Symbian's open source journey

newsmaker How do you open up a closed codebase? Symbian research chief David Wood weighs potential and practicality, and the promise of multicore phones.
Written by David Meyer, Contributor

newsmaker Symbian, the UK-based maker of the world's most popular smartphone operating system, is going through big changes.

As well as being taken over by Nokia, the company is preparing to convert its closed code into open source.

At the Symbian Smartphone Show in London in October, Symbian's research chief David Wood told ZDNet Asia's sister site ZDNet UK that the Symbian Foundation will be established in the first half of next year, with the first version of the Foundation software available in March or April 2009. He also discussed the complications of such a process, as well as what the next few years holds for smartphone technology.

Q: What can we expect from the first Symbian Foundation software? It won't be stripped of third-party code yet, will it?
Wood: Correct. That will be available only to people who join the Foundation and who sign up to the Foundation license. There will be some parts that are open source.

So the Foundation license is not the open-source license?
The Foundation license is very similar to the open source license, but it allows the companies to share the code only within the Foundation. It's a community source license, with as much as possible in common with the eventual [open source] license that will take over.

There is some code available as open source from day one, but completion [of the open sourcing] will be some time in 2010. It's a sensible engineering approach--a stage-by-stage release of the code.

I was speaking earlier to the chief executive of a software firm whose code is currently in Symbian. He said there was no problem in having some proprietary elements within open source code, and that this was acceptable under the GNU General Public License (GPL). That doesn't sound right...
We're not using the GPL--it's the EPL [Eclipse Public License]. The EPL is indeed able to link to proprietary software. The GPL is less clear. In fact, a straight reading of the GPL says if you link to other software then that other software falls under the same license. Under the EPL, if you link to other software then there's no obligation on that other software to take the same license.

EPL is weak copyleft, whereas GPL is the most famous example of strong copyleft. So I agree with that part, that there could be code that's linked to. This is to encourage innovation.

We're not saying all software should be free of charge. We do realize that there will always be new, interesting software that people will want to monetize by selling for a license. If you change the Symbian code, that has to be given back--you can't hang onto that, so that's the copyleft part of this message.

But there is code from this company within Symbian's code--won't that have to be scraped out?
Something has to be done, and I don't really want to talk about an individual case, but in principle several things could happen. We could throw money at a supplier, and we could say to them: "We will buy this off you in perpetuity and we will make it available." Or we could say we'll leave this outside the platform and we can put something else in instead. It won't be quite the same, and we might go back to the kind of offering that we had in previous versions of Symbian.

It's always possible that someone else will come along and do comparable software and make that available. There should be plenty of ways for companies [whose code is currently within Symbian's code] to recoup their investment, either by selling the software [to Symbian], or by developing a better version and making that available for an additional fee.

Can you give any indication of how many third-party players there are whose code is currently within Symbian's?
There are scores. We have numbers, but it's not clear how many of them are serious cases. In some cases they can be dealt with probably straight away, but scores could take some serious thought. Whether that's nearer 20 or 100, we need to investigate. There's something like 100 cases that we're looking at. In many cases, it looks like it's a trivial solution. In other cases, the software is in a class of its own.

What do you count as 'trivial'? Something that can be easily replicated?
Either where it's easily replicable, or where we are confident that we can change the licensing terms. It might be code that's currently under the GPL, so we might pass that through. In reality people will make a phone by taking this and adding in other things that are easily available. So there will be other stuff, GPL, floating around in the broader community. So we will say to people: "Right, you build a phone by taking this Symbian offering and adding in these additional components." Webkit, for example--that's currently under a GPL license.

There are things we might look at and say: "Well, this is easy to solve. It's not an integral part of the system. It will be available as part of what we call a distro." So people will combine what they get from Symbian with other things that are designed to slot in.

So we will see distributions of the Symbian core and some free plug-ins?
Yes, and other plug-ins that people might even pay money for. The point is, there are many business models that are possible. Just as Linux has given rise to many companies that do their labors with Linux, there will be many companies that do things on top of Symbian. We're not looking for different flavors of Symbian in the sense of changing the core, but there will be people who are, for example, specialists in software for navigation devices.

Of course, many of the smartphones are actually navigation devices in their own right, but you can imagine some people might make a navigation device which happens to be a smartphone, as opposed to a smartphone that happens to be a navigation device. There might be people who specialize in that kind of thing and say: "Here's Symbian's offering and we've got some extra navigation stuff that we add in to make us the right starting point."

A lot of manufacturers are likely to bring out Android handsets...
Lots of people will look at bringing out an Android handset. Let's wait and see what actually happens.

Symbian has scale and popularity but Android is starting from zero. It doesn't have to strip out code. For a developer, is there not a period of uncertainty at the moment, because they're not sure of what they're addressing? The message in the keynotes was that addressing Series 60 means addressing the next version of Symbian, but is that actually the case? Some bits may have to be stripped out...
I think that the proportion that might end up being changed in that way is very small. The vast majority of software that's written can be preserved.

For our developer readers, when is that moment of uncertainty going to pass?
It's going to be stage-by-stage. We have a roadmap. We will be sharing more information sometime around the middle of H1 2009. There will be a whole lot more information shared.

What innovations can we expect from Symbian between now and the next Smartphone Show, and how much will the open-sourcing process affect that innovation?
The first phone with Symbian OS 9.4 has been shown, which is the Nokia [5800] Tube device. That runs S60 5th edition. We have already released Symbian 9.5 to our customers,

and there is considerable progress on the next two releases, which you can imagine might be called 9.6 and 9.7. At some stage, they will fall into the new numbering system that will be used for the Symbian Foundation.

What will that numbering system look like?
It's still being discussed. I quite like the idea of staying at 9 forever, because it emphasizes compatibility. But the key thing is: there are two more releases for which we have a roadmap.

The main core feature for the next releases is twofold. One of them is support for symmetrical multiprocessing (SMP), which we believe is going to be really important for the future. That comes out in two phases: in the release after 9.5 we've made all the software SMP-safe, which means it won't fall over if there are suddenly more processors in there; and the one afterward is called SMP-optimized, which is when we actually restructure some of the software to run better when there are multiple cores. That will be the software that lands on phones with multiple cores, sometime around 2010.

What is the point in having SMP on a handset?
It will allow the phone to do more without running the batteries to the same extent, because the individual cores will run at a lower clock speed. It turns out, if you have two cores running at a lower clock speed, you can actually end up calculating more but using less power.

People will use this for all kinds of things; to take one example, real-time language translation. Currently most of the real-time language translation services on smartphones tend to rely on server-side work. So you might speak into it, it might send it off to the network and it comes back with a translation. Now, imagine if you could do more of that kind of calculation on the phone.

Then there's all the multimedia applications. Graphics never get poorer--there are more and more pixels and colours, and all of that requires oomph from the processors. If you can spread that out over multiple processors, it delivers a faster user experience and more functionality without running out of battery.

Does Symbian still believe the smartphone will take over from PCs, as it predicted two years ago? The input and display issues remain, and netbooks have come into the picture.
I don't think we ever said PCs would disappear. PCs will remain-—there will be a whole host of devices that remain. But [smartphones] will be more capable and people will be more comfortable using that will more and more features. Take the [Nokia E71]-—people are often surprised that it is actually quite easy to type into it even though the keys are so small. There's some very clever hardware design in that. It's part of the overall step-by-step improvements in input.

The Nokia Tube has got pen inputs and there's about five different ways people can choose to input data into that. Will all five be equally important? Probably not, but let the market decide which ones will be most important. That will make input easier. Also, because there are more pixels on the screen and the pictures are clearer, people often say they don't need their big screen anymore. Perhaps phones will come in due course with projectors as well.

The other thing is that the new generation will just automatically be comfortable in using these devices for these extra capabilities. They won't think of it as squashing down what they are used to; they will just grow up learning how to use them, and take it for granted.

Editorial standards