The potential synergy between the GNU/Linux OS and the Palm platform is huge, but so are the potential pitfalls. Larry Garfield looks at both the pros and cons of the Next Next Palm OS.
Editorial: The Penguin, The Palm, and the PDA
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
I thing one of the biggest advantages will be the ability the port a new platform more quickly, unlike the long time it took them the get PalmOS to the ARM platform. Who knows, PowerPC might emerge a viable PDA CPU, and a linux based PalmOS would likely be able to be ported more quickly.
*yawn* Oh yeah I love having my phone crash on me 3 times a week, sometimes taking all my data with it into oblivion. I just wish I could ignore such a stupid troll….
I think this is great news. I’m not sure Palmsource will succeed but there is huge potential. Unlike the PC market, there isn’t yet so much vendor lock-in present with mobile platforms. Hopefully they can start shipping before lock-in does settle in.
>There is no ‘synergy’, as the author puts it. PocketPC and >Windows will ALWAYS tromp *nix in the arena of end user >usability, graphics, sound and innovation.
lol. stolen ui, kernel, etc….
Article: “Despite the rosy marketing picture, however, there are major hurdles to a Palm/Linux marriage. Linux and Palm OS are extremely different, architecturally. The former uses bitstream files, hierarchical directories, and other Unix-style concepts. The latter uses native databases, flat-namespace data storage, and eXecute-In-Place for near-zero program load time. Making those two talk to each other is very non-trivial, and PalmSource has of course been mute on how it will accomplish this amazing feat. “
These are the kind of things that an emulator can handle easily. You can run PalmOS on the Zaurus, for example. That said, they will probably want to skip emulation and do it the right way for performance and other reasons, though emulation at some level is not out of the question.
* Linux uses Unix file system concepts; everything is a file and every file is a stream of bits. No conflict here!
* Data and file system formats can be mapped on to any of the existing native Linux file systems or Palm can port what they have to Linux and use it unchanged. (I’d expect them to create a new file system or to extend one of the existing ones. Cramfs support would probably be a good idea as an option.)
* ‘eXecute-In-Place’ is tougher, though I’d be surprised if the ucLinux folks haven’t delt with this on some level already. (ucLinux runs on Palm hardware already and does not require a MMU. ucLinux does require a hardware modification on the current Palm hardware, though. Less of an issue if you have some control over the hardware!)
Article: “Of more concern is a fact that PalmSource has only been muttering under its breath; Garnet and Cobalt programs will run on Palm OS for Linux with only a “simple recompile”. Assuming for a moment that it really is that simple, that still means no binary compatibility. No binary compatibility means, effectively, a different platform for developers to support, to say nothing of different accessory drivers. A fragmented platform (Garnet, Cobalt, Palm OS for Linux) is the last thing PalmSource needs right now, especially as Cobalt devices still have yet to appear.”
I’d be surprised if the recompile will be necessary. They can provide the same hooks or emulate them for the older apps, and allow unemulated support for the new ones.
Recompiling might be encouraged, though, to lessen emulation overhead and take advantage of additional features of Linux that will be likely exposed on the Palm environment side.
Palm could do what Apple did with MacOS apps and OSX — though they have much less of an issue.
Which has chosen not to follow the Linux Hype, what do we do? I for one want my phone to boot in 5 secs, not 5 minutes…. will they add in the Linux kernel, remove everything in it and build it from scratch and make it fast and then call it Linux????
Geez, I chose Palms devices because of PalmOS primarily. What possible advantage do Palm now have over competition? None? Correct…. this is simply so stupid and horrible I can’t imagine how they will make a win situation out of this.
The way I see it, Palm just signed their own death warrant.
Now I guess it’s time to move on to DangerOS and Blackberry!
Wow. Where to start…
* “[And for the 95% of the market ] … has chosen not to follow the Linux Hype, what do we do?”
Tivo. Few people who own one know or care that it runs Linux. No hype about Linux at all. They follow the licences for the software they use and modifiy. Tivo made a business decision, just like Palm has.
* “I for one want my phone to boot in 5 secs, not 5 minutes….”
Why would it take 5 minutes? See Linksys’s Linux based routers (WRT54G and others).
* “will they add in the Linux kernel, remove everything in it and build it from scratch and make it fast and then call it Linux????”
Linux is the Linux kernel. Everything else is an add-on. Palm wants to add on different things. See ucLinux and Linksys (mentioned above) for examples.
* “Geez, I chose Palms devices because of PalmOS primarily. What possible advantage do Palm now have over competition? None? Correct….”
You seem to have made up your mind. For the rest of us, it’s the layer ABOVE the OS — plus what the OS provides. The article covered that.
* “this is simply so stupid and horrible I can’t imagine how they will make a win situation out of this.”
OK. I can.
* “The way I see it, Palm just signed their own death warrant.”
Little dramatic, eh? :/
* “Now I guess it’s time to move on to DangerOS and Blackberry!”
Go ahead…that’s your choice. Is it wise to judge based on undelivered tech? Hmmmm…
I think all palm are going to do is have a linux kernel with some kind of *improved* <a href=”http://www.palmos.com/dev/tools/emulator/“>POSE emulator sitting on top. That way they have very little actual porting/rewring todo and all existing apps will still run. All ‘palmos’ will be is a single linux process.
Remember palmos needs very little cpu/mem to perform well – Even on modern palms most binaries are 68k, emulated on ‘fast’ ARM cpu’s – this is more or less what POSE does too.
Using a linux kernel underneath is a very cost effective way for them to support a wide range of hardware ‘for free’ and gives them many more possibilitys ( improved multi process ‘hacks’ in palmos for one ).
This move to linux from a company who recently bought Be is typically utter technical nonsense from the financial guys. It goes from nowhere and leads exactly nowhere.
“Yeah, let’s rewrite everything we already have from scratch”.
Irk, that’s so stupid I hardly believe it…
palmsource is a tiny company. Microsoft is huge. Symbian is owned by Nokia, Siemens, Motorola, Samsung, Panasonic, Ericsson and probably others.
Palm needs to stand on the should of this giant called linux so to level the playing field.
“This move to linux from a company who recently bought Be is typically utter technical nonsense from the financial guys. It goes from nowhere and leads exactly nowhere.”
Would it be smarter to stay with the Be kernel if it would cost them more in the long run? I don’t know if this is the case or not — though I doubt you do either!
“”Yeah, let’s rewrite everything we already have from scratch”. Irk, that’s so stupid I hardly believe it…”
Doesn’t look like a total re-write. Looks like they are only going to change the base OS not the application layer. That’s what the article talked about.
does not sharp (sharps?) allready ship pdas that run linux?
getting a treo device that runs linux will be interesting to say the least
and the only reason linux takes a bit to start on a desktop is that it does hardware detection every time, unlike windows that does it at install and then maybe when its fully booted into the desktop (rest of the time it guesses that you will never change, and bluescreens if you move the os drive to a diffrent hd controller). then there is the number of daemons or whatever you want to call it that starts up after the kernel have loaded. then you log in and get the desktop enviroment is fully loaded before you see it, windows on the other hand loads the basic explorer and then drops a bombshell by starting everything from antivirus controlpanels to small tray stuff while your looking at the desktop. personaly i prefer the linux way as it keeps telling what it does rather then hide it behind some loading bar and leaves me to sit and listen to the hd working…
Geez, I chose Palms devices because of PalmOS primarily.
Why in the world would you make a choice based on the OS and not the applications? The OS shouldn’t matter. It’s the apps that matter. The OS should be transparent and I’m sure that Palm will make it so with Linux.
I think that there are a number of shortcomings with going to linux, and they have to do less with the technical underpinnings and more with the cultural attitudes of the linux developer community.
One problem is that most linux developers think that for some reason you can simply take a desktop UI and shrink it down into a PDA UI. I’ve gotten pretty well chewed out on some linux PDA forums for questioning why some linux PDA UI’s had movable windows. If Palm goes linux, they will be letting in a lot of developers through their doors. Which is might be a good thing. The bad thing is that they are also going to let in the door a lot of people who don’t see the big deal in one application requiring twice as many stylus taps as another similar application and who are going to write-off many usability issues as “subjective” or “wanting it to be familiar like X application/OS/environment”. I don’t think that most die-hard linux programmers are going to be able to grok or accept something like the Zen of Palm. http://www.palmos.com/dev/support/docs/zenofpalm/ZenTOC.html
Another problem related to the first is the “with an simple compile” idea. A lot of people into the linux developer community think that something like a gnome or kde desktop mail application is going to easily be migrated to a PDA just because you’ve got stuff in the code that’s supporting the ARM/68k/whatever architecture. You’ve got a lot of factors such as screen size, input methods, and usage scenarios that won’t be easily ported with a simple compile. The only thing portability really buys you is the ability to use libraries, not actual applications. A well designed PDA application is built from the ground up to be run on a PDA, and no amount of portability is going to change that.
A third issue is with the technical stuff. From the limited stuff I’ve read about the way the PalmOS works, it prioritizes processes (if you can call them that) based on things that are done in the UI, and this is exactly what you want in something like a PDA. I don’t think that too many kernel hackers are going to be very enthusiastic about accepting a patch that makes this kind of engineering tradeoff.
What really needs to be changed, IMHO, is the PalmOS API. The memory management stuff as it currently is is a real pain in the butt and the API is really arcane. It’s sort of like classic MacOS programming. The one (and only) thing I liked about my zaurus was that it used the Qt API for stuff. It was unbelievably easy to create custom applications quickly using Python and Qt. If Palm could lower the bar for developer Palm applications so that anyone with a little programming experience could make a useful mobile application, I think that it would put the palm platform light years ahead of their competitors.
“Another problem related to the first is the “with an simple compile” idea. A lot of people into the linux developer community think that something like a gnome or kde desktop mail application is going to easily be migrated to a PDA just because you’ve got stuff in the code that’s supporting the ARM/68k/whatever architecture.”
I don’t think any embedded Linux developer is that naive.
They have experience in porting stuff and would re-write things with pda-specific = they will cut down features – or choose very-light widgets such as fltk.
They are aware that the code-once compile-anywhere doesn’t work. Everything needs tweaking – if not optimal editting – failing that – complete rewrites!
Even the kernel is compiled extremely slim and differently for each PDA and its architecture.
Having said all that I’ve seen a nice port of Firefox on a Psion Netbook which was ported to Linux 2.6.9-r3. Very nice.
and the only reason linux takes a bit to start on a desktop is that it does hardware detection every time, unlike windows that does it at install and then maybe when its fully booted into the desktop (rest of the time it guesses that you will never change, and bluescreens if you move the os drive to a diffrent hd controller). then there is the number of daemons or whatever you want to call it that starts up after the kernel have loaded. then you log in and get the desktop enviroment is fully loaded before you see it, windows on the other hand loads the basic explorer and then drops a bombshell by starting everything from antivirus controlpanels to small tray stuff while your looking at the desktop. personaly i prefer the linux way as it keeps telling what it does rather then hide it behind some loading bar and leaves me to sit and listen to the hd working…
First, linux does not take that long to boot. Second the hw detection the kernel does takes seconds at boot. What takes a long tim eis all the services it starts (as you said), and they only take a long time because they run one after the other. If they were started at the same time (taking dependencies into account) linux would boot as fast as windows XP, probably faster.
or they would churn the machine to a halt as they are swapped in and out of memory. allso, some of them may need to have a diffrent one started before itself can start.