fanatics out here. There is some really good stuff going on at Utah. But… as an insider I’ll tell you that the modularity of OSKit is inspired by the dynamics of polygamist marriages (one wife for sex, one for dishes, another for laundry, etc…) It’s a micro-carnal approach.
its a lot of fun and you can learn quite a bit from it even if you dont actually ~use~ it for anything
as an insider I’ll tell you that the modularity of OSKit is inspired by the dynamics of polygamist marriages (one wife for sex, one for dishes, another for laundry, etc…) It’s a micro-carnal approach.
That’s a good one.
I don’t know nuch about OSKit, but isn’t is the thingy that provides platform/OS independent device drivers and device interfaces ?
IIRC it still uses linux-2.2 device drivers. Are there any plans to “update” them to Linux-2.4 or 2.5 ?
AFAIK the UDI-project is a little bit similar in approach; is there a UDI support for OSKit planned ? Or why do the GNU Hurd/Mach people not consider using UDI (directly) ?
just curious …
Couple of bits of info:
– OSKit is a project at University of Utah to build a set of components that will make kernel development easeier. It provides basic things like driver support (for linux-2.2 drivers) filesystems, basically everything you need to get a full OS up and running, all wrapped up in a nice component interface. You can use whatever parts you want, and write your own code for the parts of the kernel you’re interested in.
As for UDI, the FSF does not like UDI for philosophical reasons:http://www.gnu.org/philosophy/udi.html
The gist of their arguement is that UDI removes incentives for free OSs and drivers, since Windows users can use GPL’ed UDI drivers and it makes it easier for OSS kernels to use closed drivers.
Thanks alot.
NOW it understand
should be about that OSKit itself! Why have I never heard of this? It looks like it is trying to be the lex and yacc for OS Writers ๐
ergh.. the oskit is a bunch of source from linux, *bsd, mach, etc. its one huuuge mish mash. it has sever libc libraries in it, drivers from every os you can get source to and more.
its claimed to be modular… heh. its awful spaghetti. modular my backside.
there is a reason why its not used by anyone….
its also quite old too, if you are doing any kind of hobbyist osdev, you should have known of it.
I’ll stick with the tried-and-true monolithic carnality, thanks. The problem with the micro-carnal approach is that each server requires a full set of resources, but you’re stuck with a single “hardware interface”. And if one turns against you, it can still influence the others, bringing them down too.
The macro-carnal approach takes less of my resources, and is inherently safer. For variety I can use multi-boot-knocking. This also facilitates system swapping without having to give up half of my resources every time.
http://www.linuxjournal.com/article.php?sid=6105&mode=nested&order=…