Palm gets its OS back

"Palm Signs Perpetual License for Palm OS Garnet Source Code" -- Palm press release

Now the circle is complete.

Way back in the days before the Internet bubble, Palm was a single integrated company run by its founders, making its own hardware and OS.

Today, Palm is once again a single company, run by its founders, with its own hardware and OS.

If it weren't for Eric Benhamou being on Palm's Board of Directors, you could almost pretend the last eight years didn't happen.

The details of the license agreement, explained in an admirably detailed Q&A posted by Access, raise some interesting possibilities. Here's what I read into them (and this is just my speculation; I don't have any inside information):

--Palm OS on Windows? Most important, Palm has the right to put Palm OS Garnet on any other operating system. As others have pointed out, that means they could create a new Linux-based device and run the Palm OS applications base on it. I believe they can also put Palm OS Garnet on Windows Mobile, which is going to turn the stomachs of many Palm OS enthusiasts but is extremely interesting to me.

Palm doesn't have a Palm OS-compatible 3G phone today for the GSM countries. In those countries, it can offer only Windows Mobile. But Palm can now theoretically offer Palm OS on top of its Windows devices. There are drawbacks -- the most prominent being that Palm uses 240x240 screens on its Windows Mobile devices. So we'd need new hardware, or some sort of awkward resolution hack. But I'll bet that Palm can still do that faster than it can rewrite Palm OS itself to run on 3G GSM.

I don't know what Microsoft would say about that. Probably something unhappy; they wouldn't like being treated as plumbing for someone else's OS.

Palm could also do Palm OS on Symbian, which might be less unappetizing than you'd expect. I think you'd completely hide the underlying Symbian OS, using it just as plumbing and phone management while you let Palm OS handle the UI and applications layer.

The key task for Palm will be finding a way to get all the basic phone software support for as little cost as possible, so they can concentrate on the value-added user functionality.
Palm can now play the field and choose whichever plumbing it likes best. It's a pretty liberating thought to me, and I bet it feels that way to Palm as well.

--Is it a full divorce? Palm and Access pointedly didn't say if Palm will use the new Access Linux Platform. It's still possible they might do it, but it's also possible that this agreement is the final divorce settlement between the two companies.

--Garnet has legs. I have a deep sentimental attachment to the Palm OS Garnet code base. The OS has its limitations, but for basic applications you can get a lot done with very little programming effort, especially compared to a nightmare case like writing native Symbian apps. Palm will apparently now start adding new features to Garnet, which is great (although it does create the risk of fragmenting the code base).

We now have three companies putting various levels of investment into Palm OS Garnet: Palm itself, Access, and StyleTap. An interesting situation for a dead, obsolete OS.

--Will Access make Palm OS Garnet into a layer? Now that Palm's using Palm OS Garnet as a software layer in top of other things, will Access license other companies to do the same? I have seen no signs that they will, but I think a Palm apps layer could be a lot more useful on mobile devices than Java has been. I argued for this for years within PalmSource, and I still think it's a good idea.

Anyway, I'm sure the old-time Palm engineers are glad to have their code base back, and I'll be very interested to see what they do with it.


Anonymous said...

The old time native Palm developers like me are moving to J2ME/Java and Flash Lite for our applications. Sun's recent complete open sourcing of everything Java and Adobe's nearly-the-same-move with ActionScript show me that the OS doesn't matter anymore. Other tools like J2ME Polish are making it easier to build apps across devices. It's way more important to me that I have reach to more devices.

Using your own analogy in your next posting, do you really even know what kind of engine is in your car? Do you care?

Michael Mace said...

Anonymous wrote:

>>do you really even know what kind of engine is in your car? Do you care?

Good question!

The answer is no, unless I'm an enthusiast.

To overstress the analogy a bit, if the engine = the OS and the applications = the gas, then what I care about is that all the engines run the same gas.

In other words, if Java + Flash + Apollo + whatever turns out to be a consistent platform that lets apps run across all mobile data devices without modification, I'll be happy.

Open-sourcing something doesn't necessarily produce that sort of consistency (in fact, it can lead in the other direction if it's not managed right).

How does it look from your perspective? Do you think we're headed toward a Java/Flash environment that's consistent across devices?

Anonymous said...


I thought that I'd clear up a misconception about 240 x 240 screen resolution.

This resolution is included in the Palm OS architecture and does not require any hack. For the most part, most applications will run fine without any changes. To optimize the appearance of their applications, most developers only need to add 1.5 density bitmaps and custom fonts (if the application uses them) in the same way that they added double density bitmaps/custom fonts for 320 x 320 devices.

We can tell you this from experience since thousands of users are running their Palm OS applications on Windows Treos using StyleTap today.

A lot of the developers (such as Planetarium) we are working with are adding or have already added the necessary 240 x 240 and 480 x 480 support (also architected) so this should not really be a big issue.

Bob Chew
StyleTap Inc.

Michael Mace said...

Thanks, Bob! I'd forgotten those resolutions were supported.

PDA_Guy said...

Hey Nice Article and my thoughts exactly.. but next time you ought to give credit from where you originally got this idea from =)

Michael Mace said...

I'd be very surprised if Palm chose to standardize all its stuff on only Windows CE. I believe they would probably do one other platform as well. The choices are:

-Access Linux.
-Some other Linux.
-Something else they write in-house.

It'll be interesting to see what they do.

Anonymous said...

Here are my thoughts on what Palm may do.

1) A Builtin PalmOS layer shipped with all WinMobile devices, would be a key distinguishing feature in the crowded winmobile space

2) There is a real OS underneath the "Palm OS" PACE. I expect that Palm's license includes source code to this layer. So Palm is free to extend this underlying OS layer to add such things as protected memory, and an API framework.

3) No Symbian. There is nothing to be gained here. Nokia will always have the best Symbian devices.

4) Linux. I think this is possible. The main concern here is intellectual Property rights. Basically I don't think Palm wants to develop a mobile OS that can simply be copied by competitors. (The bluetooth, ip, graphics, and phone sacks would all have to touch the kernel. GPLed kernels are a tricky thing.) There are ways to get around this by developing an open source kernel connector plugin architecture, but it is not preferred.