1995

Jean Louis recalls vividly that "In February 1995, I was - one of the many times, four or five - looking at the whites of the repo man's eyes. We were in trouble."

Despite yet another 'near-death' experience for the company, Be Inc narrowly survived, and work continued in earnest on the new PowerPC based hardware.

After Glen Adler's initial work on the PowerPC add-on board, hardware engineer Joseph Palmer was hired to almost single-handedly transition Be to an entirely new hardware architecture based around the PowerPC. He did this with remarkable success.

The entire PowerPC motherboard design, including the use of industry-standard ISA and PCI buses, was accomplished by Joe Palmer. Some people have questioned why he used the standard (and somewhat aged) standard PC I/O system in the BeBox; in response, Joe indicated that for him, it was a matter of unavoidable compromise:

"When I designed the BeBox I used the PREP design and a PC I/O system. We didn't have the resources to invent a better DMA controller, so we didn't. Guess what? The PC one worked well enough. Would the BeBox have been "better" with a new DMA architecture? Yes. Would the end user have been able to tell? Probably not, but the delivery schedule would have slipped. End users notice that."

A lesser controversy has been how and when the 'BeBox' name was coined. Here at the BeBox Zone, we've noted some thoughts and comments on the issue.

While Joe Palmer concentrated on the hardware, the port of Be's software to the PowerPC processor was basically done by two engineers, Bob Herold and Cyril Meurillon. Bob single-handedly brought up each revision of Joe's new hardware and "debugged impossibly complex interactions between the lowest levels of our software and the new hardware". Cyril, meanwhile, rewrote a number of speed-critical, machine-dependant pieces of the kernel, including the interface to the memory management unit (MMU).

The rest of Be's "high-level" software engineers basically just continued their work on the Hobbit boxes until the day Joe showed up with new PowerPC machines. With a new compiler and literally a simple recompile all of the applications, including the Browser, were soon running on the new PowerPC boxes.

The apparent ease of the transition between the Hobbit and PowerPC architectures is a reflection both of the genius of Joe Palmer's hardware design, and the architecture and implementation of Be Inc's developing operating system software.

As the PowerPC design took hold another key engineer was hired. Brad Taylor quickly became the most popular guy on engineering row when he implemented the TCP/IP networking stack on the BeBox and ported the ftp tool. Before the port of ftp, all Be's software developers were forced to use the tar tool to transfer files on floppies from their UNIX development machines to their prototype BeBoxes. With ftp, the transfer time was reduced by at least an order of magnitude, from minutes to seconds.

After feverishly working throughout 1995, it was finally decided that the results of 5 years of low profile work were finally ready for a public launch. When one considers that little over a year earlier, Be's entire hardware platform had been made redundant by the demise of the Hobbit processors, this was a remarkable achievement.

Of course, another factor in deciding to go public was that they were again in serious debt. It was hoped that the publicity from a public airing of their work would be enough to attract further funding.

Jean Louis Gassée was, however reluctant to unveil the BeBox at Agenda '95, with its audience of industry analysts and members of the media. Gassée was instead keen to launch his baby to a crowd of geeks who, he thought, would really understand and appreciate what Be had achieved, and where they were heading. Gassée's own words on the Agenda 95 crowd were: "These are industry insiders ... They all hate what we do. I wanted audiences of geeks.".

Despite Jean Louis' reluctance, the pressure of a desperate need for new funding meant that Agenda 95 would be the venue for the BeBox launch.

In his preface to the BeOS Bible, Henry Bortman provides the following rivoting description of the BeBox launch and the days leading up to it:

Gassée and his engineers arrived a couple of days early. They brought with them everything they needed for the demo - or so they thought. It had all worked flawlessly back at the office. But when they arrived in Scottsdale, Arizona, for the big event, nothing worked. The engineers called back to California, to get people to "bring more stuff", as Bob Herold recalls. "Software, different hard disks, whatever they thought would work. It was a bit of a crapshoot." They finally got it "mostly" working.

The day of Be's presentation also happened to be the day that the O.J. Simpson jury announced its verdict. Be's demo had been timed carefully to be completed before the verdict was read. Although the demo system was still plagued with problems, Steve Horowitz, who was running the demo machine behind the scenes, "got very adept at moving things out of the way," says Herold. "If the debugger would come up, he'd move it away before anyone noticed."

Apparently the ruse worked. Be got a standing ovation, only the second time in the history of Agenda that anyone has received such accolades. Gassée was speechless. Literally. Anyone who knows Gassée knows that this is a rare event. "I wanted to say my thanks to a number of people. And I couldn't do it."

His momentary lapse of eloquence notwithstanding, Gassée's gamble had paid off. "Agenda was the turning point. That got us in the big time. We went from nothing, in terms of VCdom, to the first tier VCs. Which is very good. Not just the money, but also access." Not that the money was insignificant. While waiting for investors to come through, Gassée remembers, "we had a couple of near-death experiences. I'm not joking when I say that I've seen the whites of the repo man's eyes." Gassée keeps a Xerox copy of the four-million-dollar check from Dave Marquardt, dated April 9, 1996, pinned to the wall of his office. Marquardt had been in the Agenda audience at Be's public unveilling.

Following the public launch of the BeBox on October 3rd, Be was "a little overwhelmed" by the reaction. Their previously unknown website was swamped as hundreds of curious and eager developers applied to obtain a BeBox. Amidst this frenetic attention, Be Inc managed to ship over 100 BeBoxen to developers by the end of 1995.

The process of building the early BeBoxen was somewhat involved. The BeBox motherboards were being built by a Taiwanese contracting company called Altatron. Ironically, their Milpitas manufacturing site had actually been built and run by Apple, and JLG had in fact worked on the site during his time at Apple.

Once the mainboards were delivered to Be Inc, each BeBox was assembled by hand in their offices. In the beginning, the BeBoxen were somewhat naked - the characteristic front bezel wasn't shipped to many early developers until long after they had fallen in love with their BeBox.

The enthusiasm of developers was highlighted by the speed with which they started to setup mediums for discussion and communication amongst themselves, and with Be Inc. It took only until November before a Call For Votes (CFV) went out to establish a dedicated BeBox newsgroup (comp.sys.be). The proposal was approved by a huge margin, and the comp.sys.be newsgroup was up and running by the end of December.

 Questions? Comments? Contact Andrew Lampert (webmaster at bebox dot nu).

© 1998-2009 Andrew Lampert