Palm ‘mulled Linux’ for next-gen OS
Insiders tell Secret History of PalmOS
Exclusive Palm Inc was considering Linux as the foundation of the next-generation PalmOS as recently as last spring, sources tell us. Palm eventually acquired Be Inc's development team last August, but internal discussions on the viability of a Linux-based handheld OS were taking place up to fifteen months ago. These were squashed by the lawyers, who concluded that Palm couldn't reconcile the GPL with the in-house view of intellectual property.
"Palm had a Not Invented Here syndrome in spades," a former staffer tells us, and the company spent much of 2000 in a funk, indecisiveness reigned over what should succeed PalmOS.
Late in 1999 Palm and Symbian had agreed to base the next-generation PalmOS on Symbian's kernel, promising developers new APIs within months, but the discussions went nowhere. (Going open source doesn't necessarily compromise one's IP: as Mozilla demonstrates. And both Sharp and Samsung have marketed Linux-based PDAs/phones).
But with the board reluctant to pursue the Symbian partnership, throughout 2000 Palm discussed the alternatives.
"5.0 then was random stupid tweakings in a development branch that was allegedly 'experimental', and 6.0 involved interminable meetings where staff discussed Linux, other embedded systems,". Staff say that management failed to provide technical leadership - and many of the discussions got mired in issues such as endian issues ("mostly a triviality", says our source) and whether the next-gen OS would need a modern memory management unit (MMU).
This confirms comments made by Palm's then CTO Bill Maggs to us at Comdex in late 2000, when he was adamant that Palm didn't need modern memory management. An astonishing claim we then thought.
"Surely any idiot can see the answer in the 21st Century is 'duh, yes!'", says our source.
Maggs quit Palm two months later, and wouldn't reply to our request for comment.
In August Palm decided new blood was needed, and bought Be Inc's engineering team. Palm isn't using the cut-down embedded version BeOS, BeIA, which was targetted at the NatSemi x86 system on a chip, but a ground-up OS for ARM.
Despite withstanding the challenge of a better-funded competitor fielding more capable devices, Palm's technical indecision contributed to the departure of CEO Carl Yankowski only weeks after the Be purchase.
"Palm's problem wasn't a lack of Programmer-Gods, it was just the Red Queen's problem: too much to do and not enough engineers."
In fairness, both Microsoft and Symbian have had their share of problems, with both companies tweaking their business models amidst constant reshuffles. As we exclusively reported last year, Symbian quietly dropped the prescriptive 'DFRD' approach to licensees in favor of an approach. This saw licensees get access to the source code; and Symbian pulled out of designing UIs altogether as Nokia took the lead in creating the Series 60 "platform", which runs on top of the Symbian kernel. (Siemens is the first major licensee).
In February founder and CEO Colly Myers resigned. Microsoft has revised its PDA division and roadmap frequently, too, with Steve Ballmer taking charge last year, and subsequent PDA reorgs causing much angst amongst Pocket PC supporters.
You see, we treat all disasters equally fairly. But we wondered if we hadn't been too mean to Palm as it reeled around for a viable future OS roadmap.
"You once wrote [here] that 'so badly had Palm neglected its engineering obligation under Yankowski, that only recently we were thinking of sponsoring a Black Hawk Down-style rescue mission to pluck the beleaguered engineers out of the Black Hole....' That feels a lot more like what it was like to work there than anything else I've read," says a former staffer.
Glad to be of service. It's classic Chaos Theory in practice: a butterfly flaps its wings in Santa Clara (or London, or Seattle), and eventually a tsunami of crap arrives in our inbox.®