First of all, I am very flattered that our recent editorials over at OpenBeOS have received so many responses in the OSNews commenting section.I would like to address some of the issues brought up in the many thought provoking comments, in the order that I read them:
Doomed by C++ Virtue
Some people have opined that BeOS’ adoption of the most popular “object oriented” (yes, I know the Smalltalk people just gasped) language in the world was its doom. They talk about Fragile Base Class as one reason. Being an engineering type, the first response that I have is to say that FBC is an implementation detail of the compiler. There is nothing that *I* am aware of in the C++ spec that requires FBC to occur. And there are certainly ways around it in the compiler writer’s world. Many of them are hard and would take a fair amount of extra work. But they could be done. However, there is another side to this. There is *NO* language that will please everyone (watch for this to be a recurring theme here). C? Old school. Objective C? Too weird. No one knows it. Perl? Lisp? Awk? Forth? KSH? Cobol? If Be had happened to pick one of those, some people would have been happy, others would have hated it. The long and short of it is this – it is what it is. Be chose it. We could not even pretend to write OBOS and use any other language. If you want to argue that writing some other OS is a better idea, well, that is your decision. I would be interested in knowing where you could get a full design that is better.
Buying BeOS from Palm
This whole drama played itself out. A little known fact is that I personally ponied up some money to make it happen. I thought that this would have been a great option. Palm had/has *NO* interest. I know for a fact that many tried. And some are still trying today. Palm has liability in the code – all of the licensed code that is incorporated. They bought those rights, too, I would think. And Palm would have to, in all good conscience, have to pay people to rip that licensed code out. Not to mention the opportunity cost of paying engineers to do something that will never further *their* product line. No, the smart corporate decision from Palm is to put the code in a vault and sit on it. I don’t like it, either, but if you are trying to convince someone of something, you have to put yourself in their place. And their place is far safer here than releasing code.
Future Goals of OBOS (and copying of R5)
This is the one item that boggled my mind the most. We daily have posts and ideas about where to go from here. *EVERYONE* is interested in adding this feature or that option. I have to bat them back to reality. We have a cohesive design today. Implementing just what is already designed will be hard enough. Designing, coding, testing and documenting a dozen peoples’ vision of where the OS should go is more than we can do right now. With a stable and working R1, we would/will be in a far better position to make changes. Glass Elevator is *dedicated* to figuring out what to do with R2+. I personally have a whole notebook of changes that I would like to see made. The team leads, as they and their teams have coded, have grumbled about the way this and that works. Believe me, the biggest challenge that we will have post-R1 is to figure out what *not* to change. No matter what we do, whether we abandon compatability or remain strictly R5, half of the people will not be happy.
As an added thought, a new GUI was mentioned. For all of the moaning and groaning that I hear about the current GUIs in the world, no one seems to have actually *designed* a better approach. I hear a lot about how the desktop is dead, its’ time has come and gone, blah, blah, blah. But I don’t see a better design out there. Not even a Photoshop mockup. Why is that? I don’t think that anyone knows what the next big GUI design will be. If you do, contact me. 😉
Why protect corporate interests
Very simply, corporations write apps. Usually the big, ugly apps that no one else wants to write. Look around the open source community. Compare it to a CompUSA ad. There is a lot of software that OSS people don’t want to write. It isn’t an itch that they want to scratch. Example – accounting packages. Several corporations have come to me about using OBOS. They are very pleased about the licensing and direction that we are going in.
Apps
Come on, guys. 😉 We know we need apps. We know about the whole chicken and the egg thing. This wasn’t new when Be started. This is one reason that we have a goal of binary compatability with R5. If we do this right (and so far we have), everyone can keep Productive and Opera and so on. Furthermore, BU or maybe some rich distro maker can go to these companies and say “hey – this is an active OS with N users – can you port your software?”. Apps is a known issue. Part of the reason that OBOS tries to stay so open and up front is that we want people to get excited about our progress without promising the moon. The more excited people are, the more likely they are to write apps.
Drivers
Again – this is not a surprise to anyone. One thing that I find to be encouraging is that more and more hardware is more and more compatible. Example – video cards – there are a *lot* fewer video card chipsets than there used to be. While I don’t consider this a good thing for the consumer, it is a good thing for us. USB is another good thing. The only solution that anyone has offered, so far, is to use a Linux kernel to use their drivers. Again – this is a solution designed to anger 1/2 of the people.
Releasing under GPL
The way that I see it, GPL takes rights away that the MIT license grants. GPL grants nothing that MIT does not. So, the way that I see it, we could do very much what Perl does, and dual license. It would neither help nor harm anyone. The only reason that I don’t is that I don’t see the point.
New Name
You would be very surprised how hard it is to come up with a *good* new name. We had > 3000 suggestions. Not one was so impressive that we all just said “this is it”. Many were acceptable. Some were too geeky, some were too bland. Some were to Be-like, some were too close to other similar products (Bindows, for example). We are still working away at this.
Dorks in Management
This is one of the biggest advantages of being an open source project. Sure, some OSS projects have management issues, too. But there are no accountants or MBAs who will tell the devs what to release and when with OBOS. That is both good and bad, I understand, but this particular issue is pretty easy to put to bed for OBOS, right now.
Corporate vs Hobbiest
“I am not in this for your revolution, Princess” — Han Solo
The point of OBOS is not to become Be, Inc. Go back and read that again.
We are not Be, Inc. We are here to make an operating system that is what we think is the best way to make a desktop OS. We all came together knowing that Be did a heck of a job on R5 and that continuing in that vein is a good direction. I want an OS that I can use happily. And recommend to my friends and family. I am not in this to form a company and make a billion dollars. I *certainly* would have picked an easier industry. I hope that some Red Hat comes along and makes OBOS a household name. I wouldn’t mind too much if they contributed back, either. But I am not counting on it. I am not quiting my full time, paying job to do this. This is all in my (and all of our) spare time and out of the desires we have to own and run a sweet OS. Being able to say that you built it doesn’t hurt, either.
I would just like to take a moment to thank all of those who encourage the OBOS developers and myself. Reading those emails straightens our backs a little, helps us to raise our heads and march on. For all of those who challenge our worldview, I thank you for keeping us honest. It would be wonderful to live in OBOS-world, where there was no other OS and the entire world was using R5 and awaiting our first release with a baited breath. 😉
Michael Phipps, OpenBeOS Project Leader
It’s gonna take more than Gobe or Opera to attract anyone besides the Be faithful
For me personally ….
* The ability to open my .doc and .mdb files
* A text editor at least as powerful as UltraEdit
* A sound editor at least as powerful as CoolEdit Pro 2.0
* An IM program that’s better than Trillian
* An MP3 player as good as Winamp
* DVD software that’s as good as PowerDVD
* CD burning software as good as Nero
* Ability to play mpeg4/divx/real media/etc
So, uhhhh … good luck – we’ll all be rooting for ya
some of these exist, though not all
* The ability to open my .doc and .mdb files
Gobe 2.0 can read and write Word and Excel files. Excuse my ignorance, but what is .mdb?
* A text editor at least as powerful as UltraEdit
Havent used ultraedit, so I cant say. Pe is a very nice editor however, and is/will be opensourced soon IIRC so work will continue on it.
* A sound editor at least as powerful as CoolEdit Pro 2.0
yeah, that would be nice. maybe some of the pro sound editors will continue their promised ports when obos R1 is finished
* An IM program that’s better than Trillian
that would be great
* An MP3 player as good as Winamp
check out soundplay, it can do stuff that winamp can only dream of
* DVD software that’s as good as PowerDVD
there is some work in this direction, but nothing quite there yet
* CD burning software as good as Nero
same as the DVD issue
* Ability to play mpeg4/divx/real media/etc
divx works, real media sorta does, i expect the others to follow soon from the discussion in the mediakit and translation kit lists
HaveFun!
WhiteRabbit
What’s different between what OBOS is doing and what Be was doing before it switched to internet appliances? “Build the most compelling desktop OS” ? In what time frame? For what audience? To what end? Who is involved and why? What do they expect out of it? Why should I want to be involved?
Personally, I’ve used R4 and R5. Neither one was useful to the point of being a better desktop solution than OS X, Linux, or Windows. The OS sure ran fast. But it wasn’t capable of doing much. All the clever little Be-isms got annoying after a while. Even things like running another instance of an application were often confusing.
Before Be switched to appliances, there was no mass interest in what Be had to offer. What’s different now? All of Be’s competition for the desktop has only gotten better since then.
How long will it take for OBOS to create what Be had at the R5 or Dano milestone? And I mean “fully debugged commercial quality code” not “just checked into CVS”. Is this 2003? 2004? 2005? Be realistic.
How is this OBOS-1 going to compete? The professed audience is the hobbyist. Sort of. Because Red Hat doesn’t pick up hobbyist operating systems, yet there is some desire for a Red Knight to come along and rescue OBOS.
What app vendors want to write apps for desktop Linux? Or for OBOS? Who were these corporations that came to OBOS? What did they want? What did they say? What money would they put up to support OBOS?
OBOS does have a chance to do something wonderful. If there’s enough creativity, focus, and drive to make it happen.
Get OBOS-1 done and shipped. Then let’s see what ingenuity and invention there is for OBOS-2.
Show me something compelling rather than the latest copy-someone-else’s-os-or-apps-or-api’s borg-replication movement.
#m
What if OBOS talked with say the OpenOffice.org or GoBe people to try and see they are interested in working together to ensure compatibility with OBOS?
As a user, I only require two pieces of software to use an OS more than half the time – a mailer and browser. We all know that BeOS currently has no world-class, totally functional offerings for either. I’ve used Mail-It, Postmaster, and BeMail. They all crash occasionally, have so-so filtering, fail to comply with standards to some extent, etc. Same situation width Mozilla, Opera, and NetPositive – they kind of work to some extent. But would you file your taxes electronically with any of them? That would be very brave. So sure, I want all the other apps that Darius mentioned, but I need these two first.
Oh, one more thought about drivers. Check out this old newsletter article by Pierre about why video drivers are so hard:
http://www.beatjapan.org/mirror/www.be.com/aboutbe/benewsletter/Iss…
Then remember he’s only talking about 2D support. So multiply that by at least three for a full OpenGL driver. OpenBeOS _will_ have serious trouble with hardware support if they keep going down their present path, which is a shame.
I would rather call Be a toy OS, Linux a server OS, Mac a lady OS (for it looks great but not necessarily works great), and Windows w/NT kernal a true OS.
what do you think?
So what path do you suggest?
…that alt operating systems should be called troll OSes because there seem to be so many making comments about them.
It doesn’t look great and not necessarily works great.
Bloat OS
Forced on you OS
Constant reboot OS
True OS?? What others are “ghost” OSes?
Because majority of PC users use it?
I wish I had the answers David. It seems like leveraging all the work and code in the Linux community as a base would get you much closer to your goal, much faster. You have a booting kernel with networking and video on day one. But more importantly, those same people will keep adding support for future hardware – you get a huge engineering team for free. And you get employees actually at Intel, NVidia, ATI, and so on writing the drivers, so you’re guaranteed a better product than you could write yourself. To me, the kernel is not the point of the OS, so long as it is reasonably modern and stable (preemptive scheduling, protected memory, etc.). I’d do some research first, but on the surface Linux seems good enough and the advantages are enormous.
Would you say LGPL, BSD, or MIT license to best ensure the success of an operating system?
Also, I don’t disagree with the efforts to write OpenBeOS, but if it is not going to compete on the desktop, what is? Will L4 be the “next Linux” with an actual change at the desktop?
It sounds like you’re staying pragmatic and making all the sensible decisions. Keep up the great work! The world needs a really usable open source OS (and we understand it’ll take a little while).
Well the spec. doesn’t clearly say
– “Though shallt only inherit from classes in constant flux.”
but it doesn’t need to. It’s a simple result of C++ undynamic nature. True, you could write a C++ compiler and linker that doesn’t suffer from these problems but isn’t that huge piece of work and end up with something that looks like C++ (somewhat) but is more like Objective-C or some other “true” object oriented language in nature?
What I think BeOS lacked was an object/component model and this is something I think has to be rectified. Windows has an object model in COM and an even better one in the .NET
ho-ha. KDE has a component model in DCOP/KOM. Gnome has a component model in Bonobo. What these first generation desktop O/C models have in common are they provide poor support for cross language inheritance, smooth data exchange and poor support for fine grained objects.
In this respect MS actually got it right on the money with .NET, you get all these kinds of things for free since you need to target the byte code. Unfortunately the fuX0Rd it up royaly when it came to byte code design. They should have deferred the decisiions about what is the API what is the language to the implementors of that particular .NET language. That is why things as returning multiple results (Dylan, Lua), lists and lambda calculus (LISP family and others) among other things are such a bitch to support in .NET.
If OBOS really want to attract the vast hordes of corporate and open source developers it should provide something technically interesting. Get people curious about OBOS development and the current BeOS C++ API is just plain boring.
You would be very surprised how hard it is to come up with a *good* new name. We had > 3000 suggestions.
> 3001… I propose the name “AfterShock”. The shock was Be folding up, the aftershock OBOS…
People ask, what userbase for a new OS?
My answer.. make it low on bloat and aim it as a multimedia OS for low powered hardware. If you’ve ever tried running a stock distro of Linux 2.4.x on a Cyrix M2 or a Pentium MMX you’ll know what I mean. I don’t like OS’ that choose me to abandon usable desktops just to get an acceptable speed.
The original BeOS ran well on slow machines.. a legacy worth maintaining. And BTW the Interface was very good too as far as I recall.. it looked good, was easy to use and rendered very quickly
my guess is that it will be say 2 years till r1 is out. thats august of 2004 and another 6-12 months of subtle improvements until the real r5 close is out.
Ladies and Gentlemen!
I think first of all that we should congratulate the OBOS team on huge progress rather than to try to kill it. These guys have reproduced huge parts of R5 with even BETTER code (according to old Be engineers) than Be had. So when all this talk about commercial help drops in, I think someone has an opinion rather than any facts.
Target group, how to market, how to succeed is all very important and extremely difficult questions which not just OBOS is tampering with but all operating systems which compete with Microsoft. But I’d love to help out with some facts…
1. BeOS/OBOS is without any doubt the fastest and most user friendly OS on the market (explains no 1 position on MS threatlist).
2. Commercial say no at the moment, 3000 c++ devs who are still growing says YES!! Because they can make their apps faster, better, more responsive and more smooth in Be than in any other operative system they know of and make installation a laugh afterwards (Goodbye all *nix).
3. Entire Asia are looking for alternatives to Windows and let’s face it, one part will like Linux (that 0.5 % as always) and the others? Yepp, they are likely to BE in a while… And if you’re talking commercial companies saying no to a 1 billion user market, then I’d be seriously impressed…..
Still not convinced OBOS will get along well? I think no one is, but the point which is worth noting is that there aren’t anything better for Joe User on the desktop market today, and looking at the progress of others, we’re talking 5 years before they reach Be technology, that’d mean OBOS can do little to none before they loose the edge on what they have…
Think about that…..
P.s Apps is a problem, but Be’ish apps have a much more simple way of upgrading themselves than other systems… For instance Instant Messaging which you’ll soon find a lot better than Trillian (2 months away?), don’t worry about it. D.s
Michael, just go for it, and don’t care that much what other people think.
You have to build you’re own OS, not using any other or so, why, what’s the use of making ‘copies’ of others?
I know it’s going to take much more time, but it also means that there’s the possibility you’re going to have a better (and not an equal) OS at the end to show for it.
Torvalds didn’t start with a lot of code from someone else, he started from scratch.
Try to incorporate the best idea’s (not code) from other OS’es and see what you can do with it.
Darius has strange comments, I once had a big wishlist of BeOS features witch I wanted in my windows, I know now that’s never going to happen. So why should beos HAVE to support M$????
Use the BeOS momentum and make a good (maybe the best) OS.
Good luck
Roland
Who cares about taking over the world and replacing Windows ??? Why should OBOS attract the masses ??? Why not developp OBOS just for the fun, to learn OS construction by doing (and with an excellent architectural model, at that) ?
I find the performances and architecture of BeOs and OBOS utterly interesting and don’t give a shit about OBOS being largely used ! If OBOS remains “only” a geek OS, let it Be so ! I don’t think there is place for a new big player in Oses anyway : Windows rules on the desktop (even Rasterman, of Enlightenment’s fame, acknowledged it recently), Linux offers a good alternative on servers and I dont think any new player could gather as many drivers as Windows and Linux already have.
But again, I don’t care, it’s much more interesting to work on OBOS than going back to the 70s and re-creating a big old monolithic UNIX !
* The ability to open my .doc and .mdb files
Productive 2.0 handled .doc and .xls files quite well, at least for my needs. .mdb files (Microsoft Access) don’t really have an equivalent in BeOS, as there’s no low-end database software. DataWing provided a database API, and there are several SQL databases that run on BeOS, such as PostgreSQL and MySQL. gobeProductive 3.0 could very well show up on OpenBeOS seeing as that the program’s mostly written in the BeAPI, and wrapper classes were created to allow compatibility with Windows and Linux.
* A text editor at least as powerful as UltraEdit
Again, as mentioned, Pe is an excellent editor that’s already on BeOS, and by the same guy that did Pepper (which runs on MacOS, Linux, and Windows(?).
* A sound editor at least as powerful as CoolEdit Pro 2.0
SampleStudio is an audio editing app in the works presently that will at least handle stereo audio, with multi-chanel audio coming soon, IIRC.
* An IM program that’s better than Trillian
I personlly use BeAIM for my IMing needs. Just as good as Adium, but without some of Adium’s nice UI elements. There’s also the mutli-protocol Jabber for BeOS.
* An MP3 player as good as Winamp
SoundPlay beats all other audio players hands-down. It’s phenomenal. And, if you’re hung up on Winamp, it can even use Winamp skins.
* DVD software that’s as good as PowerDVD
That would be nice. The VideoLAN client is making excellent progress, but it’s not quite there yet.
* CD burning software as good as Nero
The BeOS architecture allows for some nifty things to be down with burning apps, as BurnIT Now demonstrated. It was a front-end to cdrecord, but I think development on it stopped. CD Manager was another app that did it all natively. I could test it properly, as my burner has issues in non-MS operating systems.
* Ability to play mpeg4/divx/real media/etc
Real media files were behind, because the RealPlayer that was included with BeOS 5 Pro was sub-par. However, I believe progress is being made in the way of DivX abilities.
Also, the same guys that are doing SampleStudio are making Refraction, a cool image editor in beta. It looks very promising.
— Rob
till obos is sold at http://www.walmart.com ?
BeOS is old technology. C++ and microkernels are from the 80s. There’s new and better technology. Take a look at the Plan9, Eros OS and Exokernel, SML (Fox project), OCaml and Cyclone.
Linux doesn’t stand still. The Kernel gets better for the desktop. X is getting better, it already has anti-aliasing and Open GL support. Graphics support in the kernel (new framebuffer API) is getting improved. There are projects to improve graphics support in the kernel and userspace: KGI and GGI. There’s a project to create a new windowing and GUI toolkit: Fresco.
There’s nothing fundamentally wrong with Unix, that makes it unsuitable for the Desktop. Mac OSX proved that.
Is R5 binary compatibility really THAT important? For what? Opera? – C’mon… it’s just Opera 3.6 – who cares? Better work on embedding BeZilla into a Net+ like app.
Gobe? – When OBOS is “finished”, the wold has Gobe 4 and *we* are so happy with *our* Gobe 2. Remember that Gobe is a commercial product? How can you “conquer” the desktop with an office app that’s no longer for sale? Encouraging piracy? I doubt that would help OBOS. The port of Abiword is allready pretty good. Sum-It (http://sourceforge.net/projects/sum-it ) a an open source spreadsheat app… it’s not very advanced and it’s outdated, but if a handfull of serious developers took care of it, it could become quite good.
There are not so many must-have closed source apps for BeOS that are no longer supported. The still supported closed source apps will surely be recompiled for OBOS (Refraction as an example).
The next question is important to me and I hope that somebody can answer it:
Is it possible to write a binary-compatibility layer for the various BeOSes (OBOS, Blue Eyed OS,…)? All those OSes will be BeOS compatible on source level, but not bin-compatible with each other. I’d like to see some kind of application or _server that understands a binary format that the developers of the BeOSes have agreed on. The APIs will be the same everywhere anyway, so wrapping a binary to each OS shouldn’t be so compute intensive.
I hope you understand what I’m trying to say. My English is not the best.
BerretOS… EiffelOS… BastilleOS… GarlicOS… MoulesEtFritesOS… ChampagneOS… MyChildrenNeedWineOS…
SurrenderOS!
I think it’s a good to have projects like OpenBeOS rather than just throw all the eggs in the Linux basket. Even if the project does not rival Windows, it is useful in terms of research alone. Working outside the box may bring some new ideas (both technical and UI/concept related) to the table and at worst make a new set of mistakes for others to learn from. With research, failure can be every bit as educational as success. I wish the team luck in hopefully bringing some new ideas to the table.
I think the project OBOS has made is amazing. I want to see it done soon but i don’t want to see them taking short cuts (using linux bits) . I want to see beos and it be a quality OS, it does not need mass use. If it’s small and progressing thats fine. It’s about quality not bizzilion features and running on everything and trying to do everything. Thats how you make an OS a POS, this is the problem of linux. It has no path, it tries to do so many things and looses nice clean uniform stable path. OBOS stands to be a nice clean OS just like Beos, It will have more appeal to most than linux if it works like Beos, Be’s flop was no one knew of Beos. I agree with the MIT license, the GPL is evil and not what you want if you want to see corparate support, corparate support in apps and such is what will make you live. If the OBOS team can pull off the R1 and continue they will pass linux in use, linux will never be a mac or windows compitition because it is on the wrong path. it has no path. Keap the work up guys and keap talking to the community and letting everyone knowing how it is going and what you can use. Also any word on when you will have a new name? if so will there then be shirts, I think it is time for my first geeky shirt.
Yes, writing drivers can be a real pain in the ass – been there, done that. But what other way is there? Source compatibility with XFree86?
BeOS is old technology. C++ and microkernels are from the 80s. There’s new and better technology. Take a look at the Plan9, Eros OS and Exokernel, SML (Fox project), OCaml and Cyclone.
Yeah so be it… but there can only Be one….
And as long as all I can see out there is BloatOS (Windows, Linux etc.)… OBOS will Be that one…
I would totally agree that OBOS shouldn’t shortcut and use bits borrowed off Linux. My personal opinion is that Linux has become a fair kernel surrounded by thousands of apps which only do part of a job. BeOS was designed to reduce the clutter and bloat and instead intelligently approach the OS market through a sensible and extensible API and good coding methodology.
If all OBOS-1 achieves is binary compatibility with R5 it will do great things for the development community. I look forward to the results of the OBOS team.
OBOS is definetely the way to go, but looks like it is going to take more than a year (aprox. september 2003?) till there is an alpha Release1.
Remember the McEvan’s Amiga interview? “Stay Alive!, we will find you”. BeOSR5 has to stay alive with efforts like the YellowTab team are doing. I liked to see that BeUnited was “behind”, that is till I found that BeUnited had no corporate interest, now I’m not very sure of what exactly are they doing at BeUnited. How is BeUnited developing and promoting Open Source BeOS Standards? How do they back YellowTab or OBOS?
It is much clearer to me the YellowTab roadmap with their HOME EDITION, DEVELOPER EDITION and DELUXE EDITION, due in November 2002 and January 2003. That’s about the only BeOS development at reach of hand for the end user, the rest is still vapor, not real yet.
The OBOS developers are doiing a great job and they should not borrow stuff from linux / *nix or whatever , as long as they can build there own code, which looks very nice when i read the current progress / comments …
no offence at linux , but there are too many projects that didn’t work out , are to slow , or still are , and will be forever in progress …. linux is evoluting, but that’s sometimes very irritating to , and there are too many different groups that don’t listen to each other …
keep the good work goiing OBOS team !
Gobe 2.0 can read and write Word and Excel files. Excuse my ignorance, but what is .mdb?
Gobe’s MS format support for version 2.0 is outdated and old, plus it couldn’t open 50% of the documents I throw at it (mostly .xls) and 25% couldn’t be opened perfectly.
* A text editor at least as powerful as UltraEdit
You obviously haven’t tried EMACS. A port to OBOS should be easy, since EMACS runs on POSIX systems. Just add a UI, and it would be fine.
* An IM program that’s better than Trillian
That would be extremely nice. Right now I’m using Trillian cause I have friends on both ICQ and MSN. On Windows, I’m using Gaim, going to move to Kopete once it is more stable.
* An MP3 player as good as Winamp
The only way I could see SoundPlay being inferior than Winamp is that SoundPlay doesn’t skin. Oh wait, sorry, it does skin.
* DVD software that’s as good as PowerDVD
That would be hard to be done by any OSS project. Thank the Congress for the support they gave us. Oh wait, when they were passing the DMCA, they were all tuned into the Bill and Monica soap opera.
* CD burning software as good as Nero
That would be REALLY nice, hope someone sees to this.
* Ability to play mpeg4/divx/real media/etc
Except for DivX, the others are impossible. MPEG 4 for example is hounded by royalties that an OSS project could never provide.
How long will it take for OBOS to create what Be had at the R5 or Dano milestone? And I mean “fully debugged commercial quality code” not “just checked into CVS”. Is this 2003? 2004? 2005? Be realistic.
Nobody could tell. My guess is that by late 2003 or early 2004 they would have “just checked into CVS” – if they are using CVS anyway. “fully debugged commercial quality code” would come out later after that, but it really depends on the quality of the “just checked into CVS” code.
Besides, at least it would be better than Microsoft and Apple – “Oh look, it compiles, let’s ship it” (I’m kidding, I’m kidding).
Bloat OS
you talking about Windows XP or Mac OS X? Because a normal install of OS X takes 1.5gb of HDD space, while XP takes 1.0gb.
Forced on you OS
Obviously, you aren’t forced to use this OS. I’m forced to use Windows right now because of my family than Microsoft.
Constant reboot OS
You are obviously outdated. I never rebooted Windows in a long time (sure I shut it off every night, but it is for safety sake, not because the OS can’t take it).
It seems like leveraging all the work and code in the Linux community as a base would get you much closer to your goal, much faster. You have a booting kernel with networking and video on day one. But more importantly, those same people will keep adding support for future hardware – you get a huge engineering team for free. And you get employees actually at Intel, NVidia, ATI, and so on writing the drivers, so you’re guaranteed a better product than you could write yourself. To me, the kernel is not the point of the OS, so long as it is reasonably modern and stable (preemptive scheduling, protected memory, etc.).
Well, there is BlueEyed OS that is leveraging Linux and XFree. There is Leonardo that is leveraging Linux kernel. I don’t see why OBOS must reinvent the whell. Besides, for BeOS functions, OBOS would have to change the kernel significantly. Plus, if it wants it to be completely binary compatible with R5, I don’t think XFree is the right way. But without XFree, there goes your drivers from NVidia and ATI. Plus, Intel couldn’t care less about Linux (or any other Linux-based OS for all that matters) on the desktop, they support Linux for one purpose only – Linux is one of the dominant forces in the server market.
Would you say LGPL, BSD, or MIT license to best ensure the success of an operating system?
BSD and MIT is relatively the same, except BSD has that advertising clause. If I was to submit my code, I would do it under the BSD license.
Will L4 be the “next Linux” with an actual change at the desktop?
IIRC, L4 is to replace MACH and microkernels for nanokernels.
Windows has an object model in COM and an even better one in the .NET ho-ha.
.NET doesn’t actually have an object model. For example, Mono uses CORBA/Bonobo, instead of DCOM.
My answer.. make it low on bloat and aim it as a multimedia OS for low powered hardware.
I don’t think the success of OBOS lies there. After all, newer, faster PCs are always coming out, and they becoming more and more dirt cheap.
1. BeOS/OBOS is without any doubt the fastest and most user friendly OS on the market (explains no 1 position on MS threatlist).
For the toughest competitor, Microsoft didn’t have any Competition Manager, like they do for Linux. And plus, having not mentioned anywhere once about Be or Be’s products in any way except in the court… is kinda hard to believe since Microsoft always tried to put down its competitors.
But sure, BeOS is fast. It had a cool FS. In Dano, it had a wonderful graphics server. But really, does it even matter to normal users? If they can’t run their apps, nor have anything that rivals the functionality of their apps from Windows on Be OS, why should they move? What BeOS needed and what OBOS needs is a killer app. Like spreadsheets, Internet, etc. that sold so many computers.
3. Entire Asia are looking for alternatives to Windows and let’s face it, one part will like Linux (that 0.5 % as always) and the others?
True, but the emerging markets are growing much faster than OBOS. Plus, you notice in regions where the Linux market is growing very fast, like in China, you would see there are Linux companies behind it. There aren’t any distributors of OBOS.
Face it. In this world, nobody cares about technical merits. Why do you think Windows won over OS/2 and Macintosh? Both Apple and IBM where in a position much better than Microsoft. Frankly because they don’t care who’s best, but who allows them to do their work.
I once had a big wishlist of BeOS features witch I wanted in my windows, I know now that’s never going to happen. So why should beos HAVE to support M$????
Most of the features I liked in Be OS is going to be available in future releases of Windows. Big deal.
There are projects to improve graphics support in the kernel and userspace: KGI and GGI. There’s a project to create a new windowing and GUI toolkit: Fresco.
KGI and GGI success will very much ensure the success of Fresco. but however, Fresco is no where near 1.0. It’s API is still incomplete, and it seems nobody is working on it. If there were more developers, it would be much better.
Anyway, about X, I don’t think little things like AA and OpenGL would fix things. AA only works with apps that write support for it, via XFT. But XFT’s extensions aren’t standard. Also, the OpenGL support is laughable. Also, it would be very hard, I mean very very hard, to fix the inherited problems of X: fragmentation of toolkits and APIs.
Is it possible to write a binary-compatibility layer for the various BeOSes (OBOS, Blue Eyed OS,…)?
I don’t think this is a good idea because all of this Be OS clones don’t have the same goal and direction after they have manage to have some support of the Be OS API.
not to be picky but XP is 1.36 Gig not 1 gig, but i suppose you could axe enough stuff quick enough to get it there.
I agree with the person that was unconcerned about the availability of apps, saying that what he mostly needed was a browser and mail client, except in my case I would say a browser and simple notepad since I use web-based mail. My recomendation is that something like the QNX demo disk would be the best choice for a first public release, something easy to download, easy to get into the hands of others, and undoubtedly the kind of distro you could complete the soonest. Make sure every pc user gets one as an emergency way to get on the internet in case their usual OS goes down. MAC was resurrected among home users by packaging it expressly as a way to get on the internet, the strategy works. Some people need little more than email from the web, so there have been a couple attempts at little email appliances. If a free email service was incorporated into the demo disk it might help get that little part of the market used to using the new OS too; for some of these people it would even be their first OS. Worrying about media players at this stage will only muck things up. People have tv’s and stereos for that. If you market it as a painless way to get on the net, you’ll get your user base, and all the other apps will follow.
And to the people that complain that the tech or programming language is ‘old’, are you on Linux as you post that???
And the name someone suggested, “AfterShock”, has a point. A name that sounds powerful or new, or suggests that the user is an able person for using the OS, or that the OS supercedes what came before it, that kind of psychological craftiness can’t hurt. “OBOS” just plain doesn’t sound serious. Makes you think of an Oboe. Be had this name problem too. They spelled it like Geos, yet expected you to pronounce it “Bee Oh Ess”, which when spoken sounds like you’re saying “B.O.S.”, and if you instead said “Be Operating System” it sounded like you were saying “B” Operating system, suggesting that it is a second choice or second rate next to the primary operating system. Just plain awkward, and not very well thought out.
“Take a look at the Plan9, Eros OS and Exokernel, SML (Fox project), OCaml and Cyclone.”
And exactly how many people have heard of any of these, are using any of these, and is the list of apps that would cause OBOS to collapse if not implemented tommorow not also true for these. Ok O’Caml looks interesting!
Give me a break, right now BeOS is perfectly useable to some degree by quite a few people, but most of us also maintain another cpu or partition or HD for the other OSes of choice. I can live with that indefinitely so all my apps are covered one way or another. If you absolutely only want one cpu, then BeOS will not be for you for a few years, nor will Plan9 or the others mentioned. If BeOS wasn’t around, I would still be switching between Windows & something else since I really am kinda of tired of MS control of my PC & life & w2k is the end of the road for me, not even sp3 with hidden updates. W2K doesn’t even install on my Athlon BeOS box no matter how hard I try to get it in, (the mobo is shit).
And isn’t Windows, Linux & MacOSX like really old too with some layers of recent eye candy on top. I for one like that the BeOS iso is tiny, the PE download is 40-50MB or so. Win2k eats a Gig of HD & I am clueless about what 99.9% of that does except some of the .exe files. Same would be true for Linux & OSX.
Remember BeOS PE I think had a million or so downloads, I don’t expect OBOS to conquer the world, but 1% is a pretty reasonable target.
The dev community didn’t drop dead after the collapse either, & I won’t be surprised to see a few x devs come back, & plenty of new developers. The original R5 made alot of the cross platform stuff painfully difficult, alot of those pts are being addressed so those barriers will be reduced. I am hoping that QT3 will get ported when possible.
Perhaps the biggest issue for the future of the larger apps is the multi threading issue.
On the issues of drivers. For Windows the HW company just passes the cost of development which can be a man yr or more per device to the customer. For a Linux or non Windows driver the cost goes up as the no of users is so much smaller. There is a universal driver platform UDL IIRC that could help in the long term to make writing drivers more portable.
MP & OBOS are doing exactly the right thing, as far as replicating R5 for now, MIT license, bin compatibility etc.
I can only wish them the best of luck.
I’m impressed that everyone finds the time to develop OBOS and how well they are working together. They have my deepest respect and admiration.
Keep up the good (if not amazing) work…
not to be picky but XP is 1.36 Gig not 1 gig, but i suppose you could axe enough stuff quick enough to get it there.
You would also notice a lot of my numbers are nicely rounded up. :p. (Anyway, strangely enough, not that it matters anyway, my HP OEM install takes 1.32GB of HDD space, not including all the updates)
Remember BeOS PE I think had a million or so downloads, I don’t expect OBOS to conquer the world, but 1% is a pretty reasonable target.
That would be 20 million people by year 2006.
I am hoping that QT3 will get ported when possible.
That would need to be done by a third party, because OBOS market isn’t significant enough for OBOS. This isn’t a lame port from Windows/X11/OS X version, it should be implemented from scratch for OBOS.
There is a universal driver platform UDL IIRC that could help in the long term to make writing drivers more portable.
Linux doesn’t support UDL, and IIRC, same with Windows – correct me if I’m wrong.
There is nothing that *I* am aware of in the C++ spec that requires FBC to occur. And there are certainly ways around it in the compiler writer’s world…
I am not a compiler designer, so anyone please correct any of my next statements if I’m in error. In the article, the author states that it is possible to write a C++ compiler that doesn’t have FBC problems. I suppose the real issue is that nobody does this. However, I’m not even sure how easy or beneficial this would be in the first place. One of the benefits of C++ much of the object association is done at compile time, not runtime. This makes for much longer compiles, and can make for much larger binaries. However it minimizes the runtime burden on the system.
I took a detailed class on C++ programming and I was astonished to learn that you can basically replace “class” with “struct” in your object definitions, and all you’ve really changed is whether methods/variables are private or public by default. In essence it is more probably that you can achieve C-level speeds, because it is possible to have the compiler take care of a lot of these operations that are done at runtime in Java-like OOP implementations (I’m thinking Obj-C is like that too). FBC is the tradeoff however, that one has to deal with.
“That would be 20 million people by year 2006.”
Yep,I am suggesting that only as an upper bound. I’m always amazed at the BeOS enthusiasm I find in Russia, Japan & other non English speaking countries web sites.
The QT2.3 was ported by Zenja, but it needed Bone & something else, its on BeBits but I can’t use it. The issue has more to do with the licenses & if the QT3 source is available.
Actually once the networking & Bone stuff is out of the way, I would be as happy to give QT2.3 a try.
If I was interested in writing drivers, (which I’m not yet) this is what I would be studying, the pdfs are at
http://www.projectudi.org/
This is 1999 Intel/HPCompaq/Caldera/Sun/IBM/lynuxworks/etc/ inititative so it seems slow on take up. I don’t think Linux can use it yet, but the Linux people need this as much as anybody else. UDI has been discussed in some of the OBOS forums as well. If Linux had already been there, then OBOS could follow on, so it may be premature right now.
Googling for xxx drivers yyy always helps.
Anybody whos been around will remember that C++ used to be Cpp a pre processor that pre compiled the newish OO extensions into plain C, so as you say, class & struct are very similar, class being higher up the food chain than struct.
That is why OO is possible with just plain C, its just that the developer ends up doing in style & hard work what the C++ compiler could have done for you. Things have moved on so much that few would do OO in plain C. Jast as asm can do anything C can do, just takes x more effort & code.
If BeOS is standing on weak ground because of FBC, it’s in pretty bad company, MFC, wx, QT, & a zillion other frameworks. Objective C may claim the higher ground, but the only mainstream platform for it is OSX & NextStep, and clearly OSX is known to have severe speed issue, although that has little to do with OC, more the over use of eye candy & lack of optimisation as yet. If OSX ever moves like a rocket, the interest in OC will surely increase.
One thing that is essential for OBOS is support for the ogg formats. Especially when ogg theora (their video codec)is released.
@rajan r:
The only way I could see SoundPlay being inferior than Winamp is that SoundPlay doesn’t skin.
SoundPlay is shareware. If not registered, it sometimes stops playing the music and plays a voice that says to register it.
I prefer CL-Amp.
* DVD software that’s as good as PowerDVD
That would be hard to be done by any OSS project. Thank the Congress for the support they gave us. Oh wait, when they were passing the DMCA, they were all tuned into the Bill and Monica soap opera.
There’s no stupid thing like the DMCA here in Germany or any other part of the world. YellowTAB is a German company. Don’t be so USA-focussed.
* Ability to play mpeg4/divx/real media/etc
Except for DivX, the others are impossible.
Yeah, DivX ain’t MPEG-4. *LOL*
Real Media isn’t impossible. DivX for BeOS used Windows-DLLs. The same could be done with Real.
MPEG 4 for example is hounded by royalties that an OSS project could never provide.
Ever heard of XviD or FFMPEG? No?
@JJ:
Don’t count on UDI. Look at the calendar on the web page. It’s from 2001.
I think UDI is dead.
Vorbis is allready supported by BeOS (check BeBits).
Theora is based on VP3 – VP3 also is open source.
Someone with enough skill can start porting it now.
Rajan .NET doesn’t need an object model a la CORBA or COM since every object executing on the .NET runtime inherits from System.Object, it’s part of .NET specification. One could argue that the .NET approach is more of an object system but it does solve the same problem CORBA and COM does.
So the approach to remote object access that DCOM and CORBA solves doesn’t materialize in .NET since you can take the more natural approach to remote objects in .NET by using proxies. And as to components, yes I agree there is a need to have cross application sharing of objects on a platform and these are also part of .NET but most applications are interested in having only their own instance of this object and as such both COM and CORBA are lacking.
I believe there is a need for another OS.
Linux is too difficult for most of the desktop world. MAC comes with a hefty price tag and there really are people that want an easy to use alternative to Windblows. XP is particularly intrusive.
In addition, audio/video market would still welcome it as long there were decent apps, which is a problem i admit.
Mixing multi-track audio and adding effects like reverbs is still damn demanding. Audio pro’s benefit from every last drop of performance which a lean OS can give. They also benefit from stability and an OS that is tuned to prioritize audio, as opposed to networking or office apps. The audio community’s needs are at odds with the general computing world, or perhaps just Microsoft/intel’s needs. Audio needs ever more performance. MS and intel want to use up more and more performance with the OS to force you into buying a new PC.
Mainstream audio is using 24 bit and it sounds good but it still seems to lack some of the dynamic range that TOP NOTCH tape has. That tells me digital audio has longer to go for better sound and that could drive up performance needs even more. Just take plug ins. Those emulations of old analog gear such as compressors, EQ’s, etc still fail to live up to the sound of the old analog boxes. There is room to improve on digital audio, and i for one would welcome back an OS that is made for multimedia as be was. So thank you OBOS. keep up the good work and stay positive. Your efforts are greatly appreciated.
All I need in an OS is.
Internet applications:
*) A good email client
*) A good browser (which would be on IE or Mozilla level)
*) A good communication program (I don’t mind if it’s a own made protocol or a existing one like ICQ or MSN)
Standard apps:
*) Calculator
*) Simple text editor (like notepad)
*) Standard text editor (like wordpad)
*) Character map
*) Media+CD player
*) System maintain utilities (Defragmentation, Diskscan etc)
Network support:
*) Easy setup like BeOS had
*) Share options with Linux and Windows machines
*) Internet share options
Operation System
*) Efficience
*) Stable
*) User friendly
This is what people need. And now I look at it, hey! It’s just like Windows!
Personnaly, I would like to see PalmSource continue the development of DANO, license the binaries and publish all necessary APIs.
To this day, I am not angry with them for not announcing or releasing anything because the desktop market is still under the grip of the 800lbs Gorilla aka Microsoft.
I hope that some PalmSource employees (preferably (CPO) Chief Products Officer Steve Sakoman) keep an awareness of the demand for the OS in mind as the market changes. I hold him responsible for convincing the Board of Directors to release a desktop OS at the appropriate time.
There *is* *still* hope. PalmSource is about to split from PalmSolutions. Judge Kollar-Kotelli will soon render a verdict on the Microsoft remedies that could potentially open the desktop market! Microsoft will no doubt appeal but Time will tell!
In the mean time, Kudos to the OBOS developers. OBOS would be great even if PalmSource releases a desktop OS.
ciao
yc
If it is a good solid design, and works without compromise – people will use it. If it is a good solid design but requires effort – people will toy with it. Be did very well to accommodate a quality design. However, they did leave some things to be desired, making it a toy. Is there a commercial niche for BeOS/OBOS? No. Not as a desktop environment in which the requirement for MS Office is so great. Not when much larger masses can’t create interoperability (Linux). So, as Mr. Phipps states OBOS _can_ be something to be proud of, something to be shared, something to continually grow. But, to one’s personal satisfaction, not the masses.
If the OBOS project’s objective is to create a toy OS – success is achievable. If the OBOS project’s objective is the create a revolution based on technological marvel – forget it.
@KAMiKAZOW
I can’t say if UDI is dead or alive, perhaps not yet got any traction, I will look some more at it.
Remember Intels USB1.0 took a while to take off, & Apple pushed it out the door in a big way long before Wintel. These things take time. Gee FireWire has been around 10yrs I think.
As an ASIC guy I can say something about drivers & their development. I have been the chief architect of network ASIC devices that were extremely difficult for the SW guys to write drivers for in house. If the chip had gotten out to the real world, only Windows would be supported, that is all the SW guys knew or had time to support, & Windows driver development is a big thing to learn esp Win95 then NT driver model. The idea of supporting some outside driver guys for other OSs now seems utterly unworkeable in hindsight.
One thing is for sure, most HW guys are not usually that good at SW and vice versa. HW guys design chips to meet the specifications they are given, & will fill in missing gaps in the simplest way possible. SW guys get a hold of the chip after it’s done so there is little chance to go back & change anything. SW guys do not generally understand the ASIC in anything but a superficial way, usually the stuff they control & the general purpose of the chip. HW guys are usually just as clueless about networking stacks TCP, kernals etc.
Now on later projects, after torturing the SW guys on the 1st round, they had something that barely works & the ASIC is about to be redesigned as an FPGA for rapid prototyping (1 day turn around) & easy spec changes. Now the SW guys can participate in the changes they want to make to FPGA so the driver will be better simpler faster cheaper etc.
Guess what, the SW guys don’t speak HW, can’t think how to present what they want unless it’s real trivial. The HW guys can’t guess what they want either. The SW guys start asking for feature changes that they think are simple but will wreak the ASIC design schedule but they also want changes that are incredibly simply to do in HW, but because they don’t know this, they don’t dare ask.
In the end, the principal driver guy had to learn basic Verilog HDL & then at least try to write in HDL what they thought they wanted. After exchanging this a few times the design settles in to something that costs very little in HW but makes the SW guys very happy to be in control of the interface.
The above story could apply to most ASICs I believe so this explains why the HW often looks so screwed up from the SW side. If you can write drivers & are familiar with some HW & a HDL, there are jobs waiting for you as the HW-SW architect.
Just as there is science in most SW fields, there is certainly the same in HW design, but the interface & drivers will remain a minefield till there are far more HW-SW experts in that area. If the Windows driver is easy to build because the SW guys helped on the HW side, the other OS drivers will benefit as well. It still remains difficult though to release docs for the interface as that is almost part of the ASIC database if it is anyway accurate.
JJ
“SoundPlay is shareware. If not registered, it sometimes stops playing the music and plays a voice that says to register it. I prefer CL-Amp.”
Can you not afford the whopping $12.00 registration fee? What ever happened to buying products, especially when they are best in class and have an incredible price point?
i understand the decision that palm has made. In their position is makes sense. however, i would rather see OBOS suceed then the release of beos from Palm.
In the latter case, the longevity of the project will always be under scrutiny. Developers will not come. We already saw be pull the plug whats to stop palm from doing the same? Plus, Palm would probably try to turn it into another office destop OS. who wants that. not me. OBOS is the future not palm. I think we should all forget about palm. Open source is the only way to go on the x86. nothing else will ever survive as long as MS does.
Is UDI something that could be wrappered? ie: make a UDI wrapper for Linux and one for Windows and one for the other OSes like OBOS and whatnot…and then open source these wrappers using a very liberal and business friendly license like BSD or MIT…
this could potentially be very benificial to all the OSS OSes…tell the HW vendors that they can write just one driver for Windows, Linux, FreeBSD, OBOS, etc.
just a thought
-bytes256
don’t complain, you don’t have to use OBOS.
There were so many open source OS projects and most of them failed or will fail. I’m quite sure OBOS will not. It seems that they know what they do, the website is alive and they coding, coding and coding and the results are astonishing.
Licensing: MIT license or BSD style license is so much better than GPL for an OS.
One thing that has almost gotten lost in this is that, it is assumed by many that OBOS is trying recreate BeOS 5…and then go from there. Just as he said that OBOS is not Be, Inc, I do not see any reason why OBOS is trying to just duplicate r5 and be years behind. I’m assuming they are addressing that as they go along.
The only thing I want from Palm is release the BeOS name.
I think OBOS is already better in many respects than would be release Dano code. The longer Palm is sitting on BeOS code the better OBOS became.
But I really like the name OpenBeOS – that’s what the OBOS project actually is. If Palm sells BeOS trademark to BeUnited or OpenBeOS team that would be good enough.
If you think about what you said it sound like all the open OSs should share the driver model part of the OS. I don’t see that happening, but I don’t see why UDI couldn’t be made to work after all Intel doesn’t usually float dumb ideas (ignoring the original 8086 design).
Since Intel does exhibit strong Linux interest, it has an interest in pushing common driver development, but as I said in my article, drivers are usually dirty even when done internally. I don’t yet know enough about UDI to know if my x company could have used it.
First, nobody here is trying to create a revolution. We are trying to create an OS that we all love to use, and can hopefully make a full time job using it someday. If we capture 20% market share, holy bats cowman! I’ll consider OBOS wildly over-successful if we capture a mere 2% of the market.
Second, beunited.org and it’s place. beunited.org is not a for-profit company, it is a non-profit entity. beunited is not looking to become the RedHat of the OBOS community. We are looking to make sure that OSBOS, no matter what flavor (be it yellowTab or another distributor, and be it Cosmoe, OBOS, B.E.OS, or Zeta) all have a familiarity about them when it comes down to the user, and that programs that compile on one will easily compile on another. Beyond that, we will help bring applications and drivers to the OSBOS platforms. We will help with marketing and branding. We will not be a commercial company. Think UnitedLinux, but starting from the get-go with a bit more clout (since the major planned distributors are playing with us from the start).
Third, there are a few new features that have crept into OBOS R1, but to keep a reality check on it, Michael will not allow feature creep as a rule. There is so much that needs to be done, fixed, added, and innovated – if they tried to do it all at once, it would take them 5 more years. R5 is the first step. So no, Jay, officially they are not attempting beyond that at this point. You have to start somewhere, and you have to have a first release sometime, and Michael is doing a wonderful job of keeping feature creep in check. Everyone has the same goal – R5. Until that is reached, there is no OBOS, only bits and peices.
GlassElevator is the part where new features are discussed and prototyped – separate project, but mainly the idea is to have “improvements” ready for an R1.5 or something soon after an R1 release. So technically, they are thinking about it and pushing beyond R5, just in a different space.
And lastly, the application chicken and the egg problem will not magically disappear. Be, Inc. dealt with it, and so will OBOS. However, with standards in the APIs, any application on one OSBOS platform will also be capable of being on another, thus spreading the wealth. BeWINE work can continue with the availability of mmap in the kernel (which was not in the BeOS kernel), and with Java, that will bring a few more apps around. The MS outcome may have some additional small affect of bringing down the application barrier to entry some as well. All that we can do is try to work it down bit by bit, and hope for a point of critical mass to be reached.
Windows w/NT kernal a true OS.
Yes… this makes perfect sense seeing as how all the others listed are false OS. Embrace, extend, destroy…
>> The only thing I want from Palm is release the BeOS name.
You’ve got to be kidding! Dano is so much better than R5.
When you take a close look at Dano, you see a beautiful network architecture, a nice start on new services, improved GUI with lots of new GUI classes (some experimental), XML Classes, SmartCard classes, improved multimedia support, not to mention the excluded revamped OpenGL environment and much more.
Palm has something definitely worth releasing at the appropriate time! Dano would be a very competitive OS if finished. It would have had a good unix feel but without all the baggage and a slick super fast UI.
It will take OBOS a long time to get to where Dano is today.
ciao
yc
Just like the previous BeOS thread, this one generates a lot of comments, complaints, information, misinformation, FUD, etc.
I have to wonder how many people commenting have actually used BeOS. The Personal Edition is still freely available from BeBits, including a version that can be installed in Linux instead of Windows. I didn’t instantly decide to switch to BeOS after trying PE, but it didn’t take long before I decided to get the Pro version and install it in its own partition. I’ve tried Linux, I’ve tried FreeBSD, and for the home user, BeOS is definitely better, faster, easier to configure, easier to use than those. Obviously, BeOS was not intended to be a major server OS, and is not as good as Linux and *BSD in that area. But it sure knocks the socks off of the X-based desktop.
No, I can’t do everything I want to in BeOS, but it’s hardly devoid of apps. Yes it does need more, *good* apps, but there are people who are working on them, and the apps that already exist are good enough for many uses.
BeOS was a well-designed OS, and because of that, the job of creating OBOS is much easier than creating a brand-new OS from scratch. And as has already been mentioned elsewhere, some stuff, like the OBOS net_server, already seem to be working better than the BeOS R5 versions, so when OBOS R1 comes out, it should already be better than BeOS R5.
Yes, there are inevitable problems, like the drivers problem, and giving people reasons to switch from other OS’s, but as I’ve said, OBOS is in a very different situation than Linux. OBOS has a “leg up” thanks to the work of Be, Inc., and by going open-source, OBOS opens the doors to greater possibilities. Even if these possibilities don’t pan out, even if OBOS is *never* a major success, open source means that OBOS will continue to survive for many years to come, since they won’t have to rely upon making a profit.
But given the nature of the OS, it’s hard for me to believe that it won’t find a good niche market or two, markets that Windows and Linux can’t really touch, for various reasons. And who knows? With such a good desktop, it might indeed make major inroads against Windows. It certainly has a better chance at it than Linux.
I *like* using BeOS, and I eagerly await the release of OBOS…
In a perfect world, we wouldn’t even need the MIT license; all software would be public domain. Unfortunately, lots of people like to take software and not give back. That’s why we need the GPL.
The entire idea of intellectual property is stupid. The reason we have property is to make sure that people are not denied their property unfairly; but with IP, if you give someone else your idea, you still have your idea. Better for everybody to share all their ideas for free than to have to introduce money into it. And that’s what the GPL is for.
That being said, all of the reasons Phipps gave for not wanting the GPL seem to say that LGPL is the perfect license. The LGPL says that any changes to the source itself have to be contributed, but any applications that link against that source can be closed. That means that a developer who started with the OBOS kernel, and rebuilt everything else on top of it (for some strange reason), would have to release their modifications of the kernel, but could keep the program itself private.
Of course, the whole thing’s moot. After all, the way the MIT license works, anybody who wanted to could make a new Sourceforge page, called it the “GNU Modern Desktop OS”, fork the OBOS, and start developing; and if people moved over there, there’s nothing the OBOS guys would be able to do about it. If a significant number of developers switched over to there, then OBOS wouldn’t even be able to use the code.
And what developers would want to join this fork of the OBOS code? Already it is clear that most of the naysayers on OBOS are programmers, so they can’t do anything. Mose of the developers already working on OBOS are already there and had lots of chances to branch before this and did not. The other BeOS clones out there are getting thier developers from thier original branch source Aethos(sp) and Linux. Most Linux developers want to develop Linux not BeOS and will not join this fork either.
But the main reasons developers will stay with is 1) It has a fixed goal to develop towards. Only non-programmers talk about adding features onto a project before the framework is finished and working. Real programmers like having a fixed goal and are not going to join a project with the idea of adding all the features BeOS is missing before a stable platform is written. 2) OBOS has lasted a lot of hard knocks and still is plugging along. One guy last year complained that OBOS was not good enough and started his own project, he folded in two months. Anyone who joined him just wasted 2 month programming as he was using a model (custom objects to replace kits running on top of Linux) that did not produce useful that could run in other OSes.
The OBOS team is doing a great job, far better than 98% of the other OSS projects out there.
If a significant number of developers switched over to there, then OBOS wouldn’t even be able to use the code.
You cant take MIT licensed sources, bring them somewheres else, and change the license. It doesnt work that way. Just like you cant go and take GPL sources, turn them into MIT and close the sources. The license stays with the code until the author decides to change it. Im glad to see them choosing a non-gpl open source license.
“Mixing multi-track audio and adding effects like reverbs is still damn demanding. Audio pro’s benefit from every last drop of
performance which a lean OS can give. They also benefit from stability and an OS that is tuned to prioritize audio, as opposed
to networking or office apps.”
I’ve been beta-testing an Amiga program that is similar to CollEdit
etc.
On the amithlon emulator, using an Athlon 1700, this can mix up to 60
stereo tracks. Tests on the real-time Reverb (which is preliminary
code) show that it can manage Reverb on up to ten stereo tracks.
Beyond this, you get a polite warning message.
No glitches or hitches in the current beta.
I would expect equal or better performance on the forthcoming PPC
Amiga, which the program is planned for.
How does that compare with what you expect?
AMIGAOS does seem to have most of the things people are asking for in
BeOS now, the main problem being that the browsers are about 85%
finished. JS more or less works, but no CSS yet.
But all the stuff like CD burning programs, text editors, Samba, etc
are there.
I would still like to see a version of BeOS succeed too, but I have
doubts if it will attract enough software developers.
First of all, I’m all pro OpenBeOS. Don’t think that any of my criticism is meant to detract from the project, I think they’ve done a phenomenal job. Second, I can only laugh at some of the comments I’ve seen. People are saying, for example, that “Linux is too hard, let’s make another OS.” That’s an insane statement. Isn’t it much more work to make another OS than to make Linux easier to use? Then, there are comments that using parts of the Linux kernel (licensing issues aside for a moment) would be shortcuts. The Linux kernel is a very mature, high-performance piece of code. Using it would make for an excellent foundation that won’t be matched by the OBEOS kernel for years. I think the key thing that everyone has to distinguish is exactly why does OpenBeOS exist? If its all about making a fast, stable, desktop replacement OS, then there is little question that the best course of action would be to start with the Linux kernel and XFree86 as a base, and build upon that, much like BlueEyedOS is doing. The advantage to this would be three-fold. First, it would be easier. Second, it would be far more stable far more quickly. Third, it would be faster. Barring any advances in kernel design that the OBEOS people might come up with, one has to remember that both XFree86 and Linux have years of tweeking behind them, and starting a new project loses all those years of optimization. Additionally, since most of the performance issues on the current Linux desktops are due to toolkits and applications written without performance in mind, writing a BeOS userspace would eliminate the #1 performance bottleneck. On the other hand, it seems that OpenBeOS’s goal is not just to make a good desktop OS, but to recreate the BeOS user experience, and have a lot of fun in the process of developing it. In that case, its unquestionable that the OBEOS people have chosen the right course of action. It is important, then, to keep in mind this double-sided set of priorities. Trying to understand or comment on OBEOS while coming from the first perspective will likely lead to incorrect conclusions.
@mjang:
Can you not afford the whopping $12.00 registration fee? What ever happened to buying products, especially when they are best in class and have an incredible price point?
CL-Amp does everything I want. Why should I spend any money, if I can get the same thing for free.
I want support for MP3, OGG Vorbis, and M3U-Playlists.
I don’t need to play files backwards or with another speed. Just useless features IMHO.
@Richard Fillion:
You cant take MIT licensed sources, bring them somewheres else, and change the license.
Yes, you can! The only duty is to keep the copyright notice.
Apple re-licensed BSD/MIT source under the Apple Source License.
If anybode doen’t like the MIT license then just re-license it under GPL. You can relicense MIT sources to anything you like. That’s no problem – just keep the original copyright notice.
I can see myself using any or all of these & will install them only as they become useable & stable. Both aproaches have advantages, OBOS is pure & the others have the chance to run some or maybe all Linux apps & maybe even Carbon apps too. But the wait will be worth it either way. By the time these are released, maybe Atheos/Syllable will be worth a look.
Just to let people know. ive been looking at UDI a bit. Im planning on getting aroudn to reading teh specs. they have released reference software that is available from a link on their sit to a SourceForge site. What it does is it wraps around current Operating System’s driver models to give them UDI. What I encourage peopl to do is to downloiad it and port drivers to it whcih would help increase it influence. Also Adding a warpper for anotehr OS like OBOS or Syllable would start to give them support and I would assume make a native port easier since you then understand how the native and UDI APIs map together. UDI would solve the Number 1 problem of hobby OS’s, drivers. The only otehr problem then is software.
I belive someoen made the comment of making OBOS teh clint of Linux. I agree with taht and for R2 they should add multiuser and LDAP support. Also maybe driver wrappers for stuff liek QT and GTK– whcih would allow more apps while giving them a native feel will greatly solvet the software problem for short term andt hen work on making programs cross platform ro writing nw apps would be for teh long term. I think they should actually decouple the GUI and try and push it for a standard cross platform GUI replacement for X esp if networking support is added. Throw on a source comatible wrapper for X drivers. I would see this as the short term replacemernt for X with longterm being Fresco.
My overall vision is Linux(X11)->OBOS(or Syllable)->L4/HURD(Fresco)
As I liek to comment a lot i have been working on an GUI design that tries to incorporate the best ideas form many OS and even goes beyond them
http://cubesoft.sf.net/Design.html
The Linux kernel is a very mature, high-performance piece of code. Using it would make for an excellent foundation that won’t be matched by the OBEOS kernel for years
My guess – you don’t know what OBOS kernel is if you make such statement.
OBOS is using fork of NewOS kernel which was developped by ex-Be engineer. It may looks like kernel is only 2 months old but in reality there are more than 10 years of kernel design behind it.
Calling Linux kernel “high-performance piece of code” is asking for flame war on this forum.
If people let you slide this comment here it doesn’t mean they agree with it , it means they tired of flame wars.
“Can you not afford the whopping $12.00 registration fee? What ever happened to buying products, especially when they are best in class and have an incredible price point?”
It’s not a matter of being willing to pay the cash, it’s a matter of comparing a Freeware product to a Shareware product. All other things being the same, who wouldn’t want the Freeware?
As for me personally, I don’t mind spending money on commercial software – I do frequently. But when purchasing software I expect to get a hard copy, a written manual, and support.
OBOS is using fork of NewOS kernel which was developped by ex-Be engineer. It may looks like kernel is only 2 months old but in reality there are more than 10 years of kernel design behind it.
You do realize, I assume, that NewOS was written as essentially a “what if” and was not written to recreate the BeOS kernel. Not that it invalidates it; just that being written by a Be engineer doesn’t make it the Be kernel.
i’m with most of the folks here, and i don’t expect to pay for soundplay, simply because it’s simple app that thrives in a small pond that is BeOS.
it’s an app that plays mp3’s and whatever other format you chuck at it, it is not ‘money costing ‘ software.
it’s a given nowadays, having a media player. i mean, if marco had written the “protools” replacement, yeah, sure, give the guy the cash, but a sound player? something that plays mp3s backwards?
that’s my opinion, but maybe i’m just a bit to miserly?
http://newos.sourceforge.net/
I dropped into NewOS land for a bit on SourceForge. It looks like a good start. So far the NewOS has implemented the following features:
* Multithreaded
* Multiprocessor
* Fully reentrant kernel
* Protected memory
* Separate user and kernel space
* Text-based console
* Full locking primitives
* Kernel debugging support with installable debugger commands and remote gdb support
* Modern VM design (demand paging, swapping, memory mapping), with support for full filesystem cache integration
* Dynamically loadable kernel modules: drivers, filesystems, generic modules
* Full virtual filesystem layer, device file system
* Initial support for iso9660 and ext2fs filesystems
* Full user space shared lib support
* Kernel based network stack (UDP/IP for now)
* Remote block device
* Basic VESA mode support with generic framebuffer console
Perhaps there is a comparison of all the kernels that are out there… I wonder how NewOS differs from the QNX/Neutrino kernel. And I wonder if any of these kernels is as fast and tight as the Amiga kernel was.
Anyhow, with all this Brownian code motion, something has to come of it.
#m
>> but a sound player? something that plays mp3s backwards?
C’mon people give the guy some credit! He started this and the CDDA file system before joining Be. CDDA is now part of BeOS! That’s probably why he got hired! He is also very helpful in helping other freeware/shareware developers with BeOS.
SoundPlay is much more than your typical MP3 Player! It supports Mixing, Network Remote Control via browser, Network streaming etc…
I use CL-Amp myself but I think soundplay is more feature rich and well worth the registration fee. Here is the description from BeBits:
Award winning audio player (mp3, Ogg Vorbis, wav, midi, about 15 different mod formats, and a variety of other formats) with winamp skin support, pitchcontrol, crossfading, audio effects, VST plugin support, visualization plugins, mixing, network remote control, network streaming (client AND server) and lots more.
ciao
yc
Calling Linux kernel “high-performance piece of code” is asking for flame war on this forum.
>>>>>>>>>>>>
Why? It is. Take a look at any benchmarks. For the stuff that a kernel is supposed to do (manage memory, map files, do file I/O, etc) the Linux kernel is damn good. Generally, its faster than the BeOS kernel ever was (and that’s pretty fast!) I/O is faster, networking is faster, and yes, even IPC is faster. Read the interview with the BlueOS guy posted earlier on OSNews: http://www.osnews.com/story.php?news_id=215.
I’m interested to see a lot of mention of Mp3 players in this thread.. I think there’s a useful comparison there.
Windows media Player users – these guys use what came on their PC, it works reasonably well and no complication.
These are the guys who will stick with Win98 / ME / XPHome
Corona beta testers. Cutting edge guys.. want the latest features etc. Most likely to migrate to a bleeding edge Linux kernel in the long run.
RealOne – Willing to pay for something thats heavily marketed and looks better. Likely to upgrade to WinXP.
Winamp. Ultra light, works well, looks good, uses no resources and the missing features are bearable on account of its speed and size.
These are the guys who might upgrade to Open BeOS
They don’t mind that it can’t do everything the opposition can do, as long as it makes their old PCs run well, and reliably.
And BeOS R4 and R5. But I certainly wouldn’t switch to BeOS. Maybe Dano, but I’ve only seen screenshots and read stories about it. The BeOS I tried had too many strange problems. And neither R4 or R5 ever properly recognized my second CPU on my dual P3-450. The UI was fast as can be but apps were less usable than Windows. I loved the speed, the low latency, and the kick ass file system. But beyond that, it had miles to go before I’d switch to it for full time work.
For hobby purposes, I would much rather use the new Amiga system. It somehow seems to have kept its spirit alive while the BeOS people… I don’t know what they are doing… probably because Be never knew what they wanted to be when they grew up… and since they never grew up…. none of it matters anyway…
The BeLinux frankenstein sounds cool, especially since it can leverage much of Linux.
Perhaps it is ingrained marketing, but I don’t see any of these systems achieving critical mass. Perhaps it is a lack of patience. I think together they could achieve more than they are doing alone.
#m
I did say – As long as it makes their PCs run well and reliably…
I think there are pretty good chances that the OS will succeed in some way.
Basket and the egg.
I think the problem only applies to big software packages. The other day I was looking for a FTP client, and a quick search found _a_lot_ of programs (same goes for other types of applications). A lot of these developers are properly looking for other ways of making a profit of the their product(s), and OBOS might be the an answer.
As for Office-like applications, the platform definitely needs a user base. But on the other hand, MS decided to release OfficeX because they (thought they) could sell 700.000 copies. Gobe might want to reconsider their support to BeOS.
C++ or not to C++
Be chose C++, and obviously OBOS will be using the same type of API!!! Some like it, and some don’t. One think to keep in mind is that C++ can be used in a more “civilised” way by not using things that might make it difficult to some developers (templates, whatever).
All languages have pitfalls. For example ObjC have an odd way of deleting object, by maintaining a reference count. A simple concept, but it is actually easy to design code that end up with an un-referenced object (and ObjC have no garbage collector).
The end result
Some comments have been made about the quality of the finished product. Because OBOS R1 is going to be a clone an existing and well proved/designed OS, it is pretty clear when R1 is finished.
As someone pointed out, Dano have a lot of new features (classes etc.), all of which would be nice to have in OBOS. These things have to wait until R2, and just by looking at Cocoa/MFC it should be pretty easy to make a list of new and useful classes and functionality.
Drivers
OBOS will definitely not support as many drivers as Windows. But supporting the most popular hardware will get them a long way, and OBOS’s ‘supported hardware list’ will give users a list choices of OBOS-friendly hardware before they purchase new stuff.
So in my book OBOS will succeed in some way, and my only problem is that I would like the OS tomorrow ;-). The Open Source approach makes sure that the OS will live forever.
If BeOS Dano were purchased back from Palm and whipped into shape with a bunch of new drivers and support for the latest hardware, a BeOS package system/installer were built, and a BeOS apps depot was put together…
Why not BeOS instead of Xandros?
BeOS offers a GUI today. OpenGL today. Cool multimedia today.
BeOS would offer a much more fun, more interesting, more compelling consumer experience.
#m
>>>the Linux kernel is damn good. Generally, its faster than the BeOS kernel ever was (and that’s pretty fast!) I/O is faster, networking is faster, and yes, even IPC is faster
Ok, ok, nobody means that the linux kernel isn’t good, now, how comes that BeOs is so much more responsive than Linux ? I have been using Linux since 1993, I like it, but it’s way slower than BeOs, from an “end user point of view”.
It’s a real question and I have been thinking about it a lot, because I was so impressed by BeOs quickness. Sure, BeOs has no security subsystem, so it doesn’t need to check anything like user validation, access rights, ACLs, etc. Perhaps when BeOs has security inside (as soon as possible please, because security is not an option but a must), it will be a slower…
In addition, Linux, like every Server-Os, launches lots of services/daemons at start time, whereas BeOs, as a good “Client-Os”, only launches a minimum of processes when starting.
Any other ideas about that ???
Ok, ok, nobody means that the linux kernel isn’t good, now, how comes that BeOs is so much more responsive than Linux ? I have been using Linux since 1993, I like it, but it’s way slower than BeOs, from an “end user point of view”.
That is mainly because of slow XFree86 and slow ALSA. Having a fast kernel doesn’t mean the entire OS is fast. Everything else must be fast, BeOS managed that.
It’s simple. When I’m using BeOS I feel better, that’s enough for me.
Impotant for OSBOS succes'(the boss of the OS’s ;^)!)=>
>homogenity – so much people can help you out if they’re looking at the GUI.Be jeans one size fit’s all, hey almost everybody is wearing jeans.
>Be the fast multitasking MediaOS: which one really is it? Irix, Linux, XP haha! If it works on older hw, the 2nd and/or 3rd third world will enjoy it too, do you know *how many* people that are?
The first thing a new user should notice : it’s name and what it does or can, the installer, than the GUI and a happy feeling!!!
Paul
It’s simple. When I’m using BeOS I feel better, that’s enough for me. But I’m not alone (luckily)!
Impotant for OSBOS succes'(the boss of the OS’s ;^)!)=>
>homogenity – so much people can help you out if they’re looking at the same GUI. Be jeans one size fit’s all, hey almost everybody has got jeans.
>Be the fast multitasking MediaOS: which one really is it? Irix, Linux, XP haha! If it works on older hw, the 2nd and/or 3rd third world will enjoy it too, do you know *how many* people that are?
>The first thing a new user should notice : it’s name and what it does or can, the installer, than the GUI and after that a happy feeling!!!
Paul
You are right, I forgot the “XFree86 factor”… But until Fresco is available, I prefer to stick to BeOs 😎