Submitted by Not too long ago, Robert Szeleney put the development of SkyOS on a temporary hold. The challenges in keeping up with the ever-changing world of hardware support were simply too big to continue SkyOS then-current development model. As a result, Szeleney recently came up with the idea of using a Linux or NetBSD kernel as the base for SkyOS. Well, we’ve got a progress report on that one, and in true Szeleney fashion, a lot of work has already been done.
While Szeleney first talked about either a Linux or NetBSD kernel as a base, it seems like he’s now working on the Linux kernel – at least, the progress report only mentions Linux, it might of course be the case that he’s done work with NetBSD as well. In any case, the progress reports are posted on his blog, and the first one details what has already been achieved. The base SkyOS/Linux system (I came up with that all by myself…) has the following features:
- Linux 2.6.27.4 kernel with an init ramdisk using usplash to display and animate the boot progress.
- LSB with essential tools like coreutils, etc.
- DHCP, NFS, iputils, etc.
- Initial parts of the SkyOS C++ API ported to Linux.
- Ported the SkyOS messaging system (formally known as DataExchangePorts, now called MessagePort) to allow bidirectional interprocess communication.
- Initial appserver using svgalib as graphic backend (for VESA and other dedicated chipsets). (this appserver replaces the SkyOS GUI part which resisted in the kernel)
- Initial test application using SkyOS C++ Application class to register and talk to app server.
Szeleney also detailed what’s next for in this journey: “Teaching appserver key/mouse input, basic window handling” and “Porting rest of SkyOS C++ Classes (primarily all classes needed for Window handling)”.
Exciting news, and it seems like SkyOS development is on track again. I’m anxiously awaiting any first test releases I can play with. There’s also a small personal note that I’d like to make: I want to apologise for how I have approached the matter of the lack of progress in SkyOS. While I still stand behind the general idea of what I wrote, I could’ve presented it in a less aggressive and more polite manner. By not doing so, I needlessly put SkyOS and Szeleney in a bad light, and for that, I’m sorry.
At the same time, I also know full well that OSAlert and I are way too insignificant to ever negatively influence Szeleney or SkyOS (or anything else, for that matter). From the brief moments I talked to him, Szeleney didn’t really seem like the kind of guy who would care too much about what some silly blogger like me writes down.
I have a question for anyone familiar with the GPL: Will using the Linux Kernel force Robert to publish his source code? Just being curious.
No it wont force him to publish the code, just like nvidia dont publish their driver code and SciTech dont publish the code for their XServer.
Eh, Depends on how he’s doing it. Binary kernel modules for the kernel ( which is what Nvidia does) are somewhat controversial. I can’t explain it better than linus : http://kerneltrap.org/node/1735
Yeah, in theory Nvidia should open up their code but in practice Linus has said he is willing to look the other way – he isn’t particularly religious about these sorts of things. I can understand that he wants to be pragmatic about it but at the same time it creates a pretty bad precedent when one is willing to ignore some technical violations whilst complaining about others (although Linus not directly doing the complaining (as with the case of Tivo)).
It brings me back to what I said around 5 years ago that Linux needs to be re-licensed under LGPL so that there is definite clarity in the licence regarding binary modules rather than the merky “I’ll let it slide” position taken today.
Edited 2009-06-26 13:42 UTC
good job!. Im surprised this hasn’t happened sooner, there are a few windowing replacements for linux but none are that usable.
DirectFB maybe but my impression is it doesnt give a framework for GUI apps like OSX or SkyOS does/will.
http://www.cosmoe.com seems dead or merged into haiku?
– Anyway, glad to see this progression.
Congrats Robert, glad to see you back up to bat. Looking forward to hearing more as you go along.
“Will using the Linux Kernel force Robert to publish his source code? Just being curious.”
Yes, if he’s actually changing the kernel. No, not if he’s just writing various daemons to provide app compat.
Changing the kernel code has no effect on if the proprietary SkyOS code needs to be GPL’d too. It’s only the changes in the kernel he needs to release, whilst the rest can remain closed.
Good on you Thom for having the guts to apologise like that. I agree you may have had a point, but were quite aggressive about it.
I personally would have been quite interested to have seen work done on the NetBSD option (NetBSD is the OS I use the most after OSX), but this is just as promising.
OpenBSD would have been an interesting option if there was little interest in scalability given how much hardware OpenBSD supports. Hardcore security matched with a sexy easy to use UI would have been a great lynch pin for future sales.
Frankly said I don’t know what this guy is trying to achieve. Is this a hobbyist, commercial or whatever-else OS? I guess he just can’t decide, but he seems to be little greedy, which I don’t like at all when it comes to the human beings. I can perfectly understand the need for making money for living, but this guy’s behaviour is just beyond of my understanding.
He has a life, a family, a job.
He had a new addition to his family, development stalled.
Programmers often suffer from “writer’s block” as well. Or simply the lack of inspiration / drive. I’ve suffered this often as my own life was being formed.
I’m now lucky to code once a month for a day. I have to work to pay bills, I need to relax sometimes, and then I have to attend to family matters.
The only upshot is the amount of time put into planning my coding projects. I often plan a month or so for each little thing I code, which has greatly improved quality ( I wrote a basic FTP client first try without mistakes or intermediate compiles – Build 1 was lifted to version 1.0.0 ).
Of course, that single little app, for all intents and purposes, took me a month to write.
–The loon
Why are you calling him greedy?
Are you kidding?
Firstly, this guy asks for money for a Beta – Come-On! Even Microsoft gives those away for free…
Secondly, this guy promises a free full version – ah, yes, there’s a tiny catch: “(…) once available”. Aha. Yeah, right. Which millennium are we talking?
Thirdly, in the middle of everything (kinda like “yaddayadda … coming along great … blabla”) he takes a creative break. Really. What’s going on there – is he trying to run a business or get somebody to pay for his hobby? Or maybe he had enough fools pay for the beta, so one day he decided to spend all the money on Austrian wine (so I hope, their beer sucks! Really!) and now he suffered from a tremendous hangover.
And now Lastly, he decides to move over to Linux. Would have been smart in the beginning (where were his senses then? Of course it’s nasty to support all the different hardware – even the major player screws up constantly in that field), but then he wanted to sell his idea as a new OS.
Now what are we talking? Yet another Linux window manager with a few extras, yawn.
Dude, I hope all that wine didn’t have glycol in it…
So, after all this ranting … maybe you’re right – “greed” may be incorrect! He seems to be more like an ingenious fool, or a foolish genius? Or maybe just a fool who can program like a devil. Who knows.
Cheers (or better: Zum Wohl).
What Window manager? Looks like SkyOS/Linux will use its own graphic engine. Or not?
What I actually said is that he *seems* to be little greedy. From what I know he is trying to sold out his OS for some amount of money, although it is based on OSS efforts. I know BSD license is far more liberal for sure, but he decided to use GPL, which isn’t that liberal when it comes to keeping your own code closed and letting you simply sold your work [and that seems to be the case here, as an author wants to sold his OS in a closed-source model and earn some money for living].
Another thing is that SkyOS doesn’t seem to be any innovational. Sure it has a nice GUI and all, but that doesn’t make secure/desktop/server OS. It just has to be functional and secure [and putting your GUI on top of the linux kernel doesn’t make a wise choice I guess. NetBSD kernel would be far more secure, fast; and it has linux_compat mode too].
To [Be] honest, this thing seems to be quite interesting. I still like the idea of kernel extensions with no recompilation and stuff (just like the Darwin/OSX) but that is good to see. At least the SkyOS project didn’t just silently died.
How will this make SkyOS different from other graphical desktop environments such as KDE or Gnome? Kinda sounds to me like OS X which is built using parts from FreeBSD’s and NetBSD’s implementation of Unix
Well, the fact that it won’t use X is a great difference to me… Whether you consider that a good or bad thing is a different matter.
I’m wondeirng whether this is just another desktop environment which would replace xfce, gnome, or kde….or is it something that replaces GNU?
Basically if you run this… will it be called SkyBuntu?… or will it be called Sky/Linux rather than GNU/Linux?
nope, this will be comparable to syllable server edition.
it will use linux kernel for compatibility and some bits and bobs to get it all dance and rest of the toolset is his.
Okay then … he’s replacing two things. So if you called Kubuntu KDE/GNU/Linux, then this would be Sky/Sky/Linux…because he has his own toolset and desktop environment?
I seriously doubt he’s replacing all GNU Software. That would really be counter productive. I know you were going for Window manager/Userland/kernel. So it would probably be Sky/GNUsky/Linux.
Speaking of which, Gnusky would be a pretty cool name for the whole thing. ( pronounced either `new-ski` or `nuski`)
To clarify, parts of the Gnu Tool chain are used widely by many OSes including all of the BSDs. That is not meant to diminish the extraordinary accomplishments and hard work that the creator of Sky Os is putting in. Its just an emphasis on how good of a job has been done on many gnu utilities ( including gcc). They are what I’m confrontation with and having them available would increase my likely hood in donating ( whether required by license or not) to the project.
This will not be something like SkyBuntu. He will not be using X.org or any X server. He is using the Linux kernel to driver the rest of the operating system.
My guess about the GNU userland apps is that they will probably be used as well as this automatically gives you a large amount of apps. Also it makes porting software easier if you have the GNU userland.
Good luck to Robert.
p.s. Screenshots?
What’s the point in using the Linux kernel?
I always thought that diversity in the OS field lies in kernels.
What makes Haiku, or syllable, or SkyOS so special was the fact they are completely different worlds based on their own kernels.
At least this is what thrill me.
Now SkyOS, using linux kernel, is less interesting (to me)… It’s strange that nobody noticed this “issue”.
OSes aren’t kernels
If you think OS=kernel, then can you explain me how MacOS is close to FreeBSD ????
I regard my Mac as a Unix box with good support for commercial software. The kernel sets the personality of an OS even if the modern trend of running userland program in a vm levels this a bit.
I wholly agree… SkyOS used to be interesting for me until this announcement… Well, at least there’s still Haiku around
Except that Haiku is in perpetual pre-alpha and has a quickly detoriating code base. Unless they manage to bring in the original NewOS developer to fix all the breakage (esp. SMP), the project is doomed as far as I can tell.
Edited 2009-06-26 13:31 UTC
This is why some people would rather pressure the company that has BeOS’s asset to completely open source it or revamp it. There is a fear that Haiku doesn’t look like it’ll be matured.
Care to point out how any of that is true?
Any proof for your point, sir?
Well, I don’t pay a whole lot of attention to the project. But as an OSAlert reader I do notice it. They’ve been working on it for 8 years, and does Haiku really matter in 2009? Is it something I can install for someone and have them be at all happy with it?
“Doomed” is, admittedly, a relative term. But the Haiku project fits at least a few possible interpretations of the word “doomed”.
But then again, SkyOS is doomed, as well.
Edited 2009-06-29 04:46 UTC
My honest opinion here: You’re talking out of your ass.
Which basically translates to: “I only know what I hear from others.”
Haiku is quite stable, including on SMP systems… and code doesn’t “rot” – in fact Haiku’s codebase seems to be more modular and well-designed than most systems out there. I hear new developers coming into the project discussing how surprised they are at how well it’s designed and maintained.
I suppose this depends on the person. I’ve been actually *using* Haiku now for a couple years. Admittedly, it’s not the ONLY OS I use, but usable for sure. I think Haiku is not something you’re going to be installing for Grandma any time soon (just imagine if you installed Linux for Grandma 10 years ago when it was “8 years old”) – but it’s *very* usable already, and some claim it makes a hell of a lot more sense for noobs.
IMO, SkyOS was “doomed” for a different reason – being closed source.
Haiku is as doomed as the users/developers want it to be. If nobody uses it, or nobody develops it, then it will die. On the other hand, if it is both used/developed, then it will live on and remain relevant. That’s how open source generally works.
But that niche is now filled. By Linux. Now Haiku has to beat Linux or find its own niche.
I would be very surprised to see Haiku or SkyOS gain anywhere near as much public recognition as Linux has now. Especially since they are competeing not on the server, but directly with Microsoft (and Apple, and Linux, and *BSD) on the desktop. They don’t even register as also-rans except on niche sites like this one, visited by weird people like us. And it would surprise me if that status changed in the next 10 years.
Edited 2009-06-29 18:36 UTC
You’ve even changed the topic and context again… Originally, you hopped on a thread that was claiming that Haiku was broken and “doomed”.
First, you asked if it was relevant… Yes. IMO Linux hasn’t adequately filled said niche… and people want/use Haiku already.
Then, you asked if someone would be happy with it if you installed it for them… Certainly no less than Linux, IMO, possibly more. Let’s be clear, we’re talking about the usability of the OS here… and not necessarily the applications that run on it.
In the end, you just sound like the other people who believe nothing can possibly innovate and succeed in this day and age.
Edit: wording
Edited 2009-06-29 23:51 UTC
Hey! There’s Syllable too, don’t forget…
no everyone but hardcore enthusiastics it’s same what kernel an OS uses, what matters is if their hardware works with the OS. It wouldve taken decade worth of work hours from SkyOS to get hardware support to comparable levels as in Linux, therefore they changed to linux kernel. Normal users wont see the difference in any other way than the fact that the OS actually installs in their system now.
You could say it’ll be Sky/Linux instead of Gnu/Linux
Using the Linux kernel makes sense, sort of, in order to have access to the large selection of Linux drivers.
But how about the applications? Will the ‘new’ SkyOS/Linux have original and unique SkyOS applications and tools, or will it use the same applications as every other operating system?
Ok, so let me get this straight.
SkyOS is in the process of porting / creating:
Their C++ class library.
Their Message passing system.
An “App server” to handle GUI rendering as a userland process overtop a “kernel”…
Which happens to be Linux.
Is it me, or does this sound like the BlueEyedOS Approach to re-creating BeOS that died a long time ago?
Seriously cool, either way. I wish him luck.
Actually, this system is starting to become interesting to me now.
I’ve been following SkyOS since the beginning and this is definitely an interesting development. As long as Robert retains the rest of the SkyOS environment, I can’t see any downside to bringing Linux in as the Kernel. I’m looking forward to seeing the “finished” product.
I would not say I have a full grasp on the approach of moving an OS to a different kernel. I really think that the best examples would be OSX on Darwin.
It is true to say that OSX is Unix by citing http://www.opengroup.org/openbrand/register/brand3555.htm . I would not say it is your standard Unix though, but the took the core and built upon that.
Really, Linux is nothing without GNU. Many people complain that they call it Linux, when they should say Linux/GNU. Without the GNU tools/app/whatever Linux is just something (IMHO) that talks to hardware.
Isn’t that what Robert is shooting for? He is trying to save time to be able to focus on the GUI, not making sure that hardware works.
Just my thoughts….
Really, Linux is nothing without GNU. Many people complain that they call it Linux, when they should say Linux/GNU. Without the GNU tools/app/whatever Linux is just something (IMHO) that talks to hardware.
You’re quite missing the point. There’s people who use Linux and have built a complete userland system on top of it without any GNU tools in there. GNU tools are not required. Such OSes are f.ex. in embedded systems.
Sorry to be off topic but there were a few things said about Haiku that I feel need to be cleared up.
<div class=”cquote”>Except that Haiku is in perpetual pre-alpha and has a quickly detoriating code base. Unless they manage to bring in the original NewOS developer to fix all the breakage (esp. SMP), the project is doomed as far as I can tell.</div>
This is completely untrue. The main Haiku developers are really good at what they do and I can say from personal experience that the Haiku code base is some of the cleanest code I have worked with. Also the original NewOS developer aka giest, he contributes to Haiku from time to time and follows the progress.
As far as the perpetual pre-alpha, why don’t you give it a try. I have been using Haiku for several months full time on my development machine and have not had a crash once. Not to mention if you follow the Haiku mailing list you will see that releasing an alpha is really high on the list of things to do. Also our build system can produce iso’s so when we clear up a few remaining things it will be easy to make an alpha.
<div class=”cquote”>This is why some people would rather pressure the company that has BeOS’s asset to completely open source it or revamp it. There is a fear that Haiku doesn’t look like it’ll be matured.</div>
This is never going to happen. If Access was going to release the source code they would have done so by now. Haiku runs almost all of the BeOS software flawlessly and Haiku itself is becoming stable.
My last point is, please stop spreading FUD about Haiku. If you would like to know more about Haiku or see it progress go to the website, do some research ie read the mailing lists. Look at the commit log. Please though give some respect to the people who are working on the project and research before you write.
This sounds to me like an interesting direction for SkyOS to take and I hope it bears fruit. It certainly sounds like things are progressing quite quickly!
In partial response to those who are worrying about the move to a more common kernel, I had a few thoughts on this:
Although it’s nice to have lots of different kernels in the world, there is not that much variation in the popular operating systems. To clarify, there are many good kernels out there and they all have different feature sets, specialities, etc. Some are developed in different ways or using different tools / languages. Some are using different tradeoffs and some are arguably more cleanly designed. But the vast majority of popular kernels used on server and desktop class hardware share quite a lot in common – they’re written in C or a subset of C++, they share a single address space with limited protections. Many open source kernels have a Unix-like philosophy, which increases the similarity still further in terms of their behaviour and their syscall API.
I don’t know how SkyOS was architected but I have no reason to suppose that it was architecturally that different to the other players out there. I’m sure it was a nice piece of work, since all of Robert’s stuff seems quite impressive. But whilst SkyOS might be “losing” a unique kernel, the fundamental design of it is unlikely to have had massive benefits over just using Linux / BSD; presumably this is why Robert is considering this change of direction.
So overall, I’m not too worried about this. It’d be nice to maybe see the SkyOS kernel live on somehow as an option, or incorporated into SkyOS’s Linux kernel, or even released in source form in some way. But I think a large amount of what makes SkyOS clever and attractive is in the design of the other system components and the skill of its main developer.
Another project which is using the Linux kernel for a “non-Linux” OS is Glendix (http://www.glendix.org/). The Glendix people are trying to bring the main benefits of Plan 9 to more people by building on the Linux kernel to get better hardware support. They’re enhancing the kernel in ways it needs to support Plan 9 functionality. To my mind enhancing an existing kernel to suit your particular needs is usually going to be more efficient (although maybe not quite as fun).
There’s definitely still a need for alternative kernel designs out there (approaches using different languages, managed code, employing more sophisticated protection or extensibility, more completely microkernelised systems, etc). And there’s a real benefit to having smaller, fresher alternatives to the big players.