Slack Hurd
 

Lately, I've noticed an upsurge in interest about GNU's Hurd OS. Arch has extended support to Hurd, i.e. www.archhurd.org, and Debian has been doing so for years. Over the past 5 years, I've grown tired of monolithic kernels like the one included in Linux. Evidently, I'm not alone... thankfully. I've been tracking both Minix 3 and GNU's Hurd and both are rapidly progressing into usable operating systems.

So ... if Pat eventually releases a version of GNU's Hurd, I vote for the name, SlackHurd. :)

Pat, give us SlackHurd!

Linux is a kernel, not an OS. Don't confuse things. Hurd is also a kernel.

Slackware Linux is an OS.

So, the name would be Slackware Hurd (if the Hurd will ever become usable).

Obviously ... no confusion exists at least to me. ;) For the uninitiated, Hurd uses the Mach micro-kernel and GNU/Linux uses the bloated Linux kernel... ;)

Nah, I want SlackHurd ...

Bloated? Compared to what?

I don't think the Linux kernel is bloated. It simply supports a hell of a lot of stuff, which is necessary if you're going to compete in the real world.

Hurd supports next to nothing, and hence is less "bloated".

The Hurd project is an interesting one, and I hope for its success. But it does not, IMHO, belong anywhere near the Slackware brand, yet.

Oh, and why have you grown tired of monolithic kernels? If it works, then what is there to grow tired of?

I am always interested in whats going on or not going on in our world... the demise of Open Solaris by the evil Oracle is interesting,especially when there are two forks emerging,illuminos and Openindiana...

ps I have been waiting for a sign of HURD for donkeys years.

Not really, its just Arch.
Highly doubt Pat himself will start a project like that. All other ports are efforts of other individuals, even x86_64 which has been made official eventually, but thats entirely different.

hurd may be ready shortly after grub2 is... been waitin' 10 years already.

I don't mean to start a flame war, but is GNU Hurd just an attempt to get rid of the name Linux, since GNU has always been bitter as long as I can remember that the majority of people do not call Linux OS's GNU/Linux, but simply Linux, since Linux can mean the kernel, or a distribution?

Which will we see first: HURD or Duke Nukem Forever?

If you look at the L4linux project, it aims at porting the Linux kernel to the L4 µ-kernel API. It is a "(para-)virtualized Linux running on top of a hypervisor."

http://os.inf.tu-dresden.de/L4/LinuxOnL4/


I haven't tried it or know what would be involved in converting a Gnu/Linux OS to use it, but it says it's "binary-compatible" with any x86 gnu/linux distribution.

But really, what would be the advantage of this? The linux kernel would still be running as it self. Unless the L4 hypervisor can relaunch the linux kernel if it crashes, I wouldn't see any extra stability.

Duke Nukem Forever.

@ lewis,

as I understand it, GNU's original intent was for the Hurd kernel to be the kernel used in GNU operating systems (those things we now so often tend to refer to inappropriately as "Linux" operating systems :) ) - but for whatever reason(s), the Hurd kernel wasn't ready for mainstream use (or any use?) while the Linux kernel had been created and was becoming usable and was under active development and attracting more developers. People began creating operating systems (or maybe 'application suites would be a better term) around the Linux kernel. So, the Linux kernel became the kernel for the many GNU operating systems.

So, it isn't merely a name thing - it's actually a kernel thing. There's a link around here, I believe it was a story that appeared in the News forum on LQ, posted by LXer magazine, which leads to a recent interview with Richard Stallman, wherein he (Stallman) talks a bit about the early days of the GNU OS and of the FSF (Free Software Foundation) and about the situation regarding Linux vs Hurd as the kernel in our OS's - have a look around for it; if you can't find it I'll have a look for it later on. In the article, he says something like (I'm not quoting except for the words, works OK), "The Hurd kernel was intended to be the kernel for GNU OS's, but the Linux kernel works OK meanwhile".

The attempt to get rid of the name linux seem to have started at a very early point:
from here:
http://www.gnu.org/software/hurd/hurd-talk.html#int

Microkernels "appear" to be superior way to go but if you look at the development -- everyone tries to make a true microkernel then several years later the development ceases. It's almost like development reaches a point and then the problems in microkernels are too difficult to overcome and the project ceases.

It kind of reminds me of communism vs. capitalism. One sounds really good in theory but only one works well in practice. One is always way behind and it looks like it will never work as good, in reality, as the other.

@ GrapefruiTgirl

Fair enough explanation. I realize that GNU wrote the majority of the tools in a CLI only Linux environment. Other organizations wrote tools/software also though. And in a graphical environment, there are tons of orginizations/individuals writing the software, can we give everyone credit in the name of the OS? Linux is just a name that stuck, it's not taking a way credit from anyone in my opinion.

Ok, I secede now, instead of Slackware, we should give everyone credit who wrote software, lets call it: GNU/KDE/Fluxbox/OpenOffice/wicd/Linux.

Do you see my point, Linux is just a cool sounding name. When I use the term Linux for OS instead of for kernel, I am not trying to give Linus all the credit, or I am not trying to withhold credit from anyone, it just works, is easy to say, and sounds cool. Instead of talking to my buddy and saying, you should run "Genoolinux" on your PC.

Ok, I am going to quit now, :-)

Heh, no I understand what you're saying. I'm as bad as the next person for referring to the GNU OS's generally and loosely as "Linux", and also with no disrespect intended. Verbally lazy I guess.. There are 1000's of contributors to thank for what we have as an OS today.

Much of what I wrote above is based on what I recall from that RMS interview, so if there are problems of any sort with what I wrote, it's entirely my doing. ;)

By the way, my vote goes for "Slackwhurd" :p

On another note, I've finally watched 'Patent Absurdity' from the GNU site and it was worth watching. :)

<offtopic>
Just for the record, that's illumos not illuminos. OpenIndiana which was released today is the second distribution (soon to be based) on illumos. They aren't that much forks in the sense their (current) goal isn't to diverge from their original counterparts (Solaris OS/Network consolidation and OpenSolaris/Solaris 11 Express) but on the opposite to be as much as possible close to and compatible with them. They are often called sporks ...
</offtopic>

I can understand that with the small memories on computers (as compared to today)in the 80s how micro kernels would be the way to go
but today we can afford to waste a few MB on machine code that will rarely be used

like the justification of using assembly code to write whole applications is over so should be the justification of micro kernels

That's not the rationale behind micro kernels.

OK then what is the rationale behind micro kernels

You can easily run your system without KDE/Fluxbox/OpenOffice/wicd.
You can't run it without Gnu.
Skip to minute ~50:
http://www.archive.org/details/Richa...anchester.2008

So, you can easily run it without those packages. But:
Think about what you will have got without the packages mentioned here:
http://en.wikipedia.org/wiki/List_of_GNU_packages

Sums it up:
http://forums.debian.net/viewtopic.p...art=15#p316716

It's encouraging to see you qualify your sentiment with the term, yet. ;) However, I'll have to try Arch's latest version of the Hurd before deciding if it belongs anywhere near the Slackware "brand". :)

I could go on for ages and already have in several threads. One of my VIA motherboard computers wouldn't even boot for more than 6 months around 2006 while kernel developers decided on the most appropriate course of action... Lately, I've been plagued with optical disc mounting problems with my older nVidia MCP motherboard and SW64 13.1 ... probably another kernel problem.

Maybe, note the maybe. Regardless, I aim to give Arch's version of the Hurd I try. From recent experience with Linux kernels, the Mach/Hurd experience can't be much worse. ;)

Probably... I suspect Slackware will be largely ported to the Hurd eventually even if done by other parties. Personally, I'd love to see it happen ... now! :)

You might want to try Arch's latest version of the Hurd first? ;)

Well, we have Apple and QNX as successful distributors of micro-kernels ... Micro-kernels do work in desktop operating systems. Even pico-kernels such as Forth are possible to use.

Try reading this, which was pointed to in an earlier post.

Agreed ... simply a shortened version of a name and human nature to do so. ;)

Yeah, but it won't look like anything that you're used to. Unless you're used to Forth. (I ported a copy of Laxen & Perry Forth 83 to a TRS-80 model 4p back in the day and have a copy of pfe installed.)

Arrrrrrrgh, no! SlackHurd (Slackard) or SlackHur (Slacker) ;)

Terrific, Richard! We'll likely agree that Chuck Moore is brilliant.

Modularity and Security/Reliability.
The modularity concept was obviously a clear advantage and has been adopted by most so called monolithic kernels like Unix and later Linux by the means of kernel modules that can be loaded and unloaded on demand.
About security/reliability, the idea is buggy code running at the kernel level has the power to crash the whole machine while the same code running in user mode can be restarted after failure with minor disruption if any. Isolating portions of code prevent also components to misuse/lock shared structures and objects. The current trend toward virtualization (hypervisors, containers and the likes), which is similar in concept shows that the idea was not that bad after all.

http://www.cs.vu.nl/~ast/reliable-os/

I don't have much experience outside of the GNU Linux world, but I am guessing there are plenty of packages out there that can replace GNU, such as ones used in Unix or BSD systems that can easily replace the GNU equivalents. Let me know if I am wrong.

Sure. You could even replace all those base ones with low feature versions from busybox. I have a PoE camera that does this. Additionally, slackware's initrd images use busybox.

You just have to be very careful about what features you try to use if writing a system script. Like recursively searching files for a string has to be done with
[code]
grep string `find -type f`
[code]

essentially passing all file names as the list of files to search.

EDIT:
It seems like newer versions have the recurse on grep. My camera doesn't have it.

(edited to remove already answered questions)

Why reach out to Pat? Why not start your own Hurd distro? Base it on the Slack philosophy maybe, but do it your self?

As far i know: no, there are no such packages.
Without the Gnu-base system and development you get exactly that:
Nothing.
Inside or outside of a system with a Linux kernel.

In Debian you can easily replace the Linux-kernel with a kFreeBSD-kernel: You still use Gnu. But you can't do it the other way around.
To a certain degree you can make use of the Hurd too.

So: afaik you will need Gnu.
There is no way around it.
As long you want a free OS, of course.

I agree all mainstream and almost all other Linux based OSes are strongly depending on Gnu software and are so Gnu/Linux OSes. There is however the heirloom project with the goal to provide a Unix (non Gnu) userland:

http://heirloom.sourceforge.net/index.html

The *BSDs cut down their GNU dependence quite a bit and if put in a rough spot they could probably live without GNU as well (though at the moment they'd be missing a few packages). Of course I'm only talking about the base system. gcc is a big package but there are other compilers out there, and though living 100% GNU-free *right now* in BSD would be a little painful (since some of the possible BSD replacements are not as good as the GNU tools [yet?]), saying you cannot live in a *nix system without GNU (at least as a base system) isn't 100% correct. Just as GNU/Linux is more than Linux, it is also more than GNU, and the open source community is more than both.

Good point with QNX but I think atleast with Apple and several other attempts they would be categorized as a hybrid kernel, like a monolithic kernel on top of a micro one -- still seems there have been many attempts to make a mainstream microkernel with very little success, currently.

Android is a Linux system, and it uses zero GNU packages.

So wikipedia is wrong?
The very first sentence says:
http://en.wikipedia.org/wiki/Android...ting_system%29

Wikipedia is wrong, I got my information directly from the GNU.org: http://www.gnu.org/gnu/gnu-linux-faq.html :D

Here is the quote from the GNU page:

I have this HURD article bookmarked, I think it may be the one you mention:
http://www.h-online.com/open/feature...e-1030942.html

This would sound like an interesting project, and not to dampen anyone who is interested in HURD but, since HURD itself is just so far (very very far) behind in development (years), compared to GNU, is it really worth it? Considering the amount of development (or lack thereof) on HURD, it just seems rather irrelevant except perhaps in a case of educational or just pure curiosity. Other than that, I just don't see any real practical reason for HURD.

How is Hurd far behind in development compared to GNU, since Hurd is GNU? Maybe you meant it is far behind compared to Linux?

Yea thats what I meant sorry.

How about other microkernels like the L4 family? Aren't they more advanced technology wise (and in more common use)?

There is no practical reason for HURD, at all. Supporters of microkernels may argue that they are more secure than monolithic kernels (in theory), but the fact of the matter is Linux is legendary for its stability. Compare it to a "hybrid" kernel like the one used by Windoze and... well... you see my point.

If you ask me, GNU has created some extremely useful projects vital to FOSS and Linux, but they have failed at their original goal. They use an elitist, "cathedral" model of open source development on a lot of their projects-- sure, it's open, but only a few people can work on it. Case in point is EGCS, the community fork of GCC (which quickly surpassed the latter and was adopted as the "new" GCC by GNU).

Linux and HURD were born at around the same time. (HURD was earlier, but we'll give Stallman a few years as a handicap). Linux was created by a computer science student as a hobby, GNU HURD was supposed to be better than UNIX and lead us to a world of free software.

20 years later, Linux is widely in use and is legendary for its stability among other things. HURD is barely usable. Actually, not usable in many regards.

I would be perfectly happy if someone created replacements for GNU userland utilities so we could be done with the whole "I use Slackware as my distribution of the Linux kernel of the GNU operating system" fiasco.

That's sort of silly. For one thing, those of us who use Slackware are using "Slackware Linux." That's its name. To be correct then, Slackware Linux is a distribution of GNU/Linux. See? That's easy.

More to the point, the GNU userland utilities are pretty darned good. The only reason I'd want to see them replaced is if better ones come along.

I guess CMake is making a run on the GNU build system (autotools, make, etc.) for certain applications, but I'm not sure it's better.

Regards,

It's easy, but it's stupid. GNU shouldn't be forcing the term "GNU/Linux" simply because they can't write a functional kernel. I understand that most of the userland is GNU and that Linux is the kernel, but what does that leave us with? GNU/BSD, GNU/Hurd, GNU/OpenSolaris, the list could go on. Open source software is meant to be shared. Let Linux be Linux, BSD be BSD, etc.

GNU utilities are some of the best-- because entities outside of the FSF are working to improve them. The only reason I would like to see GNU utils replaced is so I no longer have to deal with Stallman's ego. I'm afraid I've lost all respect for the man recently, no matter how many tetris-playing text editors he's created.