Anyone have any experience with SkyOS?

[VirtualLarry]

Originally posted by: Nothinman
I was merely expressing my opinion that they're wasting their time on a project that has less of a chance of survival than BeOS.

But BeOS still gave the general OS community some new features to contemplate adding to other "surviving" OSes. So in that sense, the "experiment" was still worth it, and the effort not entirely wasted.

Originally posted by: Nothinman

The irony is, 5-8 years ago, NetBSD (and FreeBSD and Open386), were considered the only "real" free *nix's, and Linux was considered an immature hacker's playtoy, not suitable for anything else but hacking on the code.

There's no irony in that, unless you're thinking that in 10 years SkyOS will be at the same spot Linux is now.

I was semi-implying that, hypothetically-speaking. Who knows what will happen?

Originally posted by: Nothinman
But I doubt that'll happen because there's no grass roots support like Linux had. What made Linux work well was the fact that it was OSS, not that it was technically better. Infact it wasn't, originally it was written to run specifically on only the hardware that Linus had in his house and he said himself it it won't be anything "big and professional like GNU", go figure.

Now that is some irony. If only BeOS had been OSS, or at least partially, like Mac OS X/Darwin is, then perhaps it could have gotten the hardware support that it needed to survive. OSS is good, I do happen to like OSS.

Originally posted by: Nothinman

XFree86 has more-or-less always been a bit of a mess.

Which is why it was forked to X.org, so that progress could be made. Are you saying that it would have been better for the X.org people to start over?

I haven't actually dug through the codebase since XFree86 3.1, so I can't really say. But back then, it was indeed a bit messy.

Originally posted by: Nothinman

At the time, it did have far better "multimedia" capabilities that either Windows or Linux

In some spots it still does, but where did that get it?

I was just saying, that even though it "died", it made some contributions to the general OS community, so in that sense, it was still "worth it" for it to exist. (Kind of brings a parallel analog to issues of human abortion, in a way. Should those who have been born into the "hard life", simply be given up on?)

Originally posted by: Nothinman

In fact, I daresay, it prompted the creation of the Linux kernel-premption and low-latency patches, in order to attempt to compete on the basis of multimedia performance with BeOS.

Definately, I'm sure it had some influence. But look, the Linux kernel devs took an already well established source tree and altered it to meet their goals instead of starting over from scratch. And now Linux has lower latency that Windows in most areas and is probably close to BeOS in many areas.

Well, NT was and probably still is ahead of Linux in terms of fine-grained locking and pre-emption, but... I'm not sure how much weight your argument has, because Linux seems to have major portions re-written reasonably often (every two years?), and sometimes, throwing out the entire codebase and starting fresh is the best path to take. (I think Linux SCSI support, and things like the Tulip NIC driver, have had that happen to them at some point in the past, haven't they?)

Originally posted by: Nothinman

What to say, that SkyOS won't develop some similar level of enhancements, and be superior to other current OSes of the day, eventually?

It might, but it'll end up like BeOS with people saying "Oh, lets make Linux do this because it works well in SkyOS" and SkyOS will fade away also like BeOS.

That's a good point, Linux is like a bizarre hybrid chimera-like OS, the open-source OS equivalent of the Borg, sort of. The only thing stopping it from becoming the world's biggest software swiss-army-knife is the careful oversight and pruning work of the maintainers, keeping the core of it with a proper shape, almost like the pruning of a bonsai tree.

Originally posted by: Nothinman

Indeed, does Linux even yet have base-level support comparable to some of the things that BeFS was capable of doing?

Didn't Be drop the database FS for 5.0 because of speed issues? But yes, reiser4 should be able to do all that even though I won't be running it any time soon. There's no reason for 99% of that crap to be in the filesystem, it should be implemented in userspace and then be portable to all filesystems.

I knew about ReiserFS4, and it's amazing that the code is actually "done", but the semantics and the capabilities provided by it, need to be both supported and fully utilized by the userspace tools to make a difference, and given the "unix traditional way of doing things", that may prove more difficult in some sense than even writing the lower-level FS code in the first place. I wasn't aware that BeOS dropped BeFS, but you could well be correct. I only tried out the freely-available "test" version of 4.0 or 5.0.

 

[Nothinman]

But BeOS still gave the general OS community some new features to contemplate adding to other "surviving" OSes. So in that sense, the "experiment" was still worth it, and the effort not entirely wasted.

Not entirely, but don't you think we would have gotten more features if they had decided to work on Linux instead of producing their own closed OS?

I haven't actually dug through the codebase since XFree86 3.1, so I can't really say. But back then, it was indeed a bit messy.

I'm sure it is messy, it's one of the most complicated parts of unix and it's designed to run on every type of unix there is. But I'm sure it's still easier to break out the modules into their own packages than it would have been to rewrite them from scratch.

I was just saying, that even though it "died", it made some contributions to the general OS community, so in that sense, it was still "worth it" for it to exist. (Kind of brings a parallel analog to issues of human abortion, in a way. Should those who have been born into the "hard life", simply be given up on?)

There's a big difference between a "hard life" and only living for 3 years, and life is hardly comparable to software.

Well, NT was and probably still is ahead of Linux in terms of fine-grained locking and pre-emption, but... I'm not sure how much weight your argument has, because Linux seems to have major portions re-written reasonably often (every two years?), and sometimes, throwing out the entire codebase and starting fresh is the best path to take. (I think Linux SCSI support, and things like the Tulip NIC driver, have had that happen to them at some point in the past, haven't they?)

I don't doubt that sometimes it is easier, but look at the IDE layer. They tried throwing it out and rewriting it and it wasn't possible because noone knows all of the quirks and workarounds implemented in all of the controller drivers and noone will ever know them all because some required a NDA for the specs, Hedrick signed it but I'm not sure how many other kernel devs are willing to sign it even if they were made an offer. But again, the kernel devs have never thrown out the entire kernel and started over.

I knew about ReiserFS4, and it's amazing that the code is actually "done", but the semantics and the capabilities provided by it, need to be both supported and fully utilized by the userspace tools to make a difference, and given the "unix traditional way of doing things", that may prove more difficult in some sense than even writing the lower-level FS code in the first place

I'm sure it's far from "done", hell I still don't trust reiser3. I know people were working on userspace tools to do what BeFS did, how far along they are I have no idea because I'm not interested in them. Technically they probably could get by with just base xattr support which is in most of the Linux filesystems, the added features of reiser4 would just give them more leeway.

 

[Netopia]

Ok.... people don't have their BeOS History down. BeOS ROCKED and should have been a player. In (1994 I believe) they demo'ed a 7 CPU machine at Mac Expo. At that point, Win95 wasn't even out (pre-emptive multitasking) and nothing in the Mac world was even close... and wasn't for YEARS until OS-X, and even then Apple had to borrow from BSD to get a good OS. On the PC side (after Apple saw them as a threat and wouldn't give them any info about thier machines any longer) BeOS did a great job on the PC. The problem was that Microsoft illegally killed Be by telling PC builders that if they installed BeOS on machines (even as a second/dual boot option) that MS wouldn't license Windows to them any longer.... that kept Be from becoming what is could have been...not that they could't get a product out, but that MS wouldn't LET them! In the end, Be sold off all their software to (IIRC) Palm and a lot of it is used in Pilots today. Be sued MS and MS quickly gave them $23,250,000.00 to settle out of court. It's a shame, because Be was fast, worked well, was secure, had a posix CLI.... oh, and by the way... I played Quake (and maybe Q2 IIRC) on BeOS and there would have been PLENTY of OpenGL support had they not been killed by MSGoliath.

Joe

 

[Megatomic]

Yeah, I played with BeOS some back in the day and it was slick. Too bad it's gone.

 

[VirtualLarry]

Originally posted by: Nothinman

But BeOS still gave the general OS community some new features to contemplate adding to other "surviving" OSes. So in that sense, the "experiment" was still worth it, and the effort not entirely wasted.

Not entirely, but don't you think we would have gotten more features if they had decided to work on Linux instead of producing their own closed OS?

Honestly? No.

I think that the thing most conducive to creativity, is a blank sheet of paper. Working from a half-finished drawing that someone else started, isn't nearly as creatively liberating. The thing that isn't being said here, and will also I'm sure be a point of debate between programmers/designers for many years to come, is that software is not just mechnical assembly of parts, it is art. But it is more than just art, because it is also (generally) meant to be functional. So it's a hybrid, and the facets of each aspect are sometimes hard to quantify. Perhaps that's also why computing "software metrics" are so difficult.

Originally posted by: Nothinman

I haven't actually dug through the codebase since XFree86 3.1, so I can't really say. But back then, it was indeed a bit messy.

I'm sure it is messy, it's one of the most complicated parts of unix and it's designed to run on every type of unix there is. But I'm sure it's still easier to break out the modules into their own packages than it would have been to rewrite them from scratch.

Considering that the project that I was working on at the time, was investigating the feasability of doing exactly that - well, let me say, "easier said than done". If you're going to have to re-write for modularity, then why not take that opportunity to (at least partially) re-architect as well. Most programmers probably think that way, it just seems to be a logical course of action. Once you're at that point, you might simply choose to re-implement from the ground-up instead.

Originally posted by: Nothinman
There's a big difference between a "hard life" and only living for 3 years, and life is hardly comparable to software.

True, but by the same token, software-development is a far more "organic" process than most would like to admit. (Especially project managers that like to set various generally unrealistic deadlines.)

Originally posted by: Nothinman
I don't doubt that sometimes it is easier, but look at the IDE layer. They tried throwing it out and rewriting it and it wasn't possible because noone knows all of the quirks and workarounds implemented in all of the controller drivers and noone will ever know them all because some required a NDA for the specs, Hedrick signed it but I'm not sure how many other kernel devs are willing to sign it even if they were made an offer. But again, the kernel devs have never thrown out the entire kernel and started over.

That's a very good point, and an excellent example of a body of code that became an intrinsic repository of accumulated applied knowledge, and due to a lack of corresponding documentation, should likely never be re-written, only ever merely adapted to new tasks/purposes. (I have a gut feeling that much of Windows' is like this - and why the 16-bit asm core of Win3.x, was mostly ported wholesale to form the core of Win9x, using the 'Win16Mutex' to wrap the new Win32 layers around the original Win16 core.) There's a good article on a web site by one of the original Netscape devs, about why the ground-up re-write planned for Netscape 5.x was possibly the worst thing that ever happened to them as a company and codebase. I've also seen references in Bugzilla, talking about Mozilla/Firefox's FTP support, and the fact that the codebase that performed the same functions in Netscape 4.x was highly-evolved, containing a lot of "embedded knowledge" about various nuances of all of the different FTP servers out there.

That's actually also an interesting way to look at software development and system-design, and my personal view is that if one is ever able to create a "machine AI", it cannot be "built", but rather, it will have to be "evolved". There's just no good way to make it happen otherwise. Much like those "highly evolved" pieces of code.

Originally posted by: Nothinman
I know people were working on userspace tools to do what BeFS did, how far along they are I have no idea because I'm not interested in them. Technically they probably could get by with just base xattr support which is in most of the Linux filesystems, the added features of reiser4 would just give them more leeway.

Interesting info, thanks. I'll have to look up "xattr support", never heard of it.

 

[VirtualLarry]

Originally posted by: Netopia
The problem was that Microsoft illegally killed Be by telling PC builders that if they installed BeOS on machines (even as a second/dual boot option) that MS wouldn't license Windows to them any longer.... that kept Be from becoming what is could have been...not that they could't get a product out, but that MS wouldn't LET them! In the end, Be sold off all their software to (IIRC) Palm and a lot of it is used in Pilots today. Be sued MS and MS quickly gave them $23,250,000.00 to settle out of court. It's a shame, because Be was fast, worked well, was secure, had a posix CLI.... oh, and by the way... I played Quake (and maybe Q2 IIRC) on BeOS and there would have been PLENTY of OpenGL support had they not been killed by MSGoliath.
Joe

It was a sad story, and truly a shining example of how MS's underhanded and un-ethical, if not down-right illegal business practices hamper competition and retard the growth of true innovation in the industry, all so that MS can continue to rake in their hoards of cash, because there is no effective competition.

If only the current crop of under-$300 Wal-Mart boxen, that ship without Windows, were around back then, and would have provided a market for BeOS to sell into, and become a competitive platform. Be, Inc. were truely ahead of their time.

 

[Hexydes]

I think it is important at this time that I point out that it is not, never has been, and never will be, a goal of SkyOS to run, or even emulate, Windows-based applications.

SkyOS has many advantages over Windows. You can find one of them on the main news page currently: Live Query Searching. SkyOS has a file system called "SkyFS", which is based on Haiku's OpenBFS (which itself is based on the BeOS file system, BFS). You can search your entire hard drive in milliseconds and filter results based on the meta data inside of the files. You can select various items from the results and add new attributes (such as "Album: Album Name"). You can then go into the viewer and arrange your files and folders according to these attributes, and even filter individual folders so that they only show certain items (for instance: only show music files with the album name "Transistor").

There are lots of other technical advantages we have, and many more to come in the near future.

If you have other questions, I would be glad to answer them. I do not frequent Anandtech often, but I will do my best to check this thread.

Regards,

Kelly Rush
Business Relations, SkyOS

 

[Megatomic]

Welcome to AT Hexydes, thanks for joining and adding to this thread.

 

[Netopia]

Originally posted by: Hexydes
I think it is important at this time that I point out that it is not, never has been, and never will be, a goal of SkyOS to run, or even emulate, Windows-based applications.

SkyOS has many advantages over Windows. You can find one of them on the main news page currently: Live Query Searching. SkyOS has a file system called "SkyFS", which is based on Haiku's OpenBFS (which itself is based on the BeOS file system, BFS). You can search your entire hard drive in milliseconds and filter results based on the meta data inside of the files. You can select various items from the results and add new attributes (such as "Album: Album Name"). You can then go into the viewer and arrange your files and folders according to these attributes, and even filter individual folders so that they only show certain items (for instance: only show music files with the album name "Transistor").

There are lots of other technical advantages we have, and many more to come in the near future.

If you have other questions, I would be glad to answer them. I do not frequent Anandtech often, but I will do my best to check this thread.

Regards,

Kelly Rush
Business Relations, SkyOS

Wow... very cool that you came by. I would like to ask a question based on curiosity alone. How did you come to knowledge about this thread? Google search?

Also... what is the market you see for SkyOS? Is it like Be, that it isn't a replacement for Windows but a side-by-side OS or do you see it being able to catch enough marketshare to stand alone?

Thanks,

Joe

 

[Hexydes]

Wow... very cool that you came by. I would like to ask a question based on curiosity alone. How did you come to knowledge about this thread? Google search?

Google search.

Also... what is the market you see for SkyOS? Is it like Be, that it isn't a replacement for Windows but a side-by-side OS or do you see it being able to catch enough marketshare to stand alone?

Definitely standalone. SkyOS isn't being marketed as a "Media" OS, as BeOS was in the early days of it's x86 existence. SkyOS can stand on it's own.

 

[Netopia]

I just went back and looked around your site some more and took the tour. Decided that I'd download and test your newest Beta release... but do I see it right that I have to pay $30 to beta test your OS? I realize that you folks need a revenue stream, but with umpteen gazillion distros of Linux out there and the BSD's to be had... are you really going to be able to get both the number of testers and MORE IMPORTANTLY the coverage by word of mouth that you would like while charging people to test?

Joe

PS It's only "Q" (with brackets) on this board for quoting. Different from phpBB and others.

 

[Hexydes]

...are you really going to be able to get both the number of testers and MORE IMPORTANTLY the coverage by word of mouth that you would like while charging people to test?

Currently we have about 400 beta testers. We get between 5 and 10 new signups a week. As far as word of mouth, we are posted on a number of websites, including OSNews.com regularly, and occasionally Slashdot.

Thanks for the heads-up on quotes.

 

[Sunner]

I agree that free beta testing would be nice, while I don't mind paying for software I may or may not use if I feel that it's for a good cause(for example, I buy every release of OpenBSD, I've only installed 2.9, 3.3, and 3.6), I'm not sure I'd wanna spend money on a beta.

Why not a free beta and a pay version of the finnished product?
Won't speak for anyone else, but if I liked the beta, I'd buy the final.

 

[Netopia]

Originally posted by: Sunner
I agree that free beta testing would be nice, while I don't mind paying for software I may or may not use if I feel that it's for a good cause(for example, I buy every release of OpenBSD, I've only installed 2.9, 3.3, and 3.6), I'm not sure I'd wanna spend money on a beta.

Why not a free beta and a pay version of the finnished product?
Won't speak for anyone else, but if I liked the beta, I'd buy the final.

You could be speaking for me, because I feel much the same.

Joe

 

[n0cmonkey]

Originally posted by: Sunner
I agree that free beta testing would be nice, while I don't mind paying for software I may or may not use if I feel that it's for a good cause(for example, I buy every release of OpenBSD, I've only installed 2.9, 3.3, and 3.6), I'm not sure I'd wanna spend money on a beta.

Why not a free beta and a pay version of the finnished product?
Won't speak for anyone else, but if I liked the beta, I'd buy the final.

Doesn't Microsoft charge for some beta testing? They need the money now, people seem to be willing, let them test.

I think I won't be buying any future openBSD releases. I'll just give them money, I don't use the cds anyhow. I will miss the stickers though.

 

[Netopia]

Maybe you could just get them to send you a few sheets of stickers and then paper your neighborhood!

Joe

 

[VirtualLarry]

Originally posted by: Sunner
Why not a free beta and a pay version of the finnished product?
Won't speak for anyone else, but if I liked the beta, I'd buy the final.

I'm sure that continued "donations" to be able to access beta versions, keep the programmers in pizza+beer money (or not, but I hope that they might). It's not too much difference than the business model of Sveasoft, although since the codebase that they hack on is mostly GPL'ed stuff, the fact they they charge "access fees" for beta versions is a bit more contentious.

Personally, I think it's great that some company wants to go against both the commercial MS OS market juggernaut, as well as the Linux/free-unix dogmatic "religious fanatics". Plus, I hear that Sony might be looking for a new OS for their new Cell-powered devices, hmm.

 

[n0cmonkey]

Originally posted by: Netopia
Maybe you could just get them to send you a few sheets of stickers and then paper your neighborhood!

Joe

I'd spend a ridiculous amount on stickers if I could. $50 for like 10 sheets of stickers... I'd be all over it.

 

[OffTopic1]

Originally posted by: n0cmonkey

Originally posted by: Netopia
Maybe you could just get them to send you a few sheets of stickers and then paper your neighborhood!

Joe

I'd spend a ridiculous amount on stickers if I could. $50 for like 10 sheets of stickers... I'd be all over it.

Yeah! It would be funny if you have a sticker on your car that say "BSD Ricer".

 

[n0cmonkey]

Originally posted by: OffTopic

Originally posted by: n0cmonkey

Originally posted by: Netopia
Maybe you could just get them to send you a few sheets of stickers and then paper your neighborhood!

Joe

I'd spend a ridiculous amount on stickers if I could. $50 for like 10 sheets of stickers... I'd be all over it.

Yeah! It would be funny if you have a sticker on your car that say "BSD Ricer".

I put them on my sparc machines to add mhz.

 

[user1234]

Originally posted by: OffTopic

No experience with SkyOS, however I like to know more. Is there a hardware compatible list available?

It look to me like another *nix variant like QNX/BEOS that isn't going to go much beyond beta.

what's up with "*nix" thing ? is your keyboard stuck, or are you some kind of moron trying to censor his own post ?

 

[n0cmonkey]

Originally posted by: user1234

Originally posted by: OffTopic

No experience with SkyOS, however I like to know more. Is there a hardware compatible list available?

It look to me like another *nix variant like QNX/BEOS that isn't going to go much beyond beta.

what's up with "*nix" thing ? is your keyboard stuck, or are you some kind of moron trying to censor his own post ?

Is it Unix? Is it unix? Is it Linux? Is it BSD? Come up with a universal name that is shorter than unix-like. Please.

 

[user1234]

I believe "unix variant" is just as clear as "*nix variant" used in the original post, minus the retarded wildcard

 

[Netopia]

Originally posted by: user1234
I believe "unix variant" is just as clear as "*nix variant" used in the original post, minus the retarded wildcard

Is linux a "variant" of unix? I thought it was written from the ground up. If I'm correct (and I may not be... but if I am) then it is your use of "unix variant" that is retarded as it would not incorporate the entire linux world even though that is what was being referred to.

Joe

 

[n0cmonkey]

Originally posted by: user1234
I believe "unix variant" is just as clear as "*nix variant" used in the original post, minus the retarded wildcard

"Unix variant" does not fit the requirement of being small.

var·i·ant ( P ) Pronunciation Key (vâr-nt, vr-)
adj.
Having or exhibiting variation; differing.
Tending or liable to vary; variable.
Deviating from a standard, usually by only a slight difference.

If it's unix, it can't deviate from the standard, but Linux isn't unix so it can't be a variant.

Linux does differ in the fact that it isn't unix.

Unix-like is probably the best way to describe it because it specifies that these OSes are _not_ unix. Unix variant gives me the impression that all of these OSes _are_ unix.