PalmSource Buys Linux-Based Mobile Phone Company, To Port PalmOS To Linux?

[Janak Parekh]

<div class='os_post_top_link'><a href='http://www.palmsource.com/press/2004/120804_cms.html' target='_blank'>http://www.palmsource.com/press/200...120804_cms.html</a><br /><br /></div><i>"PalmSource, Inc., provider of Palm OS®, a leading operating system powering next generation mobile devices and smartphones today announced entry into an agreement for the acquisition of China MobileSoft Limited (CMS), a leading Chinese mobile phone software company with business operations headquartered with its wholly-owned subsidiary, MobileSoft Technology (Nanjing), in China...PalmSource will continue to offer both Palm OS® Garnet and Palm OS® Cobalt to support a broad range of mobile devices including smartphones. PalmSource also plans to implement Palm OS on top of Linux, bringing the benefits of Palm OS to the Linux community, including the award winning user interface, software frameworks based on the best of Palm OS and BeOS®, a large base of professional and consumer applications, and an enthusiastic community of more than 25 million users and over 360,000 registered developers."</i><br /><br />I'm personally a fan of Linux -- I use it daily in my work and research. But this press release smacks of tossing it around as a buzzword. It's no panacea, and it's no trivial thing to take an OS like Palm OS and run it "on top of Linux" -- it's <i>not</i> just running about running "make" as the PR implies. Finally, I'm not sure how they're going to reconcile this with the rich, robust kernel they already have from BeOS. If anything, this smacks me as a lack of focus, which in my opinion is the <b>last</b> thing PalmSource can afford right now; as it stands, there's no Cobalt devices out the door yet.<br /><br /><b>Update:</b> PalmSource has posted a <a href="http://www.palmsource.com/announcement/communityletter.html">letter to the community</a>. In short, "Palm OS For Linux" will be the <i>next</i> version after Cobalt, and both Garnet and Cobalt will continue to be sold alongside it. I still disagree with that principle, and I think it'll lead to confusion, but we'll see.

[surur]

This is a disastrous loss of focus! Why would anyone want to support Palm OS 6 if i one to two years they plan on replacing it with Linux.

Good bye palm, we knew thee well.

Surur

[emuelle1]

I can't claim to be any more enthusiastic about Linux than I am about Palm. A friend sent me a Pocket Pc with Linux on it. I couldn't do anything with it. I screwed up trying to reinstall the Windows ROM and had to send it back to him. Sure, it's a neat gadget, and I do hope that people will find a use for it, but I personally will stick with Windows devices for the time being. Windows has it's issues of course, but pretty much everything is written for it. Years ago I had the time to sift through command lines and configuration files, but I just don't have it in me any more.

[huangzhinong]

It is too early to say it is good news or bad news for Palm OS. This version will be only top layer on the Linux core, same as PPC on Wince. Considering PS already made the mistake to buy BeOS which didn't bring any improvement to current POS, it is a good move for POS to correct the problem. POS will get the REAL multiTask and MultiThreads instantly without any increase of costs. POS bought the CMS today too, which enable porting the POS layer to linux will be fairly easy and very low cost.

PS and POS developers and POS users will all benefit for this move in the very near future.

[felixdd]

Wait wait wait...lemme get this straight...they plan on redoing the OS structure...even before they know what to do with the current one?

And this just gets to me:
PalmSource will continue to offer both Palm OS® Garnet and Palm OS® Cobalt to support a broad range of mobile devices including smartphones.

How can you "continue" to offer something you never offered yet?

[Ed@Brighthand]

PalmSource is doing what Apple did so successfully, put its user interface on top of the Linux kernel, plus selected Linux services appropriate for handhelds and smartphones.

The Palm OS software layer will include the familiar Palm OS user interface as well as a set of middleware and PIM applications

They say that properly written Palm OS 68k applications will run unchanged on Palm OS for Linux, and that Palm OS Cobalt native applications will port with a simple recompile. We'll have to wait and see how easy this turns out to be.

In addition, Palm OS for Linux will be able to run many third-party Linux applications and services. However, applications will need to use the Palm OS APIs.

As for the argument that making this announcement will reduce companies' interest in Palm OS Cobalt (OS 6), I have two responses:
1) There's always something new coming down the pike.
2) What interest in Palm OS Cobalt? :wink:

[Janak Parekh]

PalmSource is doing what Apple did so successfully, put its user interface on top of the Linux kernel, plus selected Linux services appropriate for handhelds and smartphones.
It took Apple years and years to do that with OS X, though. It was very late in coming. I'm not so sure PalmSource can ride the situation that long. Look how long it took them to develop Cobalt on BeOS...

2) What interest in Palm OS Cobalt? :wink:
Well-put, admittedly. :P

--janak

[Janak Parekh]

It is too early to say it is good news or bad news for Palm OS.
I'll buy that, except how long do they have left before we can conclude it's good news, bad news, or irrelevant news because everyone's left the platform for WM/Symbian/whatever?

--janak

[emuelle1]

I consider it irrelevant. It would be like Mac saying they were going to develope software for the Palm platform. I don't use either, so all I would say is "best of luck" and move on.

It's always good to keep up though. You never know; this might actually lead to something that would cause a mass migration back to the platform.

[Ed Hansberry]

PalmSource is doing what Apple did so successfully, put its user interface on top of the Linux kernel
Really? I wasn't aware that Apple took OS9 and converted it to a AACE (Apple Application Compatibilty Environment) then wrote a new kernel and said you could make API calls to it but we aren't going to allow you to make full blown apps for it - and the new system would break most utilities and some apps - , then released a new OS that would allow you to make full blown API apps for it but no one would actually build a device with said new OS, and then take the whole shebang and put it on top of Linux.

Somehow, I think Apple skipped a big chunk of that mess. :lol: But I don't use Apple, so I could be mistaken. Maybe PalmSource is doing what Apple did. Then again, I doubt it. :wink:

Man, where are my PDA buzz posts that so often predicted that Palm didn't have what it took to write their own 32bit OS and transition OS4 users to it? :devilboy:

[Brad Adrian]

And what about the news about their acquiring a Chinese developer? Does anybody think this is going to open up the Chinese market and save the company?

[Jonathan1]

I'm personally a fan of Linux -- I use it daily in my work and research. But this press release smacks of tossing it around as a buzzword. It's no panacea, and it's no trivial thing to take an OS like PalmOS and run it "on top of Linux" -- it's not just running about running "make" as the PR implies. Finally, I'm not sure how they're going to reconcile this with the rich, robust kernel they already have from BeOS. If anything, this smacks me as a lack of focus, which in my opinion is the last thing PalmSource can afford right now; as it stands, there's no Cobalt devices out the door yet.


Huh. A couple thoughts. First off I’m amazed. I didn’t know Palm even had the capital to buy anything other then a deluxe deep-dish pizza from Pizza Hut at this point let along a software company. What did MobileSoft Execs lose a game of thumb war?

Secondly they must be having more problems then we could have ever imagined with Cobalt. That comment probably sounds remarkably pro Pocket PC. It isn’t. I have, like most people who “get it” have deep concerns of what may/will happen if Palm goes bye bye. But it is a valid question.
If this announcement happened a year and a half ago it would have been hailed as a bold brilliant move. But this isn’t Summer of 2003. This is the eve of 2005. With Pocket PC owning more of the market then Palm, with no shipping Cobalt devices, with declining Palm OS branded hardware, with whispers of PPC Treos. What can this purchase really do in the short term? I say short term because from the looks of it Palm doesn’t have a long term future. Not with the current snapshot of their status.

I can always hope for the best but this looks surprisingly like a National Geographic documentary I saw a few days ago. The lionesses stalking their prey. In the end when the prey knows they are pretty much screwed they get panicky and do some rather stupid things. Same plot different film.
Maybe we need a new series? When Microsoft attacks? The life and death struggle of businesses that go head to head with the Redmond giant.

[twalk]

You guys are missing some facts here...

PS already has the palmos simulator running of linux. Linux has already been ported to ARM. What PS is needing to do is to get the simulator to run on the ARM version of linux, and with lower resources.

Now why should they want to do that? Because several large cell phone companies (Motorola, etc) have decided to focus most of their efforts on linux cell phones for the future. If PS succeeds, then you'll see these linux phones also being able to run PalmOS software.

Does this mean that they're dropping OS6? No. What it means is that PS is hedging their bets, so if PalmOS specific cell phones fall behind in sales, then they've got another horse in the race with linux phones running PalmOS.

(Hey, if P1 was your major partner, wouldn't you also want to hedge your bets?)

[Janak Parekh]

PS already has the palmos simulator running of linux.
Interesting. Can you clarify, or are there articles on this you can link us to? Do you mean the PalmOS emulator, or a set of API frameworks that compile as libraries against a Linux OS? The "developer PalmOS emulator" is not even close to what they'll need -- it emulates a processor and is a complete virtual machine, whereas they need to port their libraries and stuff from the BeOS kernel to Linux, and they have to build a VMware-like product that doesn't actually do processor emulation but just syscall translation.

If PS succeeds, then you'll see these linux phones also being able to run PalmOS software.
But why would Moto, et. al. adopt Palm OS for Linux as opposed to using their Linux platform, which may be mature by then?

Does this mean that they're dropping OS6? No. What it means is that PS is hedging their bets, so if PalmOS specific cell phones fall behind in sales, then they've got another horse in the race with linux phones running PalmOS.
See, I wish they didn't follow this "multiple horse" strategy. It's confusing to me, and I can't help but feel the consumer market will be confused as well. Apple ditched previous OSes after adopting OS X very quickly, and standardized and centered around that platform to help it gain steam as quickly as possible. If PalmOS EOLed Garnet I bet we'd see Cobalt hit the market more quickly.

(Of course, I'm speaking as a user and developer. There may be OEM contracts that prevent them from doing this.)

--janak

[Ed@Brighthand]

but we aren't going to allow you to make full blown apps for it
It is my understanding that developers will be able to write regular Linux apps for Palm OS for Linux. They will also be able to recompile Cobalt-native apps for it, or use older 68k apps without modification. I'm not an OS X user, but this seems to me to actually be better than Apple's solution.

[Ed Hansberry]

but we aren't going to allow you to make full blown apps for it
It is my understanding that developers will be able to write regular Linux apps for Palm OS for Linux. They will also be able to recompile Cobalt-native apps for it, or use older 68k apps without modification. I'm not an OS X user, but this seems to me to actually be better than Apple's solution.

My comment was the on the OS5 itteration.

Their FAQ states that Older apps will run in PACE and newer apps for OS6 (of which there are none in use since there are no OS6 devices) with the Protien API can be recompiled for the new OS. Not sure how that is better than Apple's solution. If Apple had done a headfake like Palm, they might have to worry about 2 older environments to support, not one. Can you imagine if Palm had come out in 2001 and said "Ok, this is the deal. OS4 apps will still work except hacks, OS5 is coming but you can't write 32bit apps for it, you have to use PACE, OS6 will come with the help of our BeOS experts, but no one will use it. Shortly after we declare that stillborn, we'll drop this whole thing on Linux. Sound like a plan? Thanks for supporting PalmOS."

Developers would have left in droves.

[jonathanchoo]

This is good news. Finally once PalmOS for Linux mobile phone is released, my infidelity with Windows Mobile will come to an end. :twisted:

[Ed@Brighthand]

Look, I don't want to find myself in the position of being the Palm OS Defender. I'm well aware that it has many problems. For one thing, it doesn't run on the Axim X50v that has become my inseparable companion.

Still, you are overstating the situation (not without a bit of humor).

The (overstated) Pocket PC version of that is Microsoft saying in 2001, "Here's a new handheld operating system. We hope you like it because we're not going to make any real changes to it for 5 or 6 years. We'll launch another OS for mobile devices next year that is incompatible with this one. Good luck figuring out which one is right for you."

The (overstated) BlackBerry version of that is RIM saying a few years ago, "Here's our wireless handheld OS. Sorry no one but us wants to make devices running it, and we're not going to make it easy for anyone to write applications for it, but we've hidden subliminal messages in it that will make it more addictive than crack."

The point is, no OS company is perfect, and we can (and should) point out the flaws in all of them.

:wink:

[jonathanchoo]

True true Ed!

There is Windows Mobile 2003 for Smartphone and Windows Mobile 2003 for Pocket PC Phone Edition. Apparently 3rd party applications are not compatible with each other.

On the PalmOS side however, we have some applications that are properly coded that can work on PalmOS 4, 5 and 6. Most applications released today should work on OS5, 6 and the Linux version. If it can't, a simple update would solve the problem unlike on the Windows Mobile platform which is totally incompatible (last I checked). Then there are the other devices running proprietary GUI over CE .Net.

PalmOS is not perfect but to argue the move to create a GUI over LinuxOS will create confusion is totally unwarranted.

[tthiel]

Wrong. OS X is based on BSD. BSD has been around much longer than Linux and was also the basis for the Next OS. Apple had developers from Next ready to go and it still took them several years to come up with a Unix based Mac OS. I get crashes with 68k apps on my T5 (which is about to be returned) all the time and Palm is going to make 68k apps work on this mess? Right.

PalmSource is doing what Apple did so successfully, put its user interface on top of the Linux kernel, plus selected Linux services appropriate for handhelds and smartphones.

The Palm OS software layer will include the familiar Palm OS user interface as well as a set of middleware and PIM applications

They say that properly written Palm OS 68k applications will run unchanged on Palm OS for Linux, and that Palm OS Cobalt native applications will port with a simple recompile. We'll have to wait and see how easy this turns out to be.

In addition, Palm OS for Linux will be able to run many third-party Linux applications and services. However, applications will need to use the Palm OS APIs.

As for the argument that making this announcement will reduce companies' interest in Palm OS Cobalt (OS 6), I have two responses:
1) There's always something new coming down the pike.
2) What interest in Palm OS Cobalt? :wink:

[tthiel]

Assuming all that works which is a huge if.

but we aren't going to allow you to make full blown apps for it
It is my understanding that developers will be able to write regular Linux apps for Palm OS for Linux. They will also be able to recompile Cobalt-native apps for it, or use older 68k apps without modification. I'm not an OS X user, but this seems to me to actually be better than Apple's solution.

[tthiel]

Has it escaped everyones attention that there have been Linux PDA's already and neither user, developers, or hardware companies are flocking to it?

[tthiel]

Um, no. Not unless it was open source Linux. The Chinese ogvernment has mandated open source or not at all.

And what about the news about their acquiring a Chinese developer? Does anybody think this is going to open up the Chinese market and save the company?

[tthiel]

Now THAT was LOL funny!


Huh. A couple thoughts. First off I’m amazed. I didn’t know Palm even had the capital to buy anything other then a deluxe deep-dish pizza from Pizza Hut at this point let along a software company. What did MobileSoft Execs lose a game of thumb war?

[Janak Parekh]

There is Windows Mobile 2003 for Smartphone and Windows Mobile 2003 for Pocket PC Phone Edition. Apparently 3rd party applications are not compatible with each other.
This is a totally different comparison. If you've used a Smartphone, you'll note applications have to be designed differently, as the UI is fundamentally different -- to support a device without a touch screen.

Palm OS 4, 5, and 6 all have almost the exact same UI, and all require a touch screen. And so will Palm OS for Linux.

PalmOS is not perfect but to argue the move to create a GUI over LinuxOS will create confusion is totally unwarranted.
Creating a GUI over Linux is not the source of confusion. Having to maintain multiple versions for similar devices is. The more accurate comparison would be to the fact that Pocket PC 2000 had different processor support and you had to have separate builds of programs for each processor type. That was endlessly confusing, and MS canned it for Pocket PC 2002.

--janak

[Ed Hansberry]

Still, you are overstating the situation (not without a bit of humor).
Not me. No. Never. :lilangel:
The (overstated) Pocket PC version of that is Microsoft saying in 2001, "Here's a new handheld operating system. We hope you like it because we're not going to make any real changes to it for 5 or 6 years. We'll launch another OS for mobile devices next year that is incompatible with this one. Good luck figuring out which one is right for you."
I guess the difference is, I don't want a new OS every few years that effectively renders apps obsolete or they have to be recompiled. The list of things MS adds to PPC each year grows. Even many Palm-Sized apps for the old PsPC OS run on Pocket PC as long as you have an ARM version. With 2002, VPN support was added, as were spell checkers, terminal server support, instant messenger, etc. 2003 gave us the stability and speed of CE 4.0, faster PIE, animaged GIF support, more VPN options, SSL in inbox, etc. 2003SE gave us more tweaks like landscape mode and VGA support. Not revolutionary. Evolutionary. Would I like them to move faster? Sure. At the expense of every OS update being as much of a switch from PsPC to PPC 2000 was? No, not really. Yet that is the course Palm seems to be on. Every rewrite is a massive undertaking for many developers and frustrating while end users wait on their favorite app to be released for support for their new device. Sony made that even worse with their additional hacks to the OS.

The point is, no OS company is perfect, and we can (and should) point out the flaws in all of them.

Agreed. And I do my share of that for PalmOS so you don't have to. :wink: :lol:

[Felix Torres]

Uh, we're talking PalmSource here, right?
So the first question is what *shipping* products does the company have and what customers it has for those products.

And, to me, the critical question is the latter; PalmSource seems to have a lot of non-shipping products (demoware?) but not much in the way of customers. Coming up with yet another product-without-a customer doesn't seem likely to help the bottom line any time soon...

At this point in time, I'd be more impressed by announcement of new customers that announcement of new products to come sometime down the road.

2cents...

[twalk]

PS already has the palmos simulator running of linux.
Interesting. Can you clarify, or are there articles on this you can link us to? Do you mean the PalmOS emulator, or a set of API frameworks that compile as libraries against a Linux OS? The "developer PalmOS emulator" is not even close to what they'll need -- it emulates a processor and is a complete virtual machine, whereas they need to port their libraries and stuff from the BeOS kernel to Linux, and they have to build a VMware-like product that doesn't actually do processor emulation but just syscall translation.

There are 2 different things: the emulator (older, does a full 68K bytecode run, limited to OS 4 at most) and the simulator (basically an api hook/framework, with support for 32-bit armlets through .dlls). The simulator is *much* faster than the emulator, which definitely would be too slow to do this. (A version of the simulator would still need to be optimized more than it currently is to run on a cell phone sized device, and they need to stomp the rest of the bugs in it :-(.)


If PS succeeds, then you'll see these linux phones also being able to run PalmOS software.
But why would Moto, et. al. adopt Palm OS for Linux as opposed to using their Linux platform, which may be mature by then?

Because despite it's problems, PalmOS still has the most apps, the most developers, the biggest installed base, and one of the best brands for mobile devices. Linux has practically no apps (for mobile devices) in comparison, and very few OSS devs are making mobile linux apps.

Does this mean that they're dropping OS6? No. What it means is that PS is hedging their bets, so if PalmOS specific cell phones fall behind in sales, then they've got another horse in the race with linux phones running PalmOS.
See, I wish they didn't follow this "multiple horse" strategy. It's confusing to me, and I can't help but feel the consumer market will be confused as well. Apple ditched previous OSes after adopting OS X very quickly, and standardized and centered around that platform to help it gain steam as quickly as possible. If PalmOS EOLed Garnet I bet we'd see Cobalt hit the market more quickly.

(Of course, I'm speaking as a user and developer. There may be OEM contracts that prevent them from doing this.)

--janak

About all I can say about that, is that I think it was really, really stupid to do 5 -> garnet and 6 -> colbalt (and then 7 -> linux/palmos ?). They're all backwardly compatible... I guess that they were just trying to copy one of MS's stupider ideas.

As for PS EOLing garnet, from what I've been able to find out, colbalt 6.0 wasn't ready for prime time, and 6.1 only came out a few months ago. (The theory is that this is what really screwed up the T5/treo650 release.)

Todd.

[maximus]

Palmsource is acquiring Symbian !

.
.
.
.

Just Kidding.

[Robert Levy]

On the PalmOS side however, we have some applications that are properly coded that can work on PalmOS 4, 5 and 6. Most applications released today should work on OS5, 6 and the Linux version. If it can't, a simple update would solve the problem unlike on the Windows Mobile platform which is totally incompatible (last I checked).

I run the Windows Mobile app compat team and would love to hear about any cases of incompatibilities between devices. Our FAQ of known issues and workarounds is at http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dnppcgen/html/migration_developers_faq.asp . At the bottom of that you'll see an email address which I invite everyone to use for letting us know if you run across anything not listed there. We aren't perfect and occassionally APIs need to be deprecated but a ton of effort goes into maintaining compatibiltiy in our development platform.

As for the Palm announcement, my **personal** take on it is that this statement is the most intruiging:


Q. Does Palm OS for Linux replace current versions of Palm OS?
A. This is an addition to our line, not a replacement. Other versions of Palm OS continue to be available. As always, we'll make decisions on their future growth path based on feedback from our licensees and other partners.

Backwards compatibility is a hard but largely solvable problem. However, the idea of simultaneously maintaining both versions of the OS presents an even bigger problem of "sideways" compatibility. Combined with their plan to expose native Linux APIs to developers (apps using these APIs would obviously not work on the non-Linux branch of the OS), it's going to be interesting to see how this plays out in their developer and user communities.

[felixdd]

They say that properly written Palm OS 68k applications will run unchanged on Palm OS for Linux, and that Palm OS Cobalt native applications will port with a simple recompile. We'll have to wait and see how easy this turns out to be.

So the whole stepping-stone-concept of Garnet, and all the tech support, will become less relevant since even OS6 apps need a recompile to work correctly? I truly hope for their sake that it really is a "simple recompile"....

OS6 is fast approaching vaporware status IMHO, and to me it makes palmsource lose credibility. I really hope that in the end palmsource has something to show for all of this.

[sbl]

On the PalmOS side however, we have some applications that are properly coded that can work on PalmOS 4, 5 and 6. Most applications released today should work on OS5, 6 and the Linux version. If it can't, a simple update would solve the problem unlike on the Windows Mobile platform which is totally incompatible (last I checked).

We have been developing cross-platform applications for Palm, Symbian and Windows Mobile since 1999, and our products has been used in hundreds of mobile applications.

We have experienced no issues moving between Pocket PC 2000, 2002, 2003 and 2003 SE on the Windows Mobile Platform. Sure, there are some new features available in the latest versions, but checking for them and using them if they are available is no more different than enabling Shader Model 3.0 on Stationary PCs if it's available. We have been digging quite deep into Windows Mobile, including low-level networking and graphics, and the platform is in my personal opinion the best out there - Palm and Symbian not even come close.

Also, IMHO current Palm operating systems are a huge mess and the sooner they scrap it all and build from scratch the better. Currently trying to get an application that runs on all versions of Palm OS and the various devices that all use different customized APIs is a nightmare to maintain.

[Ed Hansberry]

We have been developing cross-platform applications for Palm, Symbian and Windows Mobile since 1999, and our products has been used in hundreds of mobile applications.

We have experienced no issues moving between Pocket PC 2000, 2002, 2003 and 2003 SE on the Windows Mobile Platform. Sure, there are some new features available in the latest versions, but checking for them and using them if they are available is no more different than enabling Shader Model 3.0 on Stationary PCs if it's available. We have been digging quite deep into Windows Mobile, including low-level networking and graphics, and the platform is in my personal opinion the best out there - Palm and Symbian not even come close.

Also, IMHO current Palm operating systems are a huge mess and the sooner they scrap it all and build from scratch the better. Currently trying to get an application that runs on all versions of Palm OS and the various devices that all use different customized APIs is a nightmare to maintain.
I have heard this from a number of people on a number of sites. Why does Palm keep perpetuating this myth that OS4/5/6 are interchangable from an app standpoint? And worse, why do some Palm apologists help perpetuate it? IMHO, PalmSource has really made a mess going from 4 to 5 and almost to 6. Now to put it all on Linux? Just how powerful will devices have to be to run an OS4 app, which will be in a compatibility environment on OS5/6 that itself will be running inside of Linux?

[reidme]

I think PalmSource has done an excellent job of hiding the real motivation for this move: outsourcing. "This acquisition will almost double our software development resources" translates "now we can lay-off lots of our highly-paid US employees."

[Ed@Brighthand]

Backwards compatibility is a hard but largely solvable problem. However, the idea of simultaneously maintaining both versions of the OS presents an even bigger problem of "sideways" compatibility.
It's even worse than that. PalmSource is already supporting -- and continuing development on -- two versions of the Palm OS: Garnet and Cobalt. The Linux version makes three.

[felixdd]

Does anyone think this move by PalmSource will make it more likely for PalmOne to run alternative OSs?

[Jonathon Watkins]

Does anyone think this move by PalmSource will make it more likely for PalmOne to run alternative OSs?

Heck, they are already doing three OSs. One more (PPC) should prove no problem! :wink:

[Reggie]

I'd like to sneak in the first podcast of 1SRC.com. Jeff Kirvin talks about his views on the Palm-Linux OS:

http://www.1src.com/scripts/show/707-1SRC_Podcast_One.html

[Deslock]

Why does Palm keep perpetuating this myth that OS4/5/6 are interchangable from an app standpoint? They said 80% off apps would carry over; that appears to be in the correct ballpark. I had >50 apps on my old Sony Clie T665C when I upgraded to the T3; while a bunch of system hacks couldn't be used, only 1 application gave me trouble... if you look on Brighthand you'll see that most users had a pretty easy time with the OS changes.

Also, while many CE applications work fine for all versions of PPC, there are also several that needed to be updated for OS revisions. Some here keep glossing over such problems. It probably stems from posts like
I guess the difference is, I don't want a new OS every few years that effectively renders apps obsolete or they have to be recompiled.
The first part of that statement "I guess the difference is" implies that changes to PPC haven't rendered apps obsolete. The reality is (like with POS) that's true for most PPC apps, but not for others.
2003 gave us the stability and speed of CE 4.0, faster PIE
Are you saying CE4 is more stable and faster or less stable and slower? Either way, it's still sluggish and temperamental. As far as PIE goes, are you saying it used to be even slower than what we're using now? That's sad... I wish Microsoft would just license NetFront, update it for VGA, and be done with it.

Back to the article, this would've been a great move 2 or 3 years ago, but at this point does it matter? POS6 is over a year late and is still not being used, Sony's pulled out of the US market, Palm has no new devices with wifi, the PDA market has shrunk (as a percentage of handheld devices), the Treo is a big hit in the USA but is not doing well elsewhere, PPC has overtaken Palm in terms of marketshare, and CE/PPC maintains its lead in terms of ease of software development.

[hackbod]

Hi all,

I'd like to post a few corrections and clarifications about what this all means.

"PalmOS for Linux" means all of the Palm OS system services and APIs (as they appear in Cobalt) running on top of the Linux kernel, very similar to MacOS X in which Apple's UI frameworks and higher-level services run on top of Mach. For the end user, this should mean basically no visible change -- the end-user really doesn't care about the kernel the OS is running on, and in fact with Cobalt we are already on our third kernel for the system (Linux will be the fourth).

The current Cobalt system does -not- use the BeOS kernel. It uses a kernel that was developed in-house at Palm. It does include many frameworks that were brought over from the Be technology, all of which we expect to continue to use when running on top of Linux. (Those frameworks were original implemented on BeOS, ported to the current kernel, and have on and off been running on Windows as well, so they are quite portable.)

The vast majority of the work we have already done in Cobalt are completely applicable to the Linux kernel -- all of the work redesigning the traditional PalmOS APIs and system architecture for a modern, protected-memory, multi-threaded operating system are needed just the same to run on Linux. If we had started working on Cobalt with Linux three years ago, it would have had very little impact on the time it took us to complete that OS because by far the majority of the work we had to do was in these higher-level pieces and not the kernel itself.

(If you want an analogy, you can compare us to Apple: OS 5 was equivalent to Apple's switch to PPC, and Cobalt was in some ways equivalent to the move to MacOS X, without exposing a lot of the new system features we are working on. For Cobalt as it stands today a better analogy is probably Microsoft's move to the first versions of Windows NT.)

Application compatibility from the current Cobalt to "PalmOS for Linux" should in fact be much better than it is going from Garnet to Cobalt. The move to Cobalt will have little impact on the application-level services, where-as significant work was required on them to move over to the modern OS and platform that we now have in Cobalt. The OS 5 and OS 6 transitions are really the two that have difficult compatibility issues: the first because of the switch to a completely different CPU (but still using the same system architecture), and the second because it required a redesign and reimplementation of most of the PalmOS system services. Changing the underlying kernel, given comparable capabilities (processes, protected memory, threads) generally has far less impact.

This announcement has nothing to do with our satisfaction with the current state of the Cobalt OS and technology. We fully expect a number of devices to ship with it, and for our licensees this announcement should have little impact on their current plans. This is a longer-term strategy, which will impact to some degree their associated long-term plans, but it doesn't have much relevance to current work on Cobalt -- basically that is all still applicable, except for things like drivers (but licensees tend to write new drivers for each device, anyway).

----------------------
Dianne Hackborn
Frameworks Manager at PalmSource
[email protected]