Insider reveals: What Does "Palm OS for Linux" Mean?

Bob Russell · 12-16-2004, 12:12 AM

What Does "Palm OS for Linux" Mean?
Dianne Hackborn
Manager, Palm OS Platform Frameworks

After the seeming bombshell announcement from PalmSource that there will be future versions of Palm OS Cobalt running on top of Linux, I have seen a lot of confusion over what this actually means, particularly to the end-user. This article will explain what you can expect and provide some technical background on how we expect to accomplish this.

Before digging into the technical details, I should say that, in a nutshell, as a user you can expect the experience of a future Palm OS Linux-based device to be basically the same as what we have now in Palm OS Cobalt. As an analogy, you can look at Mac OS X: though it sits on top of another open-source Unix operating system (in this case Mach/BSD), that is not the part a user directly sees or experiences. Likewise, in the future when Palm OS is running on Linux, the overall platform will still provide the same basic user experience and programmer APIs that we have now.

One reason for the confusion about what this announcement means is that there is just a lot of confusion over what “Linux” means. People commonly use the name “Linux” to refer to the overall platform they experience: this includes the kernel, windowing system, UI framework, multimedia components, applications, etc. However, technically Linux is only the kernel portion of this full platform, and that is the part we are generally referring to as “Palm OS for Linux”.

The kernel itself provides only the basic parts of an overall platform. It is roughly responsible for managing processes and enforcing memory protection and security between them, managing threads and providing synchronization primitives for them, providing basic IO such as “write this byte to the serial port”, providing controlled access to hardware and defining how one writes drivers for that hardware, and defining how persistent storage is structured and accessed. Everything else is the responsibility of software on top of the kernel. For example, to show something on the screen, some software outside of the kernel may ask that kernel for access to the screen, and then it is free to modify the pixels on the screen however it wants. Even a seemingly fundamental thing like the shell is not part of the kernel itself – which is why there are so many different shells available on Linux!

As it turns out, modern kernels share a lot of similarities when viewed from the outside. The kernel we currently use in Palm OS Cobalt (which was written in-house by PalmSource) is itself a modern kernel, and a large amount of our effort in developing Palm OS Cobalt was spent in re-designing and implementing the overall Palm OS platform architecture so that it would work well on top of a kernel with protected memory, threads, and other modern features. The vast majority of that work is directly relevant to Linux as well, and will carry over as-is.

Building “Palm OS for Linux” means, in essence, porting all of our existing platform frameworks from the current Palm OS Cobalt kernel to the Linux kernel. This includes such familiar things as the Exchange Manager, Window Manager, Form Manager, Event Manager, Data Manager, as well as the numerous internal frameworks (Binder, the Media Framework, view hierarchy and graphics system) that they are built on. Some of the lower-level parts of the system (such as Palm OS Cobalt’s new SysThread.h APIs and the low-level parts of the application manager) will need to be rewritten in this port, but that will be the unusual case.

In fact, Palm OS has already been running on three different kernels through its lifetime: a licensed kernel for 68k versions of the system, a new kernel used in 5.x, and finally the kernel we currently use in Palm OS Cobalt. All 68k applications should run as-is on this new platform just as well as they run today on Palm OS Cobalt. However, we have announced that Protein applications (which are written for the native ARM APIs in Palm OS Cobalt) may require a recompile to run on “Palm OS for Linux.” This statement was made because Linux typically uses a different approach to calling platform library APIs than we use in Palm OS Cobalt, and it is too early for us to be able to promise more compatibility than that.

Ultimately what this all means is that our users and developers will still have Palm OS much like it exists today in Palm OS Cobalt. This is a long-term project, and we expect to see a variety of devices using the current version of Palm OS Cobalt out in the market in the coming year. The move to Linux will be a progression of our overall development of the Palm OS Cobalt platform.

One final question I would like to address is if this means one will be able to run Linux applications on Palm OS. In theory, when the underlying Linux kernel is there, basic applications should be able to compile for it with little to no change. You need to aware, however, that the higher-level parts of the platform will be Palm OS. This means, for example, that X Windows will not be running on your handheld device – we will be using the Palm OS graphics and windowing system instead. It will be possible, with various amounts of work, to get these other software components from the open source community to work on our platform, but this is not something in which PalmSource itself will likely invest much time.

================================================== ==

Editor's note: MobileRead is delighted to present this article by Dianne Hackborn from PalmSource, which is the unexpected benefit of a recent exchange of emails. I first discovered Dianne over at Palm Infocenter where she grabbed my attention with some great information and the sanity she brought into the forum chatter about PalmOS devices and the future of PalmSource. I knew immediately that we all needed to hear more from her, and she was kind enough to respond by providing this information.

With the recent slowdown in product announcements, plenty of speculation and rumors have started to surface. And as impatient consumers and hobbyists, surely it doesn't take much to convince us that the sky is falling. We have very little to go on, so it's a joy to hear from someone who has real information. Believe me, the more information and detail you get, the more it all makes sense and the more you will be excited about the future.

There are admittedly challenges ahead for PalmOne and PalmSource, but maybe what we need is just a bit more patience. Two very significant investments have been announced towards the future of all things "Palm" -- the Cobalt OS (currently in the hands of device makers), and the Palm OS for Linux (which commitment is evidenced in the acquisition of CMS, a software company that makes Linux-based mobile phones.) In some big ways, we are seeing the "cost" now in a slowdown in product announcements, which is exaggerated by the withdrawal of Sony from the US PDA market. But we yet hope to reap the benefits for years to come over the aging Garnet internals.

Palm/PalmSource didn't succeed without taking some bold steps in the past, and personally, I'm glad they have the guts to continue innovating. Those who play it safe in a competitive market will die. But no matter how wonderful and forward-thinking Cobalt may be, we will not be convinced it's real until we hold a device in our hands. And we will probably not fully appreciate the rejuvenation of the internals until we start to see those improvements in the externals, i.e. in handhelds and software.

It's been a tough wait for all Palm fans, even for those at PalmSource. I've been one of the more vocal "worryers" myself. But there is every reason to believe devices are on their way, and after hearing some of the input from Dianne, I'm much more encouraged about the future of Cobalt.

I also want to remind everyone that we need to be careful not to jump to conclusions from insider comments about early versions of Cobalt. After "completion" of the OS, there was much work done to fine tune performance and compatibility for final release form. In addition, there is a balancing act between satisfying particular vendors enhancement requests and doing new features development. Cobalt is alive and maturing, and is prepared for the future.

I hope that as you read this insider information, that you will put aside some of your preconceived notions and any disappointments, and rekindle a bit of the excitement that so many of us have felt through the years with Palm. You may just find that there's a lot of pleasant surprises in store for PalmOS.

No matter what OS preferences you have, or what your views of the "best" path for PalmSource and PalmOne, I think we can all agree on one thing -- we are all cheering for Palm to succeed, and the more they succeed, the more the entire mobile computing world benefits.

BobR
MobileRead Editor

vrf · 12-16-2004, 03:32 AM

Quote:

Originally Posted by BobR

What Does "Palm OS for Linux" Mean?
You can expect the experience of a future Palm OS Linux-based device to be basically the same as what we have now in Palm OS Cobalt. [/I]

As in non-existent?

doctorow · 12-16-2004, 05:30 AM

Quote:

Originally Posted by BobR

[B]One final question I would like to address is if this means one will be able to run Linux applications on Palm OS. In theory, when the underlying Linux kernel is there, basic applications should be able to compile for it with little to no change. You need to aware, however, that the higher-level parts of the platform will be Palm OS. This means, for example, that X Windows will not be running on your handheld device – we will be using the Palm OS graphics and windowing system instead. It will be possible, with various amounts of work, to get these other software components from the open source community to work on our platform, but this is not something in which PalmSource itself will likely invest much time.

... but surely Linux folk will invest more time

In other words, Linux apps such as mplayer video player could be possible for Palm OS. The only part that needs to be rewritten is the output module (to use the "Palm OS graphics and windowing system" instead of X Windows). That's amazing! Now I only wish I was better at programming...

Alexander Turcic · 12-16-2004, 05:35 AM

Mrs Hackborn, Bob, thank you for this PalmOS/Linux scoop. Not only does it reveal more indepth information about how palmSource plans to integrate Palm OS with Linux, but also does it make palmSource's decision sound more and more reasonable. Now I am really looking forward to seeing the first Palm OS-Linux PDAs!

Brad · 12-16-2004, 12:49 PM

Thanks for posting that. It makes it a lot clearer to me.

A few months ago I was looking at Sharp Zaurus' PDA's (before they discontinued US sales) sort of wishing they would work with Mac OS X. Now it sounds like I will one day be able to play with a Linux based PDA and yet still retain the familier Palm GUI. Sounds good to me so long as somebody makes it all work with Mac.

superdork · 12-16-2004, 02:01 PM

I'm hoping to see some new devices soon. I've tryed the new Axim X50V and have to say the hardware rocks, but Windows Mobile is abysmall!!
I have been waiting patiently for native arm application support for better video playback as what's there now is paltry compared to Windows Mobile.
MMPlayer is great, but without native arm support there is only so much you can do with the all that horse power.
I'm hoping for a device with a thumboard and decent mpeg4 playback soon.
Thanks for a great site and the above article.

Pride Of Lions · 12-16-2004, 02:46 PM

Quote:

Originally Posted by Brad

A few months ago I was looking at Sharp Zaurus' PDA's (before they discontinued US sales) sort of wishing they would work with Mac OS X. Now it sounds like I will one day be able to play with a Linux based PDA and yet still retain the familier Palm GUI. Sounds good to me so long as somebody makes it all work with Mac.

Hopefully the folks over at Mark/Space are hip to this news and are furiously developing the means to have Mac users enjoy these future PDA's. I don't think that they even tried to develop a "Missing Sync" for the Zaurus, but with PalmSource moving towards Linux, hopefully this will move theMark/Space's programmers in that direction.
POL9A

hacker · 12-16-2004, 05:10 PM

Quote:

Originally Posted by Pride Of Lions

Hopefully the folks over at Mark/Space are hip to this news and are furiously developing the means to have Mac users enjoy these future PDA's. I don't think that they even tried to develop a "Missing Sync" for the Zaurus, but with PalmSource moving towards Linux, hopefully this will move theMark/Space's programmers in that direction.
POL9A

MarkSpace's flagship product, MissingSync, is based on pilot-link, code (which I maintain) that works on all POSIX platforms (Linux, UNIX, OSX, and others), there is no reason why they couldn't continue to use their front-end on top of another non-PalmOS-specific set of libraries, such as those used by the Zaurus products now, or MultiSync, SyncCE, KitchenSync (or combinations of the above) to continue to sync their devices.

12-16-2004, 12:12 AM	#1
Bob Russell Recovering Gadget Addict Posts: 5,381 Karma: 676161 Join Date: May 2004 Location: Pittsburgh, PA Device: iPad	Insider reveals: What Does "Palm OS for Linux" Mean? What Does "Palm OS for Linux" Mean? Dianne Hackborn Manager, Palm OS Platform Frameworks After the seeming bombshell announcement from PalmSource that there will be future versions of Palm OS Cobalt running on top of Linux, I have seen a lot of confusion over what this actually means, particularly to the end-user. This article will explain what you can expect and provide some technical background on how we expect to accomplish this. Before digging into the technical details, I should say that, in a nutshell, as a user you can expect the experience of a future Palm OS Linux-based device to be basically the same as what we have now in Palm OS Cobalt. As an analogy, you can look at Mac OS X: though it sits on top of another open-source Unix operating system (in this case Mach/BSD), that is not the part a user directly sees or experiences. Likewise, in the future when Palm OS is running on Linux, the overall platform will still provide the same basic user experience and programmer APIs that we have now. One reason for the confusion about what this announcement means is that there is just a lot of confusion over what “Linux” means. People commonly use the name “Linux” to refer to the overall platform they experience: this includes the kernel, windowing system, UI framework, multimedia components, applications, etc. However, technically Linux is only the kernel portion of this full platform, and that is the part we are generally referring to as “Palm OS for Linux”. The kernel itself provides only the basic parts of an overall platform. It is roughly responsible for managing processes and enforcing memory protection and security between them, managing threads and providing synchronization primitives for them, providing basic IO such as “write this byte to the serial port”, providing controlled access to hardware and defining how one writes drivers for that hardware, and defining how persistent storage is structured and accessed. Everything else is the responsibility of software on top of the kernel. For example, to show something on the screen, some software outside of the kernel may ask that kernel for access to the screen, and then it is free to modify the pixels on the screen however it wants. Even a seemingly fundamental thing like the shell is not part of the kernel itself – which is why there are so many different shells available on Linux! As it turns out, modern kernels share a lot of similarities when viewed from the outside. The kernel we currently use in Palm OS Cobalt (which was written in-house by PalmSource) is itself a modern kernel, and a large amount of our effort in developing Palm OS Cobalt was spent in re-designing and implementing the overall Palm OS platform architecture so that it would work well on top of a kernel with protected memory, threads, and other modern features. The vast majority of that work is directly relevant to Linux as well, and will carry over as-is. Building “Palm OS for Linux” means, in essence, porting all of our existing platform frameworks from the current Palm OS Cobalt kernel to the Linux kernel. This includes such familiar things as the Exchange Manager, Window Manager, Form Manager, Event Manager, Data Manager, as well as the numerous internal frameworks (Binder, the Media Framework, view hierarchy and graphics system) that they are built on. Some of the lower-level parts of the system (such as Palm OS Cobalt’s new SysThread.h APIs and the low-level parts of the application manager) will need to be rewritten in this port, but that will be the unusual case. In fact, Palm OS has already been running on three different kernels through its lifetime: a licensed kernel for 68k versions of the system, a new kernel used in 5.x, and finally the kernel we currently use in Palm OS Cobalt. All 68k applications should run as-is on this new platform just as well as they run today on Palm OS Cobalt. However, we have announced that Protein applications (which are written for the native ARM APIs in Palm OS Cobalt) may require a recompile to run on “Palm OS for Linux.” This statement was made because Linux typically uses a different approach to calling platform library APIs than we use in Palm OS Cobalt, and it is too early for us to be able to promise more compatibility than that. Ultimately what this all means is that our users and developers will still have Palm OS much like it exists today in Palm OS Cobalt. This is a long-term project, and we expect to see a variety of devices using the current version of Palm OS Cobalt out in the market in the coming year. The move to Linux will be a progression of our overall development of the Palm OS Cobalt platform. One final question I would like to address is if this means one will be able to run Linux applications on Palm OS. In theory, when the underlying Linux kernel is there, basic applications should be able to compile for it with little to no change. You need to aware, however, that the higher-level parts of the platform will be Palm OS. This means, for example, that X Windows will not be running on your handheld device – we will be using the Palm OS graphics and windowing system instead. It will be possible, with various amounts of work, to get these other software components from the open source community to work on our platform, but this is not something in which PalmSource itself will likely invest much time. ================================================== == Editor's note: MobileRead is delighted to present this article by Dianne Hackborn from PalmSource, which is the unexpected benefit of a recent exchange of emails. I first discovered Dianne over at Palm Infocenter where she grabbed my attention with some great information and the sanity she brought into the forum chatter about PalmOS devices and the future of PalmSource. I knew immediately that we all needed to hear more from her, and she was kind enough to respond by providing this information. With the recent slowdown in product announcements, plenty of speculation and rumors have started to surface. And as impatient consumers and hobbyists, surely it doesn't take much to convince us that the sky is falling. We have very little to go on, so it's a joy to hear from someone who has real information. Believe me, the more information and detail you get, the more it all makes sense and the more you will be excited about the future. There are admittedly challenges ahead for PalmOne and PalmSource, but maybe what we need is just a bit more patience. Two very significant investments have been announced towards the future of all things "Palm" -- the Cobalt OS (currently in the hands of device makers), and the Palm OS for Linux (which commitment is evidenced in the acquisition of CMS, a software company that makes Linux-based mobile phones.) In some big ways, we are seeing the "cost" now in a slowdown in product announcements, which is exaggerated by the withdrawal of Sony from the US PDA market. But we yet hope to reap the benefits for years to come over the aging Garnet internals. Palm/PalmSource didn't succeed without taking some bold steps in the past, and personally, I'm glad they have the guts to continue innovating. Those who play it safe in a competitive market will die. But no matter how wonderful and forward-thinking Cobalt may be, we will not be convinced it's real until we hold a device in our hands. And we will probably not fully appreciate the rejuvenation of the internals until we start to see those improvements in the externals, i.e. in handhelds and software. It's been a tough wait for all Palm fans, even for those at PalmSource. I've been one of the more vocal "worryers" myself. But there is every reason to believe devices are on their way, and after hearing some of the input from Dianne, I'm much more encouraged about the future of Cobalt. I also want to remind everyone that we need to be careful not to jump to conclusions from insider comments about early versions of Cobalt. After "completion" of the OS, there was much work done to fine tune performance and compatibility for final release form. In addition, there is a balancing act between satisfying particular vendors enhancement requests and doing new features development. Cobalt is alive and maturing, and is prepared for the future. I hope that as you read this insider information, that you will put aside some of your preconceived notions and any disappointments, and rekindle a bit of the excitement that so many of us have felt through the years with Palm. You may just find that there's a lot of pleasant surprises in store for PalmOS. No matter what OS preferences you have, or what your views of the "best" path for PalmSource and PalmOne, I think we can all agree on one thing -- we are all cheering for Palm to succeed, and the more they succeed, the more the entire mobile computing world benefits. BobR MobileRead Editor

12-16-2004, 02:01 PM	#6
superdork Junior Member Posts: 2 Karma: 10 Join Date: Dec 2004 Device: Clie UX50	Insightful stuff I'm hoping to see some new devices soon. I've tryed the new Axim X50V and have to say the hardware rocks, but Windows Mobile is abysmall!! I have been waiting patiently for native arm application support for better video playback as what's there now is paltry compared to Windows Mobile. MMPlayer is great, but without native arm support there is only so much you can do with the all that horse power. I'm hoping for a device with a thumboard and decent mpeg4 playback soon. Thanks for a great site and the above article.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Installing a "normal" Debian ARMEL Linux on the Kindle	choff	Kindle Developer's Corner	18	09-09-2014 03:42 AM
"Unable to detect PRS505 disk drive" in Linux	sarchar	Sony Reader	5	01-25-2009 04:52 PM
Silicon Alley Insider: "Kindle A Year Old, Hasn't Changed Reading... Yet"	curtw	News	1	11-20-2008 06:49 PM
SimpleMepis - "Best Free Desktop Linux"	Colin Dunstan	Lounge	0	10-31-2004 07:44 AM
Installing "Send to Palm" plug-in in Adobe Reader 6?	daught	Reading and Management	8	06-02-2003 07:15 PM

12-16-2004, 05:35 AM	#4
Alexander Turcic Fully Converged Posts: 18,171 Karma: 14021202 Join Date: Oct 2002 Location: Switzerland Device: Too many to count here.	Mrs Hackborn, Bob, thank you for this PalmOS/Linux scoop. Not only does it reveal more indepth information about how palmSource plans to integrate Palm OS with Linux, but also does it make palmSource's decision sound more and more reasonable. Now I am really looking forward to seeing the first Palm OS-Linux PDAs!

12-16-2004, 12:49 PM	#5
Brad Zealot Posts: 107 Karma: 201 Join Date: Sep 2004 Device: Cybook Gen3 / EB1150 /iPhone	Thanks for posting that. It makes it a lot clearer to me. A few months ago I was looking at Sharp Zaurus' PDA's (before they discontinued US sales) sort of wishing they would work with Mac OS X. Now it sounds like I will one day be able to play with a Linux based PDA and yet still retain the familier Palm GUI. Sounds good to me so long as somebody makes it all work with Mac.

Advert

Advert