Palm OS for Linux -- It's a Good Thing

[Ed Hardy]

Brighthand's Ed Hardy says that PalmSource's plan to base a future version of the Palm OS on Linux is necessary to save the company, and the Palm platform.

Read more at http://www.brighthand.com/article/Pa...Plan?site=Palm

[Kevin Daly]

I'm not 100% sure I understand this one.
Surely if Palm had an advantage at all OS-wise (and admittedly it's one they squandered through lack of imagination), it was in having an OS that was designed from the start with a view to the platform it was aimed at.
As a child (somewhat legitimate) of UNIX, Linux ultimately has its origins in something designed 3 decades ago for a, um, somwehat different hardware configuration and set of usage scenarios.

That doesn't mean I think Linux for handheld devices is not an excellent platform (I have no experience of it so I have no right to make such an assumption), but it does mean that I don't quite get why it's perceived to be such a great idea.

Oh well, at least this means that all those cracks about Windows Mobile being an absurd shrunk version of a desktop OS will have to go the same way as "What would you want colour for?" and "Why would you want all that memory?".

[feranick]

Linux has already proven to be the most flexible OS existing today. You can run server, workstations, embedded systems, your TiVo, ATM machines with customized version of Linux. Although derived from UNIX, being open helped it to be customized. There is no other OS (kernel, to be more precise) able to handle this. Would you prefer a proprietary kernel, which only a company have the access to the code?

Linux kernels in therir various flavors do not necessarily represent scaled down desktop versions. on the countrary they can be customized to be very compact and efficient, but powerful at the same time. Have you ever expressed a desire for a proper file system in your palm? or real multitasking?

Personally this is a great thing. Linux is a strong, modern OS. But its main strenght is the large wide range of open source developers, which can produce drivers, customization to the OS itself. Think drivers for example, being available for every piece of hardware. Conversely this is the big shortcoming of Cobalt. Porting PalmOS into a linux kernel frees Palmsource from the burdain of producing drivers, kernel optimization, to concentrate on what they know best, producing User interfaces and applications.

Does this sound good enough to you?

[robrecht]

"The single biggest criticism I've heard about this move is that it's coming too late.

I think that those who are saying this have, subconsciously or not, the belief that Linux is some kind of fad that's going to blow over in a few years."

It's not that Linux is a fad, but that OS6 still hasn't become a fad. I'm not sure PalmSource/Pa1mOne can skip a generation without going the way of the Dodo bird.

[rsc1000]

Generally a good article ED. Though technically more accurate about what PalmSource is doing than that that other editorial on this subject over at PalmInfocenter (mike canes piece), I think you are still off on some points:

"For example, it didn't support screen rotation or VGA screens"

Though the built in apps in OS 6 didnt take advantage of it, VGA and much higher res is there without device manufacturers needing to do any of the kind of work that sony did when it first did hires on Palm OS. In fact, if you download the debug version of the Sim you can set it to VGA. All of the graphics APIs in Cobalt are not fixed in any way to a particular resolution - it all scales and no company making a Cobolt device would have create any API or modify the existings one to get it to work. Its really just re-doing the exist apps so that they make sensable use of this (ie - you dont just want have the PIM app suddenly have all text and icon appear to be really tiny and on readable.)

Put simply, PalmSource is going to put the Palm OS look-and-feel on top of a Linux kernel. This will mean that its developers will no longer have to be responsible for developing an entire operating system, just the parts that users see.

Thats not really true (this is the thing that people don't quite seem to get here): The Cobolt (really the Protein APIs) APIs are not just user-interface; they are security/encryption, communications, graphics, media service, etc. Its the other way around - the kernal is a smaller part of the platform here (although an extremely complex and import one).


Somebody else posted :

Surely if Palm had an advantage at all OS-wise (and admittedly it's one they squandered through lack of imagination), it was in having an OS that was designed from the start with a view to the platform it was aimed at.

Again - misunderstanding. I've got news for you: the original Palm OS dragonball version (Palm OS 1 to Palm OS 4.2) ALL used a 3rd party kernal that Palm licensed. So - they are not using a desktop linux (they are using mLinux kernal) and they are not really using another 'OS' at all. The code will be the same. They are not 'skinning' Linux. This is a real incarnation of Palm OS BUT by using Linux as the kernal they will not have to worry about getting hardware companies lto write ow-level drivers etc as a barrier to to choosing Palm OS.

[strider_mt2k]

At first I saw this as merely a death-spasm, woeing the fate of PalmOS.
Now that I've been reading what the wider implications are I'm actually feeling a glimmer of hope.

Even if this attempt fails, at least SOMETHING is being done.
Experiments and explorations aren't always successful, but almost always interesting.
(As long as you aren't standing too close!)

[Ed Hardy]

rsc1000, good points. But sometimes when writing an editorial you have to simplify things a bit so you can get your point across without getting bogged down in details.

I deliberately skipped over the fact that PalmSource is going to put the Cobalt APIs in because, as I said, I was trying to keep this editorial from getting too technical. To non-geeks, that sort of thing gets lumped in with the user interface anyway. Maybe I was over-simplifying, but I'd rather that than drown my readers with a lot of technospeak.

As for the VGA thing, I guess it comes down to how one defines "support." With Windows Mobile SE, you can put the OS on a handheld with a VGA screen and you're pretty much done. To put Cobalt 6.0 on a handheld with a VGA screen you have to re-write the user interface for the PIM apps and more. To my mind, that doesn't qualify as "support."

p.s. And don't get me started on what my definition of "is" is. :-)

[Shawn510]

One thing worries me about this, the Linux license.

When Apple built OSX on top of BSD, they did not have any issues with the license because the BSD license is made to allow that sort of thing. But the Linux GPL license seems to prohibit doing that sort of thing without revealing everything you add into the code (which palm would never do). While no action has ever been taken, Linus himself has said that he did not think binary only device drivers (like the nvidia display driver) were legal under the GPL, for example.

I seem to remember much argument over the GPL with the pro-GPL side saying it was superior specifically because it would prevent someone from doing what apple was able to do with OSX. (why this would be superior, I don't see) How does this new deal get around the probelms of the GPL? Wouldn't one of the BSDs have been a more logical choice?

[JackAubrey]

Linux is nice. In the corporate world, Linux is gaining favor in some shops - but on the server only. On the desktop? Linux is MIA. Windows remains the only game in town. PDAs are much closer to the desktop than the server - in both configuration and mindset.

PDAs will eventually use a variant of Windows XP. Within 3 years WinCE OS's will be dropped in favor of XP (you heard it here first folks!). Even today, the processing power differences between PDAs and Notebooks make WinCE (or Palm OS for that matter) superfluous.

Consumers who shop at Best Buy will not be comfortable with a Linux OS. Hackers, hobbyists, gamers and geeks will buy Linux. No one else (i.e., the remaining 99% of the market) wants it. Bill G has already won this one.

Palm will not gain marketshare with Linux. The radically increased R&D effort may even sink them.

[strider_mt2k]

Doesn't the new Sony piece run Win XP already?

And the OQO?

I have to admit, an XP handheld to compliment and work well with my XP Desktop would be awesome to say the least!