“This page is a place to keep track of all those things that we need to improve in GNU Mach, so that it is a reliable microkernel for The Hurd, both in terms of stability and performance. If you find anything missing here, please feel free to add a entry with a short description.”
I find it interesting that what little support and interest the HURD has gathered now is fragmented in such a fashion that there are at least two differnet kernels being worked upon, Mach and L4. I really don’t see HURDs relevance, but if it makes some hacker somewhere somewhat happy, good for him/her/it.
.. i’ll bet my great great great grandkids are going to change the world with it.
DevL, development on Hurd/L4 is stalled. It might be replaced by the Coyotos (www.coyotos.org) microkernel.
Any particular reason?
From an outsider point of view, it looks strange.
I’m curious as well. I thought L4 was a promising avenue of development. It already exists as a kernel for various platforms, so what went wrong where?
Nothing went wrong (that I know of
.
The Hurd is a collection of servers that doesn’t depend on a specific microkernel (at least it’s designed with that goal). Some people might prefer running it on GNU mach, while others might run on L4. When the L4 port is mature probably everybody will run it, but until that happens there are obvious reasons for improving GNU mach. Contrary to popular belief Debian GNU(/Hurd) can do quite a lot already and people want improvements *today*.
“I can (well, almost) hear you asking yourselves ‘why?’. Hurd will be out in a year (or two, or next month, who knows), and I’ve already got minix.” (1991)
* Notes: The Hurd 0.0 was released in August 1996.
FYI, hurd still isn’t really usable for desktops/servers
Other good Linus quotes here:
http://en.wikiquote.org/wiki/Linus_Torvalds
May be you should better ask Marcus Brinkmann (marcus) on IRC-channel #hug 😉
Appears to me RMS want’s to wrestle back Linux/GNU for the masses. I don’t think the linux kernel fits completely into RMS’s FOSS world view. Probably becasue Linus is not a pinko.<joke>
Well, it would be understandable. After all, his GNU OS is pratically eclipsed by Linux, even though the latter is dependant on the former. Not that his radical positions are helping his cause. The general public is not much interested in hearing rants on how immoral it is to develop proprietary software, they want something that works. Linus delivers that and gained corporate support/credibility. I believe it’s pretty much the key to success; on that aspect, RMS got some way to go…
The Hurd doesn’t seem bad, but it should stick to a kernel. Otherwise, it will stay in perpetual development. Anyway, why do they want to switch to another microkernel or revive this project, apart for performance?
The GNU operating system, aka RMS’s baby, is a reality, no matter what kernel it uses. Linux and HURD are two totally different ways of doing things and none of them has anything to do with RMS.
What are the advantages of microkernel(Hurd) over monolithic kernel (rest(90%) of the World) ?
>What are the advantages of microkernel(Hurd) over monolithic kernel (rest(90%) of the World) ?
That’s a question you shouldn’t ask. It’s like asking which one is better Vi or Emacs.
Wikipedia has some information on this subject, though.
It’s not a rest of the world versus the Hurd thing. Very few operating systems today use monolithic kernels. Of the ones that do there are the old ones (BSD, Solaris) and Linux.
Windows NT, Mac OS X, etc, use what some people call a hybrid kernel. There’s an article on wikipedia, like _LH_ pointed out. Hurd is trying to push further than Microsoft’s kernel and XNU, but that’s harder. It should make GNU a very interesting OS when it’s ready for production use.
The microkernel issue is really a long story. To make it short: It turned out that the existing L4 doesn’t provide a useful base for secure capability-based general purpose operating systems. New variants (L4ng and L4.sec) are underway, which are expected to improve on that; however, the L4 folks have little experience with those matters so far. Jonathan Shapiro — the lead developer of Coyotos and it’s predecessor EROS — on the other hand has *lots* of experience with such systems. That’s why many of the former Hurd/L4 developers now hope it will be a better bet than sticking to L4.
Note that none of this experimentation with new kernels, be it L4, Coyotos, or yet something else, will produce a full replacement for the exiting Hurd on Mach, or in fact anything useful, for several years to come — if ever. That’s why improving the existing and working Hurd/Mach is crucial, and in that context also this “Mach revival project”.
QNX has a microkernel.