“So what is the state of the Hurd? Is it vaporware, like Duke Nukem Forever? Fortunately not: the code exists, there is still work going on (for instance as part of Google Summer of Code), and there are even some relatively functional Hurd distributions. Let’s look first at the code and the current architecture, and then at the Hurd distributions.”
I remember reading that Duke Nukem Forever is due out in 2011, eventually we’re going to need to find a new piece of vapourware.
As for Hurd, it’s probably not anywhere near as popular as even some hobby kernels infrequently advertised here on OSNews.. but no doubt there is a niche it will fill, like embedded on a probe sent to Pluto, just as an excuse to include RMS as part of the payload.
I think that GNU Hurd will replace anyway in time SkyOS (as vaporwarre) and at least regarding Hurd is that is a released kernel. Anyone whom understands it will share it’s ideas and will make a good base for future development of other OSS kernels (including Linux)
Pixel Image Editor – best vaporware of them all.
SkyOS exists and run. development halted but “it works” and anyone can run it.
Vaporware is software that is claimed to exist but no one has seen it or can run it except the people who make claims about it.
Sound like those guys talking about bricked devices while u click reset and “unbricked”
Oh god, i shut down my phone buttons don’t work except power ITS BRICKED O LORD HALP.
Stupidity at it’s best.
Well, yes, that’s what people say.
Not really, as there is no way of obtaining it. When you click “Get SkyOS” on the web site, you are sent to http://skyos.org/?q=node/4 where it says: “SkyOS is currently not available for download, for additional information please refer to this page.”
This message has been there for more than a year now.
Hurd runs too. I am running it right now in VirtualBox:
http://i51.tinypic.com/23l443.png
That is Debian GNU/Hurd. I couldn’t get ArchHurd to work, but then again, I didn’t try too hard. When DDE gets ported (hopefully in a few months), I will run it on real hardware and I hope to start contributing.
Develompent is active, and as I mentioned above, current focus is porting Device Driver Environment to GNU Mach.
http://wiki.tudos.org/DDE/DDEKit#Overview
That will enable running Linux and FreeBSD drivers unchanged as a user-space process.
Heh, you have to love it when people try to be clever. Obviously it should be ‘who’.
Why do I feel like I’ve already seen this article posted here before :-s
I’d really like a *simpler* description of HURD’s architecture though.. The manifesto was a bit tough for me to understand.
The code for duke nuke’em has also been consistently worked on since the project started on, and there have been several playable demos made for the press. There really isn’t much difference between the two, but my bet is that the duke will be kicking ass and chewing bubble gum long before we have usable HURD machines.
The problem with the Hurd, even if they do get it working perfectly, is that it’s a solution looking for a problem. Sure it may be conceptually better in some regards, but we already have serviceable OSes for virtually any market segment you could imagine it filling.
This one is far more instructive and correct:
http://www.h-online.com/open/features/GNU-HURD-Altered-visions-and-…
Little know fact: there is one production-ready widely used microkernel: the EKA2 at the heart of Symbian.
Creating a kernel is really a question of getting things right, rather than man-months of work: the EKA (symbian), XNU (OSX) and Linux kernel were all written in a short time by small teams, while the Hurd team, and also the behemoth Microsoft struggled (and often failed, as MS failed several times to produce a succesor to the NT kernel). So, getting a working Hurd may happen, one day…
This post is pretty much completely wrong.
OSX is based on Nextstep, and they didn’t write it from scratch in a very short time, they bought the damn thing and modified it. Before that Apple tried several times to replace the classic Mac OS, please google Pink, Taligent and Rhapsody.
MS has never tried several times to replace the NT kernel, they did however write it from scratch, and in 2001 used it to replace the Win9x line of Windows. Windows 7, Vista, XP and Win2k are all NT. NT is a modern kernel, with modern features, it would be stupid for MS to try and replace it.
Edited 2010-09-21 23:13 UTC
Yeah, and the NextStep kernel appeared out of thin air?
It was a huge rewrite of Mach mixed with BSD, done by a small team lead by Avadis Tevanian in a couple years. Huge enough to be considered as a “new” kernel.
And the Longhorn kernel was supposed to be a complete rewrite, but they dropped everything and went back to improving NT.
This is only my opinion, but the NT kernel is a very modern one, the userspace is mostly the problem here.
I would have better faith in the minwin initiative than in a new kernel.
The NT kernel traces its roots back to VMS for the VAX, by Digital Equipment. It’s especially obvious in the common characteristics between NTFS and Files-11.
Not entirely – at least, not in the sense that NT was an evolution of VAX. It’s just that the guy who did VMS went on to do NT at Microsoft.
Of course, using the logic of the OP, the BeOS kernel owes a lot to Mac OS, as a lot of the engineers were either ex Apple engineers or Mac developers in the beginning. In other words, as Thom says, take the “VMS” with a pinch of salt 🙂
Very true. Having actually worked on VMS it does make me feel some affinity toward NT an NTFS… tho’ I think RMS over Files-11 blew NTFS away. It was so nice to work with. Just as DCL blows DOS CMD away. In fact just about everything to do with VMS and the layered services and apps blew Windows away.
(I am a little biased in that regard).
And also as the poster below you mentioned… the relationship of Mac OS to BeOS, ditto. BeOS was so much better than Mac OS at the time… (opinion)
Of course BeOS was better than System 7. Why else would Apple consider buying Be Corp.? They went with NeXT Software because JLG demanded more than Apple were willing to pay. (Although they ended up paying more than twice as much for NeXT.)
True… and they also got a CEO who has since led them nowhere but up. I doubt Jean-Luis (spelling?) would have given them that.
Probably a very fortunate turn of events for Apple. Quite apart from getting a good OS, they got Jobs back. Love him or hate him, the man has transformed Apple. He has added $200bn to the valuation of the company since he took over.
NeXT was worth more than Be among other things because they actually had a proven product, a very very very good technical staff, and at the time NeXT was actually a profitable company. And things like WebObjects were taking off. Never mind that Steve Jobs was actually a good business man.
By contrast, Be had an unproven OS which booted real fast to basically do nothing. The programming and dev tools were nowhere near the depth and breath that NeXT had (compare the openstep API vs. the Be half assed C++ libraries and weep). It has absolutely no credibility in the business sector. And was ran by Gassee who had almost ran his Apple division to the ground. And whose bad business skills were showcased when he completely f*cked the negotiation with Apple, thinking he had the upper hand over both Apple and NeXT. What a maroon.
I feel bad for the engineers who worked for Gassee though.
No BeOS at the time was very primitive. Poor network support, poor printing support, wobbly kernel. People compare BeOS 5 with the Mac OS but at the time when Apple purchased NeXT the BeOS was very rough.
Though in time it go to be great.
Yeah, I ran it with Bone and some other modifications, but it was… I don’t know… “fun” to me. I still prefer the interface over Mac OS. I still hold out for Haiku to go Beta.
VAX is a computer architecture, I think you meant VMS its intended OS. Right?
It was never supposed to be a complete rewrite. Google 3 pillars of Longhorn.
Also, NextStep was on version 3.3, released in 1995 when Apple bought them. It was first realeased in 89, and Apple bought them in 99. NextStep also ran on mach and BSD. Mach predates even Nextstep, it was written at Carnegie Mellon, and that project went from 1985 to 1994. BSD is based on the original Berkley Software Distribution, which was written at uh..Berkley, and was first released in 1977.
Doesn’t sound like quickly, or a small team to me. All this information can be found through google.
Apple bought next in 96 and released Rhapsody in 98.
They certainly did not buy Next in 1996, they bought it in 1999. Rhapsody was released, as OS X, but that happened in 2000.
Actually, I looked it up again, and I was wrong, it was 1996, not 99. Sorry
And NextStep was in version 4.0 and named “OpenStep” by then 😉 I still have the box somewhere…
Openstep was the API:
http://gnustep.org/resources/OpenStepSpec/OpenStepSpec.html
http://en.wikipedia.org/wiki/OpenStep
Dude, are you really trying to correct someone who actually ran OPENSTEP 4.2 on a nextstation. LOL.
For the nth time, with the advent of version 4.x NS was commonly known as OPENSTEP. The name is for both, the API and the underlying OS for what used to be NeXTStep (or was it NextStep?).
Deal with it.
So, I ran NextStep on a Cube, Do you really think you are the only person on this site who has experience?
http://toastytech.com/guis/openstep.html
Anyway, So they changed the name for version 4, big deal, you win, that doesn’t change the fact that OpenStep is the name of the API, or that GNUStep is also based on OpenStep.
I don’t mean to seem like I’m constantly correcting you, but no, wrong. OpenStep was released prior to the Apple acquisition. In fact, OpenStep 4.2 was the last release IIRC. Apple continued to ship OpenStep for quite a while after they owned Next.
The other thing of note is that OPENSTEP was a specification that Next created and a bunch of manufacturers supported. So there was a version of OPENSTEP running on Mach, Intel, Windows NT, some of the Sun hardware and some of the HP RISC based work stations. The Sun and WinNT versions ran on top of the underlying OS, the others ported the entire OS (IIRC). I’ve used the NT version and it works reasonably well, even today.
OpenStep is not NextStep. OpenStep is the API and Spec that NextStep used. And just like NextStep, it WAS released prior to the acquisition, because it was first released in the 93. I never said NextStep or OpenStep wasn’t released before the acquisition, so I’m not sure what your point is. OpenStep is also what GNUStep is based on.
NextStep is the OS and OpenStep is the API.
I don’t think you are correcting me, because you aren’t quite right.
Actually, henderson101 was almost right about OpenStep and OPENSTEP, he just had them reversed: OpenStep is the spec and OPENSTEP is what NeXTstep was rebranded as in the next release after the spec was out.
http://www.facebook.com/photo.php?pid=7339524&id=571610449
Actually the previous poster was correct.
NextStep stopped at version 3.3, 4.x and on were rebranded to OpenStep (or OPENSTEP I always got confused with Jobs’s OCD capitalization shenanigans during that period).
I actually have the box with the OS somewhere in the basement.
http://gnustep.org/resources/OpenStepSpec/OpenStepSpec.html
If I’m not mistaken, the bunch was limited to NeXT and Sun – the two companies that collaborated on the specification. Sun did the version running on top of Solaris with some help from NeXT presumably, while NeXT did the one running on NT as well as the full operating system running on 4 different architectures, including Sun boxes. Sun soon lost interest in favour of Java, though
This series is quite interesting, btw;
http://www.youtube.com/watch?v=KLHr6Z35t1Q
No need to correct him here, he is saying the same thing you are, just with more mushy grammar. His sentence would read better as “NextStep was on version 3.3, which was released in 1995, when Apple bought them (in 1999).” He’s not saying 3.3 was released after Apple bought them, but that it was already released at that point
It was supposed to be a rewrite by rumor only. Microsoft’s statement on the issue was something along the lines of “we always planned to work off of Vista’s kernel, why would we abandon it after so much hard work?”
Additionally while MS Research has produced a many different kernels none of them have never been considered for implementation on commercial platforms irrespective of how good/interesting they were.
Even the most recent Singularity project which garnered a fair amount of interest was dismissed as only a test bed for incremental changes to the present kernel.
Not stupid.
MS has been working on alternative OS (including new kernels) for a while. Take a look at singularity (http://en.wikipedia.org/wiki/Singularity_%28operating_system~*~…) for example.
They’re “only” research OS but they’re pretty damn cool IMO. And so who knows, the NT kernel might be replaced someday (along with a lot of the legacy)
singularity was before midori. They’ve actually talked about migrating to midori.
http://en.wikipedia.org/wiki/Midori_%28operating_system%29
Well, OpenStep {OS, not specification}, but essentially correct. NextStep’s API was not OPENSTEP {Specification, not OS} compliant. The API in OpenStep 4.x is what Cocoa is based on/morphed from.
True.
Yes! Which was the project a few Be Engineers were working on before they joined Be Inc. There was Blue too, apparently.
Technically, Taligent was what Pink became after Apple lost direction and took the “partnership” route. I have no idea if they shared a line of code, but I don’t think Pink was an actual OS at the point they changed direction.
What you have COMPLETELY missed is Copland. This was what was to be Mac OS 8 and what was actually released Mac OS 8 ended up raping to steal a lot of the “new” features. I used to have a copy of Copland, but I never got it running because it required a serial debugger and I couldn’t be arsed to mess about with it.
Wrong. Wronger that a wrong turn in wrongton. Rhapsody *IS* Mac OS X. Rhapsody is Apples’s “first go” at making OpenStep in to a Mac alike OS. The entire system is pure STEP, it just has the Workspace manager with an Apple style menu and platinum style Icons. In Fact, Mac OS X Server 1.x looks exactly like Rhapsody, and the initial developer released of OS X have a very similar “Finder” to Rhapsody. It wasn’t till 3rd or 4th Developer Release that it started to look OS X-ish, and not till the Public Beta that it really was OS X as we know it.
The Windows NT Kernel has been altered to varying degrees (sometimes beyond recognition) on a number of occasions.
1) NT 3.x > NT 4
2) XP > Vista/7
3) Server 2008
4) Longhorn (aborted)
5) Windows Mobile
NT 3.5 did not integrate the GDI. How much of a rewrite do you think it took to put the user land GDI functionality in to the NT4 kernel? It was not “trivial”.
As to that last statement – it’s debatable. I’d say, Windows 7 is the most happy I’ve been about Windows since Windows 2000 (which I used for about 6 years.) XP just seemed like tweaks over 2000 for the most part (yeah, super over generalisation, but let me have that one please 🙂
Just because Rhapsody became OS X, does not mean that it was not an attempt to replace classic Mac OS, it was just successful
Windows 7 is NT.
Widely known fact: there is a production-ready widely known used microkernel: QNX’s kernel.
The article fails to trace the DDEKit work and ViengoOS work. Moreover sticking to Mach seems a dead end unless they aim to create an OS utilizing Apple public available sources (which I think would be a good approach).
The Apple XNU kernel is not a microkernel.
The fact they use (only parts of) the Mach kernel is misleading, it is an hybrid kernel like NT & others (linux, in a way, with FUSE).
The Mach kernel is not always a microkernel: the inter-process communications had so much overhead that the easy way to improve performance was to fuse the servers back inside the kernel, thus going back to a monolithic (or at least hybrid) design. At one time, the Mach “microkernel” was so not “micro” that Jonathan S. Shapiro coined the name “nanokernel” to designate true microkernels.
Edited 2010-09-22 10:30 UTC
Yes, I know.
Yes, I know.
But they could port more easily parts of it including IOKit CFLite and other parts free parts to Hurd, like a more complete puredarwin by filling missing parts with Hurd.
The article is about the vaporware and the never ending coming out “The Hurd”. But you people go on a tangent and now are talking about all other types of OS. Stay focus boys!
By the way, the only reason that there is still work being done (albeit very little) on the Hurd is that crazyman Stallman does not have the guts to kill it. He is too jealous with Linux success and don’t have the heart to do it hoping that maybe, somehow, one day they will finish the Hurd and he will try to push it down people’s throat as a Linux replacement.
One last thing, if you call Linux, GNU/Linux, you have bought into Stallman craziness too and drank all the cool aid!
I don’t think so. You attribute way too much power to Richard Stallman if you think that he has the power to drive so much people on a ever-failing project.
My theory is that Hurd, like some research projects in the OS world, exists because because computer science has its equivalent of theoretical physicists. People who want things to be done right, no matter if a quick-and-dirty hack exists, works well, and is widely used.
If a Hurd kernel existed and was usable on a wide range of machines, it could be a better option than Linux for many use cases. Because it would not be bloated yet, would have a more secure and robust microkernel infrastructure, and would be an easier codebase to work on. That’s because of this “if” that many people still work on the Hurd project, and that’s also because of this everything-should-be-done-right attitude and attempts at code reuse from various project that it didn’t shipped a working kernel yet.
You could say the same thing about DragonFlyBSD and it is progressing at a much faster rate.
I agree with the parent that HURD exists for non-technical reasons. Declaring HURD a dead end would be too painful for Stallman. He is extremely jealous of the success of Linux as can be seen by how he yelled at a kid over not calling it GNU/Linux. The guy is a kook and there are better and more interesting alternative Nix kernels in development.
Yeah, but that’s BSD. Some devs just prefer GPL, a current rationale being that they don’t want to give code to big corporations for free…
<insert licensing debate and history of Webkit and here>
Maybe the reasons are not technical, but not linked to RMS either ?
Edited 2010-09-26 21:22 UTC
I shouldn’t feed the troll but I will say the facts anyway.
Stallman can’t kill Hurd, it is a free software project and it has life of its own. FSF has long ago stoped hiring hackers to work on Hurd, it is entirely hobbist project now. Last FSF-paid Hurd hacker was Roland McGrath who moved to work for Red Hat in 2002 (he works on glibc now).So, Stallman have done his part on killing Hurd by not spending FSF money on it! Are you more happy with Stallman now? Do you still think he is crazy?
Why am I even asking, you delusional RMS-haters wouldn’t rest until see the guy dead. And then you would attack him because he died.
And one last thing: I will call the thing GNU/Linux just because I don’t want buy into crazy RMS-haters’ craziness. You have bought into someone else’s craziness which is way more dangerous than RMS’s.
And what is crazier than yelling at people who do not call it GNU/Linux, comparing proprietary developers to violent criminals, refusing to surf the web with images and eating your own toejam in front of a group of students?
So, is Stallman too jealous on Linus Torvalds, heh?
Stallman started the open source Unix operating system. It was done until 90%, and then a teenager with freckles came in and did the last 10% and got all the credit for all the many years of hard work from the GNU guys.
Who would not be slightly irritated? Every non computer guy thinks today Linus T wrote the entire Linux from scratch! I even met some computer guys thinking that too “Linus is super, who is this Stallman, why is he shouting? What has he to do with Linux? Ask Stallman to shut up”.
I get very irritated when I see kids nowadays trashing Stallman because he actually has the temerity of believing in something. You may agree with him or not, but GNU is his baby and he has a track record to back up his stances. I may disagree with him, but my respect he has earned. Some people nowadays seem to consider anyone who actually stands for something as being a threat to their entitlement to apathy.
Most of these kids do not comprehend that all those GNU tools they take for granted, used to be not only very expensive… but rather exclusive tools. I remember how back in the late 90s eve, SGI for example would charge a pretty penny not just for the compilers for Irix, but just the libraries themselves, on top of the hefty licensing costs for the OS and the HW.
It is easy to trash the farmer once the salad has made it to one’s table and is ready to be eaten.
Edited 2010-09-25 05:24 UTC
If anyone should be called kids it should be the people who follow him since they have a child like understanding of economics.
The heaven’s gate cult leader also had strong beliefs, maybe he should have been admired as well.
Without GNU the world would just have used FreeBSD and would probably be better off for it since it uses a standard base and wouldn’t have encouraged endless sound stack and package format wars.
Stallman does not have viable alternatives to existing software development models. This can be seen in his recent answer to proprietary game development which is to lower your standards.