The Haiku Project had its annual conference in Orlando this past weekend, and though I have little experience with BeOS or Haiku, I decided to attend and write about the conference because it's 20 minutes from my home. It's called WalterCon because "Walter" was one of the proposed names for what was to become a free replacement for BeOS. Be, Inc. may be dead and gone, and the BeOS source code may have been sold and warehoused, but the spirit of Be lives on in Haiku and WalterCon. If only it had a little more developer support, you'd be hearing about the Haiku operating system a lot more often. Heck, you might even be using it.

A little Haiku background

Haiku is an MIT-licensed replacement for the proprietary BeOS, which went down with the Be ship in September of 2001. Be, Inc. sold all of its software assets to what was at the time Palm, what later became PalmSource, and what is now ACCESS. It hasn't been put to use as far as anyone knows, but YellowTAB's ZETA operating system is rumored to be a naturally evolved, next-generation version of BeOS for x86 computers.

Since the BeOS source code was erased from the market, Haiku had to get its basis from someplace else. That ended up being NewOS, a pet project of Travis Geiselbrecht, a former Be software engineer. He eventually abandoned the project, and in August of 2001 -- shortly before the last bits of paperwork officially dissolved Be, Inc. -- the Haiku Project began. After five years, it's still not quite ready to replace more mature desktop operating systems like Microsoft Windows, Apple OS X, and various desktop BSD and GNU/Linux distributions. It is, however, dangerously close to becoming a viable desktop operating system in the near future.

Haiku is not just another open source operating system with the same open source desktop programs and scattershot approach to market focus. It's just the opposite, in fact -- it's binary compatible with the modest library of BeOS R5 applications, has a few of its own, and seeks to build more. Haiku also has a singular focus on desktop computing, which few other operating systems -- even Windows and OS X -- can claim.

Be failed for two reasons: Microsoft forced OEMs into refusing to dual-boot Windows preinstallations with BeOS, and Apple did everything it could to shut BeOS (which directly competed with NeXTSTEP to replace Mac OS 9 -- you probably know how that ended up) out of the PowerPC market, too. Only Microsoft ended being blamed for Be's failure, though, and when Be filed an anti-trust lawsuit against it, the two parties agreed to settle for a paltry U.S. $23 million (I say paltry because it's possible that BeOS could have eclipsed Windows if it had been allowed to compete properly).

WalterCon 2006 overview

Click here for a small directory of the photos I took at WalterCon 2006, and here's the official WalterCon photo database. The photos I took are of the Saturday morning setup, a group discussion on proposed Haiku application limitations and the structure of BFS, a small gathering of Haiku contributors during a break, and a live demo of Haiku on a projector.

The annual Haiku conference informally began on Friday night with a group dinner. I didn't attend that, but I did show up bright and early on Saturday morning to catch the state of Haiku address, given by project leader Michael Phipps. The half-hour speech primarily covered the improvements that Haiku has undergone in the past year (USB and networking support, mainly), and reiterated Haiku's market focus and anticipated user base. I took some video of the speech and some other impromptu interviews which can now be seen on Google Video here and here.

The day got a bit of a late start, but moved pretty quickly from there on. After the initial speech, we saw a video entitled "Haiku Rocks" that displayed Haiku's desktop capabilities. I saw a minesweeper clone, a basic sound editor and a sound playback application, a text editor, and other desktop programs. Though I'd seen screen shots of BeOS and YellowTAB ZETA over the years, I'd never really seen either of them in action. Haiku seemed to me to be a fairly capable operating system, though as I found out later, it still depends on some old BeOS modules and programs to be functionally complete.

What really amazed me about Haiku was its ability to boot from within other operating systems, and its unbelievably small footprint. The live demo ran off of a 128MB flash disk, though it didn't have more than a small set of very basic applications.

We all went to lunch a little before noon, and had some interesting discussions about Haiku's progress and direction, the development model of OpenBSD, and the increasing bloat of GNU/Linux distributions.

This being my first experience with real BeOS users and Haiku developers, I was surprised by a few discoveries:

• A minority of WalterCon '06 attendees currently owned a BeBox.
• None of the attendees were Apple users -- in fact most openly expressed disdain for Apple computers, software, and users.
• About half of the computer-toting attendees were using Microsoft Windows, the other half on GNU/Linux.

I'd been expecting more open source operating system use from people who had traveled from, in many cases, thousands of miles to participate in a conference on an open source operating system. I also expected more Apple-heads for some reason -- perhaps because Be, Inc. was started by a former Apple employee, BeBoxes ran on PowerPC hardware for most of Be's existence, and BeOS was a contender for Apple's next-generation operating system. I was a little disappointed that I didn't see anyone who really ate their own dogfood -- no full-time Haiku, BeOS, or ZETA users at all, though one faithful BeOS user did wander in late in the day, looking for help with an Internet connection problem.

The future of Haiku

As Saturday's session came to a close, discussion turned toward Haiku's future. What are the critical features and applications that Haiku needs in order to appeal to a wider audience?

In the same way that the original BSD slowly replaced proprietary Unix pieces to form its own operating system, Haiku is slowly replacing proprietary BeOS modules and applications. It's not going very quickly at the moment, though, with only about 20 active developers working on the project. Project leader Micheal Phipps hopes to get more developers interested in the project so that Haiku can progress more rapidly. The idea is to enhance the operating system in such a way that desktop applications can be easily written for it without having to re-code a lot of common algorithms. With that in mind, some of the potential advances in Haiku's future that the group discussed on Saturday are:

• Multi-user support (local logins only)
• Vector math library
• Speech synthesizer and voice recognition kits
• Standardized help system
• Synchronization kit for synchronizing data among various devices and services
• Instant message kit
• Native XML support
• Encryption plugin architecture
• Public key cache for encryption authentication
• Database kit (possibly using SQLite)
• Fax support
• Reference kit (dictionary, thesaurus, map, etc.)
• Sound effect plugins
• Color management
• Device kit (API wrappers for devices)
• Game kit (wrappers for game sound effects and hardware)
• Translation kit (translating data between different formats and modes of presentation)
• User interface enhancements: vector graphics instead of raster; internationalization; perhaps a different menu structure; improved font support
• Various improvements to the Be filesystem (BFS)
• Input kit as a wrapper for input devices

There was also a some talk of porting existing open-source applications from the GNU/Linux and *BSD realm, but Phipps shot that down, insisting that ports are usually a bad idea. The only exceptions would be major applications like Firefox and the GNU Gnash free software Flash player, but for the most part, Haiku's focus would be on the operating system itself and how to enhance it so that it would be easier to create and use native Haiku programs. For the past several years, the focus has been to replace proprietary BeOS components, and that effort must accelerate its pace if the project is to remain viable -- current Haiku software must be compiled with the ancient GCC 2.95 to maintain backwards-compatibility with old BeOS software. Making the jump to GCC 4.0 means that a lot of old programs will be left behind, and Haiku is not quite ready to do that.

If Haiku were to achieve even most of the above-listed functionality, it could reasonably come from behind to become the desktop operating system of choice not just for users, but for many OEMs as well. Since Haiku is MIT-licensed, it allows commercial derivatives to be distributed without source code. With a significant body of desktop software and a standard, documented, feature-heavy API, Haiku could easily drain Apple's market share while whittling away at the Windows and desktop GNU/Linux userbase as well.

"I'm not necessarily an open source advocate," Michael Phipps told the attendees. "Open source is a means to an end. It's the only way for BeOS to continue development."

That development does continue, but unless more developers lend their spare time to the project, it could be years before Haiku is ready to present itself as a useful desktop operating system.

As for WalterCon 2006, it was a lot of fun and I learned a lot about Haiku, BeOS, the BFS filesystem, and some pointers on writing device drivers. I look forward to reviewing Haiku when it's ready for a critical eye. At very least, it's now on my radar.

Discuss this article or get technical support on our forum.

Copyright 2006 Jem Matzan.