Haiku Project | Hacker News Reader

Haiku Project

haiku-os.org

147 pointsjordansmithnz64 comments

It seems that this project is a free and open-source inspiration of the BeOS rather than aiming to be a complete reimplementation of it. Just like what Linux was to UN*X and ReactOS was to Windows. But, judging from its recent progress: package management repos, 64 bit support and the recent software ports at ‘HaikuPorts[0]’, it seems to be already more advanced than BeOS was and already has a lot more software since its last official alpha release (5 years ago.)

But yeah a pretty interesting OS, but I’d wait for it’s so called ’beta’ when ever it arrives. I’d like to try this on a Raspberry Pi or another ARM board soon. If only it attracted more developers though.

PS. (Its actually 'Haiku' not 'Haiku OS' ;) )

[0] https://github.com/haikuports

(I'm one of the developers of Haiku -- I manage the website and am one of the leads at HaikuPorts in addition to committing to the main source tree.)

The ARM port is stalled due to a lack of developer time, and has been for over a year now... The kernel is mostly ported and initializes the framebuffer but then stalls when it gets to HID and storage initialization because the USB stack isn't ported yet (iirc). Patches wanted... but we won't have much time to spend on it ourselves anytime soon.

Yes, the beta release should be coming "real soon now". We're down to 2 blockers, both of which should not be that complicated to fix.

And yes, it is indeed "Haiku" not "Haiku OS". :)

Is there any documentation on what the current status of the ARM port is?

I know there is https://github.com/haiku/haiku/tree/master/docs/develop/port... but I think it's out of date at this point. Check the commit logs for the "arm" directories in the kernel, and then ask the developers responsible for that on IRC (Freenode#haiku).

I don't know much about the ARM port in specific, but I'm very familiar with our build system and development model in general, so if you have questions about that I can help there.

I think it started out as trying to perfectly clone BeOS, but then in the ensuing years that became less of a priority and they started improving things in ways that could even go so far as to break compatibility.

Breaking compatibility is mandatory; otherwise Haiku would be stuck with its C++98 API. Trying to interface otherwise-modern code with the vintage BeOS API is doable but does not exactly lead to rising property values in Tokyo.

Well, we are still stuck with a C++98 API... some parts of the x86_64 core code in the kernel, since that's only ever compiled with GCC5+ use C++11, but by and large we're still C++98.

(The Be API used almost none of the STL, so it doesn't really make a massive amount of difference to us. <functional> would be of use in event handling code, as well as a few other new language keywords and the like, but by and large we'd be unaffected.)

I tried running the nightly on virtualbox, got through the partitioning and then it just froze...

FWIW the following nightly worked.

A fitting link to follow up the Apple / Next merger article that's gotten so much traffic today. At the time I really thought Apple was making the wrong move. Who knows what BeOS would have been today had Apple picked it up, but the wisdom of going with something *nix-based seems to have been a great decision.

Or if Microsoft didn't actively attempt to destroy it.

MS got hit with lawsuits because of Netscape, but their targeted destruction of BeOS was so much worse considering, at the time, BeOS was such a better system that just couldn't fight back.

This. Wasn't Dell actively installing BeOS on systems but due to Windows licensing restrictions couldn't make that apparent to end users?

There's (perhaps apocryphal) stories that Gates was terrified of what a Unix with a decent UI might do to upend Windows.

It’s weird that Linux won as a platform on mobile and it was partially due to Apple making it so that phone carriers had less influence on phones so the carriers needed something to claw it back with .

Linux 'winning' on mobile has nothing to do with Apple doing anything at all. It 'won' mostly because it was chosen as the base for Android as well as several other mobile platforms (Intel's Moblin, Nokia's Maemo, the combined offspring called Meego, Samsung's Tizen, etc). Android 'won' for the same reason that Windows 'won', it was available for all to use in more or less any way they saw fit. The carriers had nothing to do with Android 'winning', that is mostly due to the availability of affordable unlocked devices from a multitude of sources, a market which Apple can not and does not want to touch as the profit margins and opportunities for lock-in are too low.

If so, that would have been NeXTStep. BeOS was (mostly?) POSIX compatible, but was less of a Unix than Linux.

John Perry Barlow described it: "I have a NeXT machine -- I expect a boo or two -- that's as close to UNIX as I've been able to get, and that's kind of like UNIX with training wheels by Armani." http://shop.oreilly.com/product/9781565929920.do

Yeah, but Apple solved the NeXTStep problem by buying NeXT and discontinuing Next step for commodity hardware. While it revived Apple, it also neutralized any serious threat NeXTStep might have posed to Microsoft.

My argument is that if Gates was "terrified of what a Unix with a decent UI might do to upend Windows" then NeXTStep must have been more terrifying than BeOS.

In the 1990s I worked in places where NeXT machines were in frequent use, including as the main desktop machine for a number of people. We also had NexT machines available for making presentations.

There was never anything like that for BeOS. And FWIW, I was a registered BeOS developer.

Remember too that Apple's A/UX was also a "Unix with a decent UI".

I remember that Microsoft described Windows as “Windows IS POSIX,” when discussing other operating systems that could be competitors… Windows had a POSIX layer back in the day, similar to the win32 or NT layers, but for federal sales.

Of course, the POSIX subsystem was replaced with Windows Subsystem for UNIX, which was too scary of an idea to ever actually use.

And, Microsoft has now come full-circle back to having a bash shell. This bash shell does far less, though, not interacting with the rest of the system...

> This bash shell does far less, though, not interacting with the rest of the system...

If you mean WSL, then it actually does (the name "Bash for Windows" is really a misnomer). Not only you get full access to the filesystem, but you can e.g. pipe output of a Win32 binary into a Unix one. It's really much more powerful than WSU.

Has WSL been “upgraded” since it was first introduced? When I first looked at it, you needed X forwarding to get a gui and couldn’t launch a windows executable, for example.

It's been steadily and continuously improved. Some of the latest improvements still require an insider build but win32 process interaction just works now (in my experience).

That's very good.

You still need to jump through hoops with GUI. I don't think that's going to change anytime soon, since their focus is mainly on developer command line scenarios. Still, when you need UI, XRDP or Xming work adequately.

Can you me more about the "targeted destruction" of this operating system by Microsoft? I worked there at the time and don't remember anything like that.

Here's an old article on it: https://web.archive.org/web/20080420193929/https://www.there...

And another old document on it: https://www.jumpjet.info/Application-Software/BeOS/msft_comp...

In essence, Microsoft used its market position to pressure PC makers to not sell PCs which allow dual booting into a second preinstalled OS. Back then, it seemed that Be Inc finally got a breakthrough in the market, with Hitachi interested in shipping it preinstalled next to Windows, with a boot manager to select which one you want on every start. This might have finally gotten BeOS the wider exposure to users which it needed to become disruptive (most people will not install an aftermarket OS, they will just use whatever comes with the computer).

OTOH, startups need to factor anticompetitive behavior into their planning; whining about being bullied didn't help Netscape or Be.

I remember thinking it was a terrible decision at the time too. It's hard to even imagine how different the world would look today if they'd gone for BeOS; I think Apple would almost certainly have died or been acquired, and we'd all probably still be using a mix of feature phones and whatever craziness Blackberry was up to.

(BeOS was really sweet though! So damn snappy. Even on way way way more powerful hardware today, macOS is still not generally snappy. It's a shame.)

Haiku is still "damn snappy", whether on a 15-year-old Athlon XP or the latest Core i7 (neither of which BeOS will even boot on). You should give it a try sometime :-)

The whole "Apple needed to buy an OS to be saved" thing was always hogwash -- they bought NeXT under the auspices of "just" porting OpenStep to PowerPC and Mac-ifying it ... but in reality that took them 5 years to accomplish. I'm pretty sure that in 5 years pre-Jobs Apple could have shipped the Copland project, which had reached Developer Release status just prior to Jobs coming on board and its cancellation.

I don't think NeXT brought anything tech wise to Apple that Apple couldn't have done already. In reality it was a Steve Jobs/Avi Tevanian coup that led to the mass layoff of a whole lot of people and the garbage binning of a lot of good tech (Newton, OpenDoc, Dylan just to name a few).

In any case if not for the iPod/iPhone/iTunes jump, Apple would probably be dead by now. They were in trouble until the revenue and brand cachet from that started to roll in.

What I mean to say by this is: Apple needed neither BeOS or NeXTstep. They needed urgently management discipline and to ship the decent tech they already had. Jobs came on board and provided discipline, killed a bunch of expenses, and kept a lifeline going through the iMac sales, but wasted that money on the OS X transition when they probably could have just finished Copland in that time. But hey, I wasn't there, so...

Yeah, definitely, the focus that Jobs' provided was far more important than the tech.

I doubt Copland was going anywhere, though, no matter how much time they gave it. The underlying tech of NextStep was just so much better than the classic macOS assembly/pascal, cooperative-multi-tasking, single-memory-sapce, GetNextEvent() universe. Just too hard a problem to try to modernize while still so closely tied to MacOS in your mind...they needed to make a clean break.

My understanding after reading a bunch of articles (now I'm an expert!) is that what became 'Carbon' in OSX was based in part on work done for Copland. And that API maintained reasonable good classic compatibility (at the source / API level) while modernizing it a whole bunch. And it shipped quite quickly after the coup, so it must have been based on something existing prior to it.

But such was the politics of Apple under Jobs/Tevanian that the (quite esoteric by industry standards) Cocoa/Objective-C ecosystem was considered superior and Carbon was actively deprecated despite being quite successful with third party developers...

Sure, Obj-C was esoteric, but also well-designed and impressively ahead of its time. Witness that a minor evolution of the same tech went on to form the foundation of iOS.

Even now, with Apple trying to transition to Swift, it's unclear to me that's an overall win. The combination of C plus message-passing was inspired.

So basically I agree Cocoa was better tech than Carbon. Carbon was popular with devs 'cuz that's what they were used to, and that's what their existing apps were written in...it was a great stop-gap measure, but wouldn't have provided nearly as good a foundation for the future (in my opinion).

I dunno, after doing iOS development for two years, I was very unimpressed with ecosystem. The UI toolkit is actually surprisingly low level for what it is.

But it has served Apple well overall I suppose. I think ObjC is a terrible language tho.

Android was started to compete with Microsoft in mobile. It still would have existed and likely still would have attained dominant market share.

What I loved about BeOS was the multithreaded user interface. Imagine and office with a single computer with dozens of gigs of RAM and 128 cores could do where everyone could collaborate on the same screen (50 feet wide) at once. Kinda like SunRay on steroids.

Yes! I've been thinking about this occasionally. I have decided that when I get time I will build on FreeBSD (my favorite operating system) and make a shared screen multi user graphical desktop that supports multiple keyboards and mouses at the same time, and there will be a permission system so that people are only allowed to interact or obscure the windows of others if they get their permission.

But there's only one GUI thread. It wouldn't work.

I'll write my own GUI layer from scratch.

That sounds like a godawful nightmare.

Until you personally need 64 cores for a particular calculation while everyone else is just checking their emails...

sounds like a mainframe

sounds like warehouse scale computing, imo.

If this ever gets past Beta someone has to recreate the original BeBox to go with it, perhaps with a dual core CPU with each core's activity (?) mapped to the LED cpu activity indicators on the front of the box. I still miss mine ;)

https://en.wikipedia.org/wiki/BeBox

There was some project which was going to do that, using the MinnowBoard Max (x86-based): https://twitter.com/20thABBx -- but we haven't heard from them in over a year now :(

Man, I’m not sure BeOS was ever on the same technical footing as NeXT, but those BeBox machines were magical. I remember getting to play with one a little—an Internet acquaintance I met through IRC had one, was home on college break, and my fragile little mind was totally blown.

I remember watching the fractal generator render the Mandelbrot set and top out both CPU LED bars. In my naïveté I figured that software everywhere would soon be making use of multiple processors, increasing performance linearly, and with no effort.

I guess we all have lots of cores now, but writing software to use them all is still a PIA.

Those felt like magical times, with one juggernaut (Windows) and a sense that any of the new computing platforms could upend everything. Nowadays, I feel less of that excitement in spite of being able to credibly choose between Linux, Mac, Windows. Maybe that’s more a function of my age than anything else. Or maybe the hegemony of mediocre web apps has just delivered us into another period of bland sameness.

Imagine if BeOS had been ported to mobile devices. What would it look like? How would it have run? Not so insane, since this was attempted:

https://en.wikipedia.org/wiki/BeIA

And also this:

https://mondaynote.com/the-operating-system-fountain-of-yout...

Well, there was a bit of hope. Palm bought BeOS, spun it off with PalmSource which all eventually got bought by ACCESS. It was hoped Palm would use BeOS as the basis for new operating systems but it never quite happened with the failed Palm OS Cobalt. Palm, going its own way, eventually created WebOS which is now owned by LG.

Some of us spent $2,000usd on a BeBox :-) I learned so much about c++, SMP, threading, etc.

Wasn't Zeta a BeOS?

Here in Germany we had much fuzz about it, because it was sold via TV commercials.

Somehow noone was really sure if they had an acutal BeOS licence and really access to the source, eventually it died.

"Zeta was a commercially available operating system based on the BeOS R5.1 codebase." - https://en.wikipedia.org/wiki/BeOS

Zeta was illegal!

magnussoft/yellowTAB never had a license but distributed a rebranded BeOS 5. After a years long drama they eventually ceased operation.

https://en.wikipedia.org/wiki/Magnussoft_ZETA#Criticism

Excerpt from Wikipedia article:

  "Zeta had no legal rights to distribute the BeOS software, 
  much less open source any of it, as the rumours had it.
  [...]
  The main reason that there hasn't been a public statement 
  previously is that dealing with this matter legally, in 
  Germany, is an expensive undertaking and--given the 
  apparently small amount of funding behind Zeta in its 
  various incarnations--we'd only be in a position to spend 
  significant money and legal time to make a point.
  [...]
  We have sent "cease and desist" letters to YellowTab on a 
  number of occasions, which have been uniformly ignored. If 
  Herr Korz [Bernd Korz, Former CEO of yellowTAB] feels that 
  he holds a legitimate license to the BeOS code he's been 
  using, we're completely unaware of it, and I'd be 
  fascinated to see him produce any substantiation for that 
  claim.”
  — David Schlesinger, Director, Open Source Technologies. 
  ACCESS Co., Ltd., Access 'Completely Unaware' of Legitimate 
  Zeta License

I was going to reference the "is computer on fire" and "is computer turned on" API calls, but couldn't find them in the documentation anymore :s All that's left seems to be this page: https://www.haiku-os.org/is_computer_on_fire/

Oh don't worry, they're still around: https://github.com/haiku/haiku/blob/master/headers/os/kernel... :)

Our own API documentation is still a massive work in progress. The old Be Book should still have them documented, though.

A certain Be executive once quipped that "we don't engage in necrophilia." And yet, here we are... ;-)

It still gives me a grin to see this project plugging forward, funky 1990s-flavored C++ programming kits and all. :-)

Haiku isn't Be, we're free to be necrophiliacs. :P

I couldn’t have helped but want to rename the NS prefix from all over the OS APIs (NSString, NSUrl, NSWhatever, etc.) but it feels good to me that it’s still there. Almost as if no software is ever perfect...

Is this still considered alpha? Last I tried this on a dell laptop it would crash everytime I tried something really banal, I can't remember what. Like visit Google with the browser.

We're very near "beta", at least in the nightly builds The browser is much more polished (it plays YouTube out of the box, etc.) And kernel crashes are at least somewhat rare.

thanks!

How does this compare to Linux performance wise?

What’s the advantage to Haiku?

I remember installing this in a VirtualBox around 2008 - cool to see it's still going! I had a friend in my CS program who was really into BeOS.

In 1999 at a convention in New Orleans, a friend of mine had BeOS 3 or 4 on an x86 and it absolutely was amazing. It booted way faster then Linux or Windows and was quite pretty. A lot of us reminisce for what could have happened with the MacOS. I love OSX. I use it every day and I'm quite happy with it. Still I wonder what the present would be like with a BeOS based system instead.

I bought a Performa 6360 (the lowest-end Mac that could run the BeOS at the time) so I could run the BeOS. Even the installer blew me away (you could move the dialog around while it was installing, and it was updating live on the screen). I was like, "You can do that with this hardware?!" This was 1997. I really loved that OS.

Reply on news.ycombinator.com