Discover more from Abort Retry Fail
'Be' is nice. End of story.
pick a name from the dictionary, it can work
Jean-Louis Gassée was born in Paris France in 1944. From 1968 to 1974, he worked for Hewlett Packard in Europe. He was in charge of a project to develop the first scientific desktop computer from HP, and he was later promoted to sales manager for the European market. From 1974 to 1981, he served as the CEO of Data General in France. In 1981, Gassée became the Director of European Operations for Apple Computer. A few years later, following the firing of Steve Jobs, Gassée was promoted to be the President of Product Development. From what I can tell, he spent a lot of time and energy thwarting bad ideas from the rest of the company while at Apple, but he also stewarded many great projects: the Newton, the Macintosh Portable, the Macintosh II line, and the much loved SE/30. Sadly, in 1990, he suffered the same fate as Steve Jobs before him, and he was pushed out of the company by Sculley and the board. Steve Sakoman (developer of the Newton) was the VP of Product Development at Apple, and he left the company with Gassée. Shortly after that, Erich Ringewald left Apple. He was the lead of the Apple “Pink” OS group, which was the group working on the next generation of Apple’s Macintosh operating system.
In an interview with the The Delivery Man, Gassée said of his journey:
I see a chain of pivotal moments: my childhood interest (1950s) in what we called radios, followed by being fired from the lycée, sent to a Breton boarding school where the Prefect of Discipline was a true geek and showed his recent acquisition, an OC 71 (germanium) transistor and encouraged my tinkering. Years later, in 1968, joining HP France to launch their first desktop computer, and finally being hired to start Apple France late 1980.
Abort, Retry, Fail is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.
These three gentlemen then set to work building a new company. The Chairman and CEO was Jean-Louis Gassée, the VP of Engineering was Steve Sakoman, and the CTO was Erich Ringewald. The name of the company was a subject of concern, and Sakoman had been going through the dictionary in alphabetical order, and he and Gassée would consider the words. They got to “be” and Gassée said “Be is nice. End of story.” It becomes rather clear in just a bit that these three minds were required for what was to be created.
Mr. Gassée is a rather opinionated man from what I can tell, and this isn’t new. He formed his opinions through experience in the industry over the course of decades. When he founded Be, he set his sights on an ambitious goal: fix the computer industry’s stagnation. From an interview for Tech Head Stories, 13th December 1995:
About the BeBox
The BeBox is a personal computer that relies on three ideas. The first idea is that we create a product that has a distinct architectural advantage in the freshness of its operating system. The most obvious example of this advantage is that every BeBox has two Power PC CPUs. Multi-processor PCs are actually quite easy to do on the hardware side of things: They're a very inexpensive way to increase computing power. And yet no one does it because they don't have the infrastructure, the the operating system, to support multiple CPUs. The other guys, Macintosh and Windows, they certainly won't be able to anytime soon. I know... I've lived inside one of these sausage factories; the layers of software “silt” are deadening, it's cancerous. It took Microsoft five years to go from Windows 3 to Windows 4. Apple will need six or seven years to move from System 7 to System 8. You know what I'm trying to say? Another example: we have a database engine built into the operating system. This is a dream of all PC makers, I can attest to that. Then there's very fast, rich I/O, multiple serial ports, MIDI ports... even a GEEK port that will let the bleeding edge hacker lift the hood and do unspeakable things to our computer.
The second idea was that we wanted to help the software developers reach the market. There are so many software developers who are frustrated by the dominance of a few large predatory birds in their ecological niche. A fledgling software developer has a hard time developing, so to speak. Today, imagine that you are a young Windows programmer and that I'm a venture capitalist and you come and see me and say, “Mr. Gasse, do I have a deal for you.” “Yes?” “I have the word processor for Windows that will kill Microsoft Word.” What am I to do if I'm a caring venture capitalist? I have to open the drawer and instead of pulling out the checkbook I should pull out the Magnum .357 and give you the coup de grace because this will stop what otherwise would be a long, ugly, expensive agony for your family. You can't compete; you won't get the money and you can't buy the shelf space. What we offer is a much different way to reach the market: You write an application and put up a demonstration version on our Web site. I see the demo, I download the demo, I use it, I like it... so what do I do then? I use the telephone. (Some day we'll have credit cards flying over the Internet, but let's rely on the existing infrastructure. ) I call you and give you three numbers: my credit card number, my Internet address, and my machine serial number so you can customize your application for my machine.
The BeBox was a seriously cool machine. It was first released in October of 1995, and it was a monster compared to other machines of the time.
Let’s start with the outside. So, there on the front, you have the traditional CDROM drive and floppy drive of the era. Then, there’s the Blinkenlights. The bottom-most right LED was used to show hard disk activity, and the other lights showed CPU load. The left array would light up corresponding to one CPU, and the right the other. Nothing there is too revolutionary (though quite cool), but let’s look at the back of this thing.
So, here, you have four 9-pin D-sub serial ports, a PS/2 mouse port, two 15-pin D-sub joystick ports, four DIN MIDI ports (two in, two out), four RCA ports (two in, two out, stereo), two 3.5mm audio jacks (one in, one out), three 4-pin mini DIN infrared I/O ports (for the younger among my audience, infrared was common in the 1990s), parallel, SCSI-2, AT keyboard, and 37-pin D-sub “GEEK port”. This port was a kind of GPIO interface implemented by Benoit Shilling.
The BeBox shipped with two PowerPC 603 CPUs clocked at 66MHz. These are 32-bit RISC microprocessors on a 0.5 micron process. They featured an 8KB code cache, and an 8KB data cache. Later models shipped with the 603e, which doubled both cache sizes, and bumped the clock to 133MHz. The 603e CPUs were on a 0.35 micron process. The BeBox allowed for eight 72-pin SIMMs which granted a maximum of 256MB of RAM. For expansion, the BeBox’s motherboard had three PCI slots, and it had five ISA slots. Another note on hardware that I feel is important is that this machine’s DAC allowed for 16-bit audio sampled at up to 48kHz; not shocking, but still rather impressive for the time.
There were other quite powerful workstation machines available in 1995, but I am not aware of any with quite so much I/O. To make full use of this beefy machine, Be Inc developed BeOS. The development team was made of twelve software engineers who hailed from companies including Apple, NeXT, and Sun. They worked for roughly five years to create a preemptive multitasking, multithreaded, symmetric multiprocessing, object-oriented, 32 bit operating system. The system was partially POSIX compliant, had an API written in C++, and used Bash for its CLI. Despite having Bash, the operating system was fully graphical. This OS even featured a 64-bit file system with database-like features (there’s a book about BFS if you’re interested).
In the end, something less than two thousand of these sweet sweet machines were delivered. The BeBox did not succeed in the market. I’ve seen a million different reasons people give for why the BeBox failed, but I think the real answer to this question is rather quite simple: the PC compatibles. I’ve mentioned in other articles, here on ARF, that the PC platform with Windows was absolutely exploding during the 90s. We have in the BeBox yet another victim. As innovative and as cool as the BeBox and BeOS were, they weren’t compatible with PC software. With MS-DOS and Windows being so dominant, there wasn’t demand for a machine on which zero already purchased software could run (in the 90s, people bought software in boxes from physical stores, and that software cost large amounts of money). The other extremely powerful systems of the time period were all UNIX systems (AIX, HP-UX, Solaris, IRIX) and these could usually easily compile code written for other UNICES. Additionally, the software market for these UNIX systems was very niche. BeOS wasn’t a UNIX either. Neither software compatibility with the PC, nor software compatibility with UNIX…
With the demise of the BeBox, BeOS shifted into a pure software play, and Be rapidly began porting BeOS to other hardware. Of particular interest to Be as targets for BeOS were PowerPC and Intel x86.
As of 1996, Apple was looking to replace their varying OS projects. There were problems within Apple that made the development of a next generation OS nearly impossible, and to solve these problems Apple sought to purchase something that was close to their own vision. With BeOS having been ported to numerous hardware platforms including some Macintosh machines, and even having shipped pre-installed on some Macintosh clones, Gil Amelio (then CEO of Apple) initially approached Be Inc about BeOS. There’s conflicting information regarding the reasons for the failure of this deal, but Apple eventually chose to purchase NeXT. Macintosh OS X and the current macOS are based upon NeXT Step (iOS, Watch OS, TVOS, and the rest are as well).
From there, things only got worse for Be. The company lingered around until 2001 when it sold its copyrights to Palm for $11 million (around $19120796 in 2023). In 2002, Be brought litigation against Microsoft for anticompetitive practices, but the suit was settled out of court for $23.25 million (around $39.76 million in 2023). After purchasing BeOS, Palm promptly discontinued the OS. Palm itself later had some problems, split, and sold. The rights to BeOS are now in the hands of Access Co along with PalmOS.
The funny thing is, BeOS was just too cool to die. Immediately upon its death, a German company called yellowTAB began developing the system as ZETA. ZETA was based upon BeOS 5.1.0. Ultimately, the company became insolvent, and Magnussoft purchased yellowTAB. Magnussoft failed to learn from the demise of yellowTAB. They continued to develop ZETA. Neither yellowTAB nor Magnussoft ever procured a license for BeOS, and Access Co claimed that ZETA was an illegal distribution of BeOS.
If you thought that that would be the end for BeOS, you are in error. Following the purchase of Be by Palm, an open source project was started whose aim was to recreate BeOS from scratch with full binary and source compatibility. This was OpenBeOS. The first release in 2002 was a community update to BeOS 5.0.3 with some open source replacements for Be code. The project name changed in 2004. With everything surrounding the demise of Be being highly litigious, it is no surprise that the project wished to avoid legal complications over their name. They chose Haiku.
Why did they choose the name Haiku? Error messages from some applications in BeOS are written in haikus (most notably, the NetPositive web browser). Additionally, they felt that the art of haiku was representative of the elegance and simplicity of BeOS.
The Haiku project continues to this day, and the system is quite usable. As of today, the project is working toward the imminent release of Haiku OS Beta 4.
Officially, Haiku only supports 32 bit and 64 bit x86 machines. Despite that and in the spirit of BeOS, Haiku does have ports to ARM, m68k, PowerPC, RISCV64, and SPARC.
Haiku has improved from where Be left off. It has some support for FreeBSD driver compatibility, WiFi, a WINE port for Windows applications, a real package manager, and so on. There are still some problems that would prevent many from using Haiku as their daily driver (primarily hardware support and a lack of 3D acceleration), but the project has moved quite quickly of late. I look forward to the day that I can run Haiku natively on M1 Macintosh.
Update - here are some images shared by the HN user Helf
“Simple is hard. Easy is harder. Invisible is hardest.” — Jean-Louis Gassée