Submitted by With many recent stability fixes and other improvements by Michael Lotz (mmlr) as well as others – he was able to finally nail down a couple last minor tweaks that allowed him to checkout the Haiku source from the SVN repository, compile a raw Haiku image, and test it in QEMU entirely from his Haiku install. This is the first time ever that Haiku has reportedly ‘self-hosted’, an unofficial important requirement for an alpha release. Please note that there are a few technicalities to be ironed out before the process can be easily reproduced by all. Update: Please note that Haiku won’t be taking over the world just yet.
There is probably no greater stress test than to retrieve the entire source code repository, compile and build it from scratch, and then run it from within an emulator, all from a single session. This test tests a lot of components of the OS, and to have it actually work proves that the OS is actually stable enough to be usable.
So Haiku is officially usable. This is the most important milestone for a new OS.
Congratulations are in order. I’m off to pop open a bottle of the bubbly stuff.
The message linked could be misinterpreted – it was not from a single session (in fact, last I heard it takes several reboots just to download the entire source)
A followup message was posted to help clarify:
http://lists.berlios.de/pipermail/haiku-commits/2008-February/01124…
Edited 2008-02-12 08:56 UTC
Thanks for the clarity Urias.
This is great news for many of us nonetheless!
What a milestone for the project! Looking at the Alpha milestone on Trac (http://dev.haiku-os.org/query?status=new&status=assigned&status=reo…) there’s not a whole lot left to do before release; only a few bugs (some of which, I believe, are already for most intents and purposes resolved) and a number of enhancements which are all in varying stages of completion.
It’s going to be exciting to see what happens once they’ve got a bootable, installable CD image out there…
The ability to compile the OS inside itself and build a usable image will GREATLY speed development. Right now you have to use BeOS 5.0.3 or YellowTab to do serious Haiku development – if it can compile itself competantly, it’s only a hop and a skip to being able to do ALL Haiku development inside Haiku itself.
Now they can finish taking over the world! Ha ha ha ha ha!
Are you quoting Day of the Tentacle by any chance? 🙂
Yeah!
Glad somebody still remembers that awesome game.
http://arstechnica.com/news.ars/post/20080212-first-look-haiku-poet…
Both this OSNews entry and the Ars article fail to quote the Haiku project web site, anybody’s got the URL handy?
Rehdon
http://www.haiku-os.org/
There ya go.
Edited 2008-02-12 14:21 UTC
Thank you.
Rehdon
Asking a honest question here: What is the big deal about Haiku/BeOS? I am young enough and new enough to OS jumping that I don’t remember BeOS (OS/2 for that matter). I fully understand not liking or even hating Windows *cough* Vista sucks *cough cough*, but there are other alternative OS’es out there already. What does Haiku offer that is not already available in at least on a Linux distro? The UI doesn’t appear to be ground breaking, that task thingy in the upper right corner actually seems to be a step backwards to me.
Are there things that Haiku/BeOS can do that no other OS can do, or does it do things more efficiently? Would the work being done to develop Haiku be better spend developing a DE for existing Linux/UNIX base? Are there politics involved? Is there a site that can compare Haiku to other OS’es?
http://www.faqs.org/docs/artu/ch03s02.html
”
Be, Inc. was founded in 1989 as a hardware vendor, building pioneering multiprocessing machines around the PowerPC chip. BeOS was Be’s attempt to add value to the hardware by inventing a new, network-ready operating system model incorporating the lessons of both Unix and the MacOS family, without being either. The result was a tasteful, clean, and exciting design with excellent performance in its chosen role as a multimedia platform.
BeOS’s unifying ideas were ‘pervasive threading’, multimedia flows, and the file system as database. BeOS was designed to minimize latency in the kernel, making it well-suited for processing large volumes of data such as audio and video streams in real time. BeOS ‘threads’ were actually lightweight processes in Unix terminology, since they supported thread-local storage and therefore did not necessarily share all address spaces. IPC via shared memory was fast and efficient.”
Edited 2008-02-12 15:09 UTC
It’s more of a political and technical statement.
Those who used BeOS found it, at the time, to be much more advanced in design and construction than they did the older Mac OSes or Windows.
Unfortunately, since the maker of that OS went out of business, the aforementioned OS lines were mostly replaced with better-constructed OSes. But those who still felt that the newer OSes were lacking in comparison to what BeOS could’ve become decided to start on their own (open source) revival project, which has increasingly become a de-facto successor to the original, closed-source BeOS.
However, they had to create this successor using archaic tools (such as a much-older version of the GNU Toolchain) in order to maintain compatibility with the last closed-source version. Because of this, they have had to work on this for around the last 5 years, and many programmers have left the project over that time.
After getting to what is called “Alpha R1”, these folks have intentions to make newer versions which will be FAR more up-to-date, hopefully in a way that will compete with the likes of today’s desktop OSes.
But that would mean breaking compatibility with the older OS, which is probably a good thing in the long run. They don’t know what the next version will look like, and there are many influences from other OSes (Linux, Mac OS X, Windows Vista), but they may retain a number of distinctive features that will sufficiently separate Haiku from the others.
If anything, this may benefit other desktop-oriented Free Software projects with new ideas.
You basically answered your own question.
Yes, there are several things about Haiku that make it very different from the currently available OSs out there.
Anyway, this comes up all the time and people seem to not get it at all. Saying that we would be better doing something else (like helping other OS projects) misses the point by several miles. We do this, work on Haiku, because that’s what we want to do and because we believe this is the best possible course of action. Of course anyone is free to disagree but it doesn’t change this simple fact. Try to tell a beer guy that wine is better and you will see what I mean.
What’s the big deal? It’s quite simple: it’s better than anything else out there, OS-wise. The ease-of-use sensibilities of OS X with the software freedom of Linux. Don’t get me wrong about Linux (typing this from Xubuntu), but the problem with existing OSes out there are:
XP / Vista: it’s Microsoft. ‘Nuff said.
OS X: Mega-expensive, limited by hardware
Linux: too complicated for regular people (particularly setup, permissions, and support), not very commercial-friendly
No, the applications aren’t there — yet — but that’s because OSes and applications are a chicken-and-egg problem.
The technologies that are going into Haiku aren’t so groundbreaking, but how they are used in a cohesive package that just works for both novice and advanced users is what sets it apart.
‘What’s the big deal? It’s quite simple: it’s better than anything else out there, OS-wise. The ease-of-use sensibilities of OS X with the software freedom of Linux’
All of that could apply to RISC OS. It was also amazing and ahead of its time
‘No, the applications aren’t there — yet –‘
This was a problem witjh RISC OS… and the apps never got ‘there’ before it died. The same will happen to Haiku. Haiku will never have a user base bigger than AmigaOS does now, the apps will never get ‘there’, neither will H.
There’s no big deal about it and there never will be.
Edited 2008-02-12 17:45 UTC
As long as it remains open source, and there are fanatic people who are supporting it – I don’t see how your logic applies.
Nobody can “kill” Haiku except its own users/developers/supporters…
‘Nobody can “kill” Haiku except its own users/developers/supporters…’
…or lack there of.
Well yes, exactly – what did you think I meant?
7 years after Be Inc. closed, 7 years since OpenBeOS then Haiku is in development, users/developers/supporters are still there when – by any measure – they had all reasons to lost faith and interest and had many motives to moved on without return.
But they didn’t.
Maybe they have a reason for that…
well we’ll see if thats the case. I have used osx, beos, windows, and linux, and when haiku becomes stable I will switch in a heartbeat (though I sadly will keep a partition available for audio production until it catches up in that regard).
BeOS wasn’t just “advanced for its time”, there is still nothing else like it and when haiku is released, even as r1, in my honest opinion, it will still be the most elegant desktop experience there is.
And good applications, owing to its awesome api (they also are very forward-looking by supporting java, hopefully they can integrate it well with the BeAPI), are also destined to come once the platform becomes readily available.
The apps, in many cases, don’t need to “get there”. We have a full Mozilla Suite port (and all the associated applications as a result), we have streaming media and DVD players, we have a massive range of applications RISC OS never had and likely never will have – comparing apples and oranges here.
‘comparing apples and oranges here.’
Fail to see the logic here. Are you trying to say (for some strange reason) that apples and oranges can’t be compared? I could compare them for you if you like.
Maybe I’ll write an article about dead horses (Fitz Law and ‘apples and oranges’) for OsNews. I’ll call it ‘An Orange… and why I won’t be getting one.’
Or you could continue trolling with pedantic, disingenuous non sequiturs. That seems to be working well for you – and it’s a whole lot easier than actually replying to any specific points.
When I was a BeOS user it got very close to having all the apps I needed. I only booted into Windows to use Photoshop, the rest of the time was in BeOS. There was quite a bit of software for it and a lot of that will still run. With better POSIX compatibility porting apps will be easier and there’s plenty of open source apps out there these days.
Someone probably said that about Linux once…
MacOS X is “mega-expensive”?
Bwa-ha-ha-ha-ha! Ok, let’s do some simple math:
MacOS X 10.5 (Full maximum edition; which for MacOS X is no problem, since there is only ONE version) = $129
Windows Vista Premium Maximum Edition (whatever it’s called… it’s the one in the black box) = $400-$500.
These are both the full retail, non-upgrade versions of each OS.
Now, pray tell, WHICH version is “mega expensive”?
Still, $129 is a lot more expensive than $0, don’t you agree?
$129 more expensive than $0? Well… Duh. Isn’t even $0.01 more expensive than $0.00? Simple math. The point I was making, was based on the insane idea that MacOS X is somehow “super expensive” at $129, when the OS (Windows) right above it, costs anywhere form $250-$500 for the ful versions. It’s a statement of utter nonsense.
$129 or $500, they’re still more expensive than $0.
J-Ho’s point still stands, as we’re talking about Haiku (and, by extension of free software, GNU/Linux). Please elaborate.
Interesting, but a useless metric, since the vast majority don’t own a Mac to actually put Leopard onto. So, that would make OS X 599 (price of the cheapest Mac in the US). It’s worse in Europe; a new Mac Mini here is 599 EUR. Yes, that’s 874 USD.
So yes, OS X might be cheap for the few people who actually own one (although in the US that number is a lot higher than in Europe) – but it’s not for everyone else. Please keep that in mind.
But yes, Vista is overpriced. No doubt.
MacOS X *upgrade* price.
Before that, you must buy a Mac, only platform supported by and bundled with MacOS X. Apple computers are not well known to be particulary cheap. Their cheapest complete computer is over $1000, while for the price of their quite naked Mac Mini system you could have a complete PC.
MacOS X is a very good operating system, and one can argue that the whole user experience worth more, that ROI is better in the end, but when it comes to cost comparison and only cost, aka investment required, there is no debate: being a MacOS X user is expensive.
That’s Apple business model for a reason, you know.
Keep in mind that nothing that you read here will actually make you understand “what’s the big deal”… You actually have to use BeOS to understand its philosophy and how much modular it was…
…That said, let me try…
BeOS was build with C++ in mind. That means kernel, drivers, services, etc… so, everything is basically what you may understand as “dynamically linked”… everything should use the standard BeOS APIs to communicate, so every process just need to send “BeMessages” to each other for just about everything…
Also, everything was very well thread, so anyone with multi-processors machines at the time would gain as much as close to 100% more processor power for a dual processor compared to a single processor machine.
That’s because the original BeBox machine used processors called “Hobbit”, that are not as powerful as Desktop processors, but they was way more cheaper and that way they opted for 4 (IIRC) of these processors instead of one, and focused in a multi-threaded design for their Operation System.
Every (native) BeOS’ GUI app has at least two threads, one for the application itself, and the other for its window and widgets.
Also, a very interesting aspect of BeOS was the “Translators”, which are a system wide codec-like system that could make any BeOS (native) app support any system-installed filetype, encapsulator, codec, etc… you could theoretically write something in C and save it in Python if your system had these two translators… but in real word, that means the you don’t have to write GIF, JPEG, PNG, MP3, AAC, MPEG, AVC, etc to your application, you just need the proper translator installed in the OS (system-wide or user-wide, so you could provide one if the user doesn’t have…). Not having to re-write the abc was very interesting, and gave BeOS Developers a way to write very powerful applications that are very small, in code and binary, easy to maintain and a one man job and small teams could actually delivery very interesting applications.
By living in the connected and social world, where searching and tagging is very important, BeOS had an extraordinary filesystem that was a mix between a traditional filesystem and a database. Searching was VERY fast, I mean, almost real time in that 100/200MHz era… enough to the default mail client store e-mails actual folders and files… These ones could be searched in full text and fields by the standard search… available to all (native) BeOS applications… Want to search all e-mails sent in 1999 to a group of people you found in another search through your mail messages and find the ones you replied? what about saving the search and have a real-time updated results? Maybe your IM contacts could be saved as files in the filesystem too, and you could just drag and drop from your search results into your IM contact to send them these files…
What about searching into your music library for artist that has “ace” in the name, has albums from 1960 to 1980 and has been played in the last year? what about this selection be available like an ordinary folder to application in your system? That may be a common feature these days for spotlight, google desktop and similar, but steel not as powerful, fast and integrated in the system…
I could go on with replicators, and other features… but you could just search for the BeOS Bible in google and have a lot of fun…
Still… I can’t explain in words about the whole BeOS experience enough that would substitute the actual use of the system…
…and let’s wait the taking over of the world!
An excellent reply! This does sound interesting and I can understand this. It reminds me of Studebaker or Packard automobile fans. Ahead of their time, but were beat out by ,technically speaking, inferior products.
Well when Haiku finally releases beta quality or RC quality, I may check it out.
My favourite recent example of why I still use BeOS and find it (and Haiku) compelling:
Not long ago I was helping to setup the computer systems for a small company – they have multiple physical locations, so they wanted me to create & setup MSN messenger accounts for each employee. While doing the setup, I added the new accounts to my contact list in Pidgin on my laptop – meaning that those users were now on my server-side MSN contact list.
So when I got home and started up the im_server (a multi-protocol IM client that uses BeOS’ central contact files for storing data), the new contacts were downloaded – which created corresponding “person” files in the process. The next day, I started to write an EMail to one of those people in BeMail, and got a weird surprise: BeMail auto-completed the EMail address, even though I hadn’t added it to address book entry (or communicated with that person by EMail before).
Turns out that it was because I had created the new MSN mess accounts as passport accounts – meaning that the usernames were the users’ actual EMail addresses, rather than [email protected]. So when the im_server downloaded my new contact list from the server, it was smart enough to add the username to the EMail address attribute for the “person” file it created. And since BeMail reads address book info from the same files, it was able to auto-complete the address.
And that’s all done by making use of functionality built into BeOS itself, without the need for add-on search/indexing tools (since person files store information in BFS attributes, which are already indexed & can be queried for).
It can be moved easily enough to any of the other three corners, or stretched along the top / bottom of the screen like a normal taskbar (by dragging it by the “grippy dots”). Some people like having it the top-right, I personally prefer to have it shrunk in the bottom-left.
Lotz states that he uses the compiler and flex from R5. As he admits, it’s only “pretty close to self-hosting.”
R5 binary compatibility is already pretty good… you can use R5’s firefox build with some issues and netpositive already works as well as under R5 for me. Still, it’s not trivial that Lotz got the build system to chug out a working Haiku image under Haiku by dropping in the R5 gcc and flex (with some hacking apparently). I’ve been amazed with how consistently(and quickly) the build system makes the entire Haiku tree under both R5 and Linux… with multiple versions of gcc no less! The fact that such cross platform building is already functioning at such a high level bodes very well for the future portability of the Haiku system IMHO.
I would say it’s a matter of opinion – everything used to “self-host” haiku at this point is open source and re-creatable. The GCC used from R5 is the same GCC in the Haiku repository (updated and maintained by the Oliver Tappe and the Haiku devs), it was simply compiled on R5 first and then copied to Haiku taking advantage of Haiku’s binary compatibility…
The headline I submitted was slightly less misleading, and longer – so I understand why it was shortened the way it was
Ah! So it’s your fault that I can’t access Haiku-os.org tonight then? I think we suffered the /. effect
typo edited
Edited 2008-02-12 17:31 UTC
Actually, I think that’s a coincidence… the Ars Technica article (and the Slashdot article that linked it) was not related to this self-hosting news…
Just good (bad?) timing I guess
Edited 2008-02-12 17:40 UTC
I ran BeOS R5 years ago and I started using Linux when BeOS when out of business. I can’t wait to install Haiku OS on my laptop again. I have been waiting for this moment for years.
Honestly, all they need are new apps. Give me a good browser and a descent Office program and I would be more than happy to switch to Haiku.
-2501
I don’t know what the excitement is about porting Java to Haiku. You may as well say, “me and a couple of buddies of mine are going to build a 15,000 sq. ft. house at the top of one of the Smokey Mountains with one pickup truck to haul supplies but we can only do it on the weekends if our wives and kids don’t obligate us to do other things”. It’s a monumental effort.
They could port Mono with 1/10th the effort it’s going to take to get that Java monster running 100% and then with two very nice front-end development tools available, SharpDevelop and MonoDevelop, you’d get the application programmer productivity level higher and a lot more people could be contributing worthwhile desktop applications faster.
The same people got Java extensively ported to BeOS R5 – it nowhere near as complicated as you seem to assume. Theres screenshots of relatively complex Java applications running under BeOS R5. Which, as it goes, is a harder OS to port things to due to worse POSIX compliance and generally being older.
There is also a video of Java running in BeOS:
http://youtube.com/watch?v=yVwc-ZSRMwQ
Enjoy!
And also because we can’t fix bugs in BeOS. Firefox for instance has a lot of ‘fixes’ to get around BeOS bugs. The most interesting one would be the way to get around loading libraries. (BeOS only had 32MB reserved memory for ALL apps that loaded ‘addon’-libraries).