Slashdot Mirror
AtheOS Fork Brings BeOS on Top of Linux
Eugenia writes: "Yup, Bill Hayden has forked AtheOS by using its app_server and Interface Kit (along with some other of its kits, like the filesystem layer) and ported it on top of the 2.4.x Linux kernel, without the need for X11. He already has the graphical environment working, and he also has some BeOS apps recompiled and working under Linux. Why BeOS applications? Because that was the reason of the fork. Exactly because AtheOS and BeOS have similar technical principles (highly multithreaded, truly preemptive, similar C++ API etc), by modifying AtheOS's API to match BeOS, Bill is trying to resurrect the BeOS. By doing so this way, Bill is already way ahead from the other two efforts to ressurect BeOS, OpenBeOS (dependant on the 'clean' NewOS kernel) and BlueOS (which depends on Linux and X11)."
That depends on how you define 'way ahead.'
People like me who really like BeOS admire the entire structure of the operating system, from top to bottom. I have zero interest in running FrankensteinBeOS, which is what this sounds like. Therefore I am content to work on the OpenBeOS project, which may be 'way behind', but should have a nicer outcome (for people who like BeOS). The project is coming together quite nicely for something so young.
(It's easier to see really far when you can stand on the shoulders of great engineers.)
"And like that
Originally found on:
8 21 5112&list=2311
http://www.geocrawler.com/mail/msg.php3?msg_id=
FROM: Bill Hayden
DATE: 03/26/2002 06:59:50
SUBJECT: [Atheos-developer] Atheos Fork Announcement
Well, it was not my intent to announce this quite this soon, but given
the recent conversation on the list, I feel that it's best not to wait
any longer.
I forked Atheos about 6 months ago and have been continuously developing
it since that time. I've taken it in some very new directions. I
should warn you that some of you will absolutely love the changes, and
some of you will perhaps feel that the "dream" of Atheos has been sold out.
The new project has had a name since the beginning, but I'm going to
hold off on releasing that until I can verify that the domain names and
trademark are secure. So I'll call it "New Atheos" for the purposes of
this e-mail.
New Atheos has the following major new features:
o Runs on top of the Linux kernel, not the Atheos kernel
o Atheos API has been merged with the BeOS API
o PowerPC support
o gcc 3.0.X compatiblity
o OpenTracker/Deskbar desktop manager
These features give the following benefits:
o Most BeOS programs compile and run with little or no changes
o Linux kernel means that CD-ROM, CD booting and installing, DHCP, etc.
work
o Linux kernel means that driver support is excellent
o Mac users get a piece of the action
Things I haven't even started on:
o Printing
o Media Kit
o Replicants
Existing Atheos programs will need changes to compile. I haven't found
one that took me longer than a few minutes to "convert". Where Atheos
and BeOS use different semantics, I chose the BeOS method.
I am going to hold off on a release until I can successfully compile and
run OpenTracker and Deskbar. They use just about every obsolete and
goofy BeOS construct that exists. I'm most of the way there, though,
especially for Deskbar. Kurt wasn't lying when he said it would be a
nightmare to port them. Of course, I'm doing an "anti-port". When some
BeOS program won't compile, I change the API to match it instead of
changing the program itself.
The first BeOS program that successfully came up was Pulse, and there
was no small amoung of satisfaction to see good ole' Pulse running on my
new system. Nostalgic BeOS users can perhaps understand.
I'm writing in a hurry, so hopefully I haven't forgotten something
important. And no, I can't give a release date yet. I hope to have a
CVS server up at the time of release.
Thanks,
Bill Hayden
Wasn't one of the touted features of BeOS its low latency, single-user kernel optimized for multimedia stuff? Demos had multiple video players all playing smoothly, while 3-D animations occured in other windows.
How well can the Linux kernel deliver such performance?
There's 10 types of people in this world, those who understand binary and those who don't.
that's GnuBe/Linux to you sir....
Which is exactly why things such as this split happen, and should happen. He can do what he likes, but if it doesn't suit enough people, they'll go elsewhere and make their own. So whether he likes it or not, this is how things should be. I'm glad to see someone doing something interesting with AtheOS.
It's not enough to bash in heads, you've got to bash in minds. - Captain Hammer
I appreciate what this guy is doing, but seriously folks, why the hell is everybody so intent on making some sort of BE/Linux hybrid? I support OpenBeos for the following very good reasons:
1) Has over 100 developers now
2) Intent on rewriting original Be api so that compile and eventual binary compatibility is attained
3) Uses an alternate liscense to GPL so that open source is maintained without frightening away commercial developers due to fear of *GPL Contamination*
4) Already has contacts with commercial developers and distributors (albeit kept well under wrap right now)
5) Misc. Beos fans don't want to touch Gnome/KDE with a ten foot pole, and I know it would be way too tempting to port them for application compatibility purposes. Beos booted on my PII400 in 15 seconds, and was fast as hell. Would a BE/Linux combo keep Beos' vastly ease of use and configuration, or would it inherit Linux's most dreaded characteristics?
I'm rambling like a rabbit with the flu. But these are some valid concerns. Check out OpenBeos right now and sign up if you have the time and skills.
Wha??? EVERYONE cares about network transparency these days! (even if they don't realise that they do)
;-)
I hardly know any windows users nowadays that don't access their email through a web interface, so they can access it from any computer they happen to be on. Outlook Exchange is the next thing up, which true - is still a corporate thing at the moment.
But think - how long will it be before your wordprocessor is running on a remote machine and you just have a dumb terminal? What would be the advantage of this? Well it doesn't take much hardware to run a dumb terminal, compared to having to carry around a hard disk, cdrom, lots of memory, etc. etc. etc
And the extension to this is distributed computing - I mean true distributed computing, where your wordprocessor uses other peoples run time if it needs it, and theirs does likewise. Eventually you end up with the idea of one _massive_ computer, distributed around the world so that it never goes down in one go, which everyone connects to using dumb terminals.
Network transparency is the future, for these and for numerous other reasons (control your fridge from your computer! Yay!)
In the same way that "Only a geek sends text messages & emailswas 5/10 years ago, the same is happening now with network transparent computing.
We can't help it if we are ahead of fashion
Right now I am running XFree86-4.2.0 on FreeBSD-4.5. I have to interact with a Solaris-8 box. With X, I am able to run Clearcase, Rational Rose, Framemaker, etc. under FreeBSD. That's awesome.
If all you're doing is running some game under Wine under Linux on a standalone box in your dorm, then you don't really need X. But the rest of us appreciate the power and simplicity that is X.
A Government Is a Body of People, Usually Notably Ungoverned
The simplicy and power of typing setenv DISPLAY mickeymouse.window.manager:0.0
...having a fully functional computer (called X-terminal) being fully subutilized because the system was planned with a thin client in mind which ended up being a thick client?
Yup! Just try doing that on Windows and you'll see the simplicity of it. My login script sets my DISPLAY, so no matter which remote machine I login to, I can display locally. Really nice and completely effortless.
Or the simplicity of the most retarded cut and paste model yet developed?
Cut and paste is a policy. By rights, it shouldn't even *be* in X. It belongs in Motif, GTK+, Qt, etc. A policy-less GUI has its disadvantages, but the advantages outweigh them.
Or the power to drag and drop between almost no applications?
Don't blame X. Blame GTK+, Qt, Motif, etc. A policy-less GUI means it won't impose a standard on you. Because of this, a KDE program can drag and drop from my FreeBSD box to a program running remotely on my Solaris box. Otherwise there would have to be some standard out there with enough teeth for Sun to adhere to. A standard with that much teeth in it is detrimental to my freedom.
As it is now, it doesn't matter which which X server is running, my X client is fine.
Oh! If only I could subutilize Windows the same way! If only I could push that everpresent IExplorer running in the background off to another machine... If only I could compile in the background without dragging my MystIII down...
A subutilized CPU is a wonderful thing! Well worth the money. But if you want a fully utilized CPU, the answer is simple. Just downgrade.
A Government Is a Body of People, Usually Notably Ungoverned
The X extentions are so damn complicated that no ones using them, KDE isnt using Xrender, Gnome isnt using Xrender, hell even Enlightenment isnt using Xrender, the only people who seem to be able to make Xrender work, are the programmers working on it.
This only goes to advertise to the world exactly how little you know about X and how little attention should be paid to your misinformed rants about it.
NO where else have I seen alpha channeling in linux than from keith packard the creator of the Xrender extention.
Open your eyes then. It's everywhere, certainly all over my desktop anyway. If you want to live in the past, feel free. If you want to ignore it, feel free. Spreading misinformed, baseless FUD about one of the most significant modernizations to have happened to X in its entire lifespan isn't appreciated however.
What good is an overly complicated undocumented hard to use API on top of a bloated badly designed implementation of X?
The Render extension is a sensible, well-thought out solution to many of X's previous shortcomings. It's not perfect, but then - it's not finished yet. As for documentation, what do you need exactly? The wire protocol for Render is pretty comprehensively documented, and if you're merely trying to use it in an Xlib program, well, there's always the source code to look at. Yes, that's not perfect, but Render is the work of just Keith, and XFree86 is short-staffed enough as it is. Again - it's not finished yet! In any case, it seems that's enough for Trolltech and the GTK+ developers...
I'm also going to argue here that XFree86 isn't bloated and neither is it badly designed. What it is is massively short of good developers, especially those that are interested in working on the internals of the X server as opposed to just getting the latest and greatest features of their new graphics card working. It's an engineering project as big as the Linux kernel or KDE or GNOME, but with only about 5 people working on the core parts. Is it any wonder it develops more slowly?
Of course, if you were that concerned about X, I'd suggest you go and start hacking code for it, because that's the only way it's going to get better. Except, you're not a graphics programmer are you? Because if you were, you wouldn't have made such baseless allegations about X, and certainly wouldn't have made such basic factual errors as you did in your post.