It's nice to see Haiku making such good progress -- I feel that the existence of a user-friendly open source OS that's not a UNIX or Windows derivative is a good thing in itself, if only because it may be keeping alive some ideas that don't fit in the legacy-oriented Linux mold.

Maybe Haiku could become a nice netbook OS? Its single-user media focus certainly fits the bill. (Although Internet appliance computers were Be, Inc.'s swan song marketing attempt as well, so maybe they don't want to reopen that wound...)

I do think that Haiku needs some kind of forward-looking showcase project with a bit of gloss. Right now the superficial impression is that they've mostly catched up to a decade-old commercial project. Consider these screenshots of the first and latest release of this system...

BeOS Developer Release 4, from 1994: http://osnews.com/img/20869/screen2.gif

Haiku, from 2009: http://osnews.com/img/20951/1.png

(I prefer the 1994 version actually -- those tight pixel fonts and prominent grids of awesome 45°/30° icons must have something to do with it... The old light-grey look actually looks fairly contemporary alongside the Leopard and Windows 7 aesthetic.)

I second netbook OS suggestion. They should port Haiku to popular netbooks e.g. acer aspire one, asus eeepc, MSI wind, so that anyone could run it without any problems.

This way there is no hardware hell(at least not that big) and you get usable system for the masses.

"the superficial impression is that they've mostly catched up to a decade-old commercial project"

I think that is pretty much where they are at. The hope is that it provides a solid base upon which to build.

Out of general curiosity, why doesn't a OS like this focus on being a VM-only OS? If they went this route, they wouldn't need to worry about driver hell, right?

And god knows I'm not going to give up my core OS anytime soon for a toy OS. But I'd love to play with it inside my VM!

Seems like an easy decision. What am I overlooking?

Because its developers want to use it as an every-day operating system without the overhead of another. They don't see it as a "toy OS" but rather something actually useful that they want to use on a day-to-day basis.

It's worth adding they already provide drivers for at least VMWare and possibly other virtual machines as well as VMWare images so it's not as though they're treating virtual machines as second-class citizens.

Couldn't there be created a hypervisor OS that translated diverse drivers to a single, unified API, and did nothing else (basically running the VM, in kernel space, without scheduling.) I could see Linux stripped down to perform this task--having all computers appear to all operating systems as the exact same hardware might push some distinct advantages along with it. For example, a network is a network is a network; it would be the hypervisor worrying about if it was a wi-fi one or a cell modem or whatever, and then providing a clean API for any additional features one might have over the other (connecting to APs for wifi.) Actually, picturing this further, it would be perfect if the hypervisor exported all its APIs over 9P. Any extra features would be immediately discoverable.

While this is an interesting idea, and definitely possible, I'm left wondering if it would be desired. By that I mean do we want all networks to behave in the same way, do we want all devices to be represented by their lowest common denominator?

Essentially creating this specific hypervisor OS would create a lot of places for leaky abstractions, and you'd end up creating a lot more special case code to detect and fix those abstractions then you would if you just coded the drivers.

I once suggested a virtual MIDI interface in WinUAE (the Amiga emulator), after having implemented one in a C64 emulator (Vice). This was mainly for Bars & Pipes (by Todor Fay, who worked on DirectX). They did it, too!

"...if the hypervisor exported all its APIs over 9P."

Please sign me up for your newsletter. I need to find more excuses to use 9P. Wmii got me hooked.

As a general thing for toy OSs it's a good idea, but a big part of the sell of BeOS was high performance media and filesystem, and the benefits of these would be reduced as a VM OS.

Haiku is a seriously impressive piece of work.

The reasoning to the "eternal pre-alpha" makes sense in my eyes since it lowers expectations. Criticism over lacking basic features (like SATA issues, CD-booting, partitioning, printing or even Wi-Fi), usability or stability issues (such as unstable filesystem drivers) can easily be dismissed without much crticism, people declaring it "useless" or a "toy OS". An alpha that's actually usable and can be used to attract developers is a good thing IMO and makes it a more significant milestone. It's much better than the alternative of a premature launch.

Also, a fun quote from an article by a different OSNews author in 2004 (http://www.osnews.com/story/8114/Three_Years_of_Haiku_The_Lo...): 'And if Haiku is "ready" by 2010, no one will care.'

I discovered BeOS about the same time I started getting into UNIX. I remember being impressed by the graphics and performance (the knoppix-like demo cd was ahead of its time). One thing that always perplexed me was that despite having a POSIX/GNU/whatever interface, they stripped out the multiuser abilities. This seemed like a tragic oversight, particularly as the need for multiple users and security tiering (administrator rights) was taking off.

It's not obvious to me what the Haiku team is doing about security in this regard, but if the same intellectual effort and creativity used to create the BeBox and BeOS was used to address spyware/malware/other desktop security concerns, it could be awesome indeed.

What if Apple had bought BeOS?

Now that is a major "what if?". For, not only did Apple actually go on to buy NeXT, but in doing so they brought Steve Jobs back into the company. I.e. we might have had Mac-BeOS and no Steve Jobs: what would have happened?

Instead of Steve Jobs, Apple would have acquired Jean-Louis Gassée... But he probably wouldn't have been in a position to take over the CEO job, not to mention being able to replace many of the important executives with those from the acquired company.

More fundamentally: Gassée probably wouldn't have been able to reinvent Apple's engineering values and sprawling product lineup the way Jobs did, simply because he had been so heavily involved in the first place in what Apple had become since Jobs's departure.

(AFAIK, Gassée was in charge of the infamous Macintosh Portable project: http://en.wikipedia.org/wiki/Macintosh_Portable ... So it's clear that he doesn't have Jobs's nearly infallible design taste for compact computers and lustworthy gadgets :))

I remember seeing BeOS in late 1999 for the first time and thinking "wow, that's really pretty".

Compared to Windows and Linux of the day it was. Now? Not so much.

Pretty could mean a lot of things.

Like pervasive multithreading, a 64-bit journaling file system. a kernel written in C++... beauty is relative.

BeOS was (is?) the best OS ever. I feel bad for not having contributed anything to the Haiku project but for the prayers to the non deity agnostics like me don't believe in.

I make it a habit to test out a Haiku build every couple of months. The improvements and progress they've made is really incredible.

If I had any free time, I'd love to work on it myself. :|

I remember when I first started using BeOS- it was version 3. My Brother purchased it at Best Buy, along with a copy of the BeOS Bible- I still have the BeOS Bible in storage. Fascinating history. I thought it was a great OS, but never had much use for it (I was in High School and primarily interested in PC games at the time).

The only other person I ever met that used BeOS was a guy that had bought one of their net appliances.