Mark Wilcox's Forum Nokia Blog

The Future Symbian UI - the most important thing not discussed in Monaco

2009-05-05T13:03:19+03:00

For those that haven't seen it, David Wood has posted about some proposals that Nokia have made to the Symbian Foundation here:

http://blog.symbian.org/2009/04/30/reviewing-the-release-plan/

These include:

1) Qt for S60 to become an official part of the Symbian platform

2) Orbit - a new Qt-based mobile widget set (this is really implementation detail)

3) Direct UI - a replacement for Avkon and the existing applications

Yes, you read that right, Qt is proposed to become the new application framework for Symbian devices (from Symbian^4 onwards).

Personally I think this is a fantastic thing, and will be all the more exciting because developers with access to a hardware reference platform (hopefully more phone-like versions available by the time the code comes along) will be able to watch the progress of the development and even get involved. However, it does imply (at least in the current proposal) a significant source and binary compatibility break. Is this also a good thing? I think so, but I'm sure not everyone agrees.

If you want to know more get yourself beta access to the new Symbian developer website and look at the proposals yourself. Then discuss the implications on the new Feature and Roadmap Council discussion board if you have some useful input to the decision making process.

If you want more background info about Qt on Nokia platforms and the general strategy, see the excellent presentation made by Mika Rytkonen on this subject at the Nokia Developer Summit. In the same session, the trolls also talked about the future of Qt, so if you're interested in this future UI based on Qt then look out for their declarative UI updates (extension of Kinetic work) over the next few weeks/months.

Please note - these are proposals for Symbian^4 which should ship in devices end-2010 or early 2011. Avkon isn't going to disappear overnight. :-)

Qt on S60 Open Source Project - Bluemapia

2008-12-17T19:59:14+02:00

Do you like to work on the cutting edge of mobile technology? Are you interested in social networks or location based services using free maps? Want to get involved in an open source project on mobile phones?

If the answer to any of those questions is "yes" then maybe this post is for you.

Learn about Qt on S60, porting to Symbian OS and open source mapping solutions. Get hands-on experience with the latest Qt on S60 technology preview and help port a social mapping application while you're learning (I find you always learn more and faster when you have a real project rather than just playing with examples).

Qt on S60 is fantastic news for S60 developers, even more so since the revelation in the webinar (hopefully it'll be at that link soon anyway) today that Nokia intend to make Qt free for commercial S60 development. However, we're going to need some good examples of Qt on S60 projects on a variety of scales. I'm starting work on one of those now, incorporating the other cutting edge trend for mobile social location (N97 anyone?). Which will mean we also have an example of extending Qt with camera and GPS functionality.

Bluemapia is a location based social network for boaters. There is an existing open source client for Windows Mobile, that uses various free maps and other popular open source libraries. We're planning to port the application (and libraries) to Symbian OS/S60 and re-write the user interface using Qt. I say "we" because some engineers from Symbian (including Symbian Ambassador Lucian Piros, who you may know from the SDN discussion boards or NewLC), an experienced open source GIS developer, an experienced Qt developer and more, will be contributing to the project. If you want to learn from some experts then volunteer too and get involved!

The idea is that we combine with existing open source mobile efforts in this area and create some components that people can use to build their own (open source) location based apps that use free maps (and maybe also not-so-free maps if/when we've got some appropriate APIs and license terms).

The project will be hosted in Launchpad and to get things going quickly I'm planning a "code camp" for some of the developers in London on 17th January. If you'd like to come along to that and get involved, or can't come but would like to get involved, or just want to know more then please get in touch - either reply to this post (with a way of contacting you) or email me via the discussion boards (user name Sorcery-ltd), or just go to the code camp event site and register (if you don't have a Bantora account yet you'll have to email me first to get an invite) - places are limited, first come, first served - final numbers must be confirmed by 6th January.

Looking forward to hearing from you and working together!

Mark

P.S. Once we're done porting to S60, hopefully we can also port to maemo and other Qt supported platforms too.

Multi-language Programming, Python, Flash Lite, C++, Java, Widgets

2008-12-16T17:03:33+02:00

I've been working with some other Forum Nokia Champions and expert developers in the last few months to create some resources for Multi-language programming (mixing your favourite runtimes). The result is a series of papers for the Symbian Developer Network.

In them you can find out about:

1) The architecture for applications mixing languages

2) The strengths and weaknesses of various combinations of languages

3) How to combine Flash Lite with C++ or Java (with examples from KuneriLite and Jarpa)

4) How to combine Python with Flash Lite or C++ (with examples from Flyer and a tutorial on writing PyS60 extensions)

5) How to extend WRT widgets

One of the suggestions I made on my blog here before and also in the papers has already been done. One of the authors of the paper covering Python, Jussi Toivola (aka GameDude) has ported pygame to PyS60 Community Edition. Now you can write games with sound effects in Python, taking advantage of the native performance provided by the SDL.

Please read and enjoy! Many thanks are due to Bernd Wiegmann, Felipe Andrade, Ugur Kaner, Jukka Hamalainen, Mikko Ohtamaa & Jussi Toivola, as well as the wonderful folks at the SDN.

Mark

Symbian getting open source friendly - almost there

2008-12-13T12:38:17+02:00

We've had some really great developments for open source on Symbian OS in the last couple of months, most of them have already been reported but I thought I'd do a quick summary and look at what's still needed.

We had Open C/C++ to let us write standard C/C++ code on Symbian, now we've got Qt on the way too. Thanks to another poster on the discussion boards I can now point to a public source that says we have GStreamer being ported to Symbian as well: http://www.scribd.com/doc/40648/S60-Extending-the-Internet-Ilari-Nurmi-Nokia (page 9). As Paul reported on Thursday, Mobile Firefox is also coming to Symbian OS, and that should mean more libraries (such as Cairo) coming with it. The more building blocks and frameworks we get in place, the easier it is to port other applications and libraries.

Very rapidly we're getting to the situation where we have a Linux-like user space environment on top of the Symbian OS kernel. To me this seems like the ideal combination for open source in mobile (I think it's going to take a few years for embedded Linux efforts to make it a serious contender in terms of power consumption and support for phone hardware, particularly at the lower end - single chip devices - and top end - the way SMP is supported needs to be quite different than for the desktop).

Now that Nokia's acquisition of Symbian has been approved, the way is clear for the Symbian Foundation and opening of the Symbian/S60 code. As long as most of the big players in the mobile industry can be the "good open source citizens" that Nokia is trying to be then eventually Symbian & Linux should just be different open OS's that you build all open source mobile projects for. There are still some things that need sorting out before we get there though:

1) Tools - Carbide.c++ v2.0 becoming free is a fantastic piece of news and a big step in the right direction. However, it's still tied to Windows and the majority of open source developers (and indeed good embedded software engineers in my experience) would rather not use Microsoft OS's if they don't have to (putting it mildly). Being Eclipse based, Carbide.c++ could theoretically move to Linux, although we'll need the (strongly rumoured but not much publicly discussed) simulator to replace the Windows emulator and some fixes to the SDK. I'm not sure how Windows-specific the on-device debugging tools are either but I expect some work is required there. This same point was made by Joel West recently.

2) Symbian Signed - I'm expecting further change here at some point since no other platform has this kind of clumsy headache of a system; it just can't be allowed to continue for too long. It would be nice to hear some official comment about reforms, even if they're going to take a while to implement. The current system is decidedly unfriendly to open source and there's really a need to stick to projects that don't need anything other than user-grantable capabilities.

However, I'm really pleased at the pace of progress with the mobile software world opening up. So much so that I think it's worth working on some big open source projects on Symbian now (and clearly Mozilla agree!).... watch this space for my first one....

Mark

Thoughts on Smartphone Show 2008

2008-10-27T11:02:30+02:00

I was at the Smartphone Show last week. It was much larger than previous years and there was more of a buzz about the place than before. I'm sure this had something to do with the soon-to-be-created Symbian Foundation.

One of the big news items from the show was the nomination (can't actually be appointed yet because the foundation won't exist until Nokia's acquisition of Symbian goes through) of Lee Williams, currently head of S60, for the role of Executive Director of the Symbian Foundation. I have mixed feelings about this. On the one hand he is almost uniquely qualified having been the first head of S60 to have other licensees than Nokia get products out in a sensible timescale. He knows the whole platform that will form the core of the foundation releases. He's a very capable and enthusiastic communicator. On the other hand, I think politically it would have been much better to appoint someone from outside Nokia, possibly even from outside the industry. Preferably someone with some open source credentials. The message from the keynotes at the show was that it's all about developers. I don't think this appointment really confirms that. Then again, I'm sure Lee will do a great job and I wish him the best of luck! He's just going to have to work extra hard to demonstrate the independence of the foundation when Nokia is doing most of the early development work on the platform.

There was a lot of exciting news for developers at the show, with plenty of new stuff to play with now, or look forward to shortly. At the Forum Nokia stand there were lots of demos, including Carbide.c++ 2.0 and Qt on S60. It was great to see the new architectures in Symbian OS (Screenplay & Freeway), that were not much more than powerpoint slides at last year's event, with impressive demos on the Symbian stand. However, the most exciting things (yes I get excited about some pretty geeky things!) on the Symbian stand were the new Symbian Analysis Workbench (SAW) and version 2 of the build system, that runs on Linux as well as Windows. The SAW tools are for debugging and performance analysis, they're available as plug-ins for Carbide.c++. They look similar to, but better than, the current Carbide performance investigator tool (part of the expensive Pro and OEM verisions). Currently they're only available to Symbian partners on the SDN++ site but the suggestion was that once the foundation is established they'll be free to everyone (hopefully with the other critical Caribde.c++ feature - on device debugging).

Of course, for me, one of the great things at the show was finally seeing my writing published in the Symbian Multimedia book, which you can now buy! There was also a very nice book on Design Patterns on Symbian OS released at the same time.

However, the very best thing at the show is meeting all the people that make these things happen and other like-minded developers. I got to meet up with several of my fellow Forum Nokia Champions again, which is always a pleasure. I also enjoyed getting to spend some time with Espen from Qt Software, who's leading the port of Qt to S60, Vinod, the technical architect of the P.I.P.S. team at Symbian and Lucian, one of the new Symbian Ambassadors (very much like a Symbian version of Forum Nokia Champion - although the first batch are all Symbian employees so I'm not sure how much use they'll get from the benefits of membership), plus many, many more - too numerous to mention.

It's really worth going to some of these events - you're bound to come away with plenty of new contacts and ideas. Don't miss next year's Smartphone Show, whatever form it takes.

Mark

Python for Series 60 Community Edition!

2008-09-04T13:36:53+03:00

Ever wanted to add modules to the core PyS60 distribution rather than having to find them from all over the place?

Ever wanted to build a single executable for your Python application that didn't need the interpreter installed separately or as an embedded SIS file AND have no clashes with other interpreter versions or problems with uninstallation and dependencies?

Ever wanted to try out a new build system for Symbian OS, where the build scripts are written in Python (SCons for Symbian anyone)?

If you had these things what would you do with them? You could:

Write commercial PyS60 applications!
Port/Intergrate new modules more easily.
Fix bugs in PyS60 core or modules that affect your apps rather than work around them.
Much, much more...

Thankfully the fantastic folks at Red Innovation have brought us Python for Series 60 Community Edition that provides all of this. I can take no credit here, I've only tested that the build system works. At the moment it's still really only for enthusiasts and early adopters as the installation and setup is a bit complex to start working on it, but there are plans to simplify things in the future.

The project is hosted at Launchpad so that's the place to go with your questions and bug reports.

Personally, I'd love to see what people could do if we add Pygame support, since the main dependency, SDL, is already ported to S60. Unfortunately I have no idea when I'll have time to work on this. If anyone else wants to get started I'll be happy to support/mentor them if they need it though - just comment this blog or drop me an email via the discussion boards. I've already added a "blueprint" for this to Launchpad, but there's no detail yet.

Perhaps one day Nokia will be able to take patches and extensions from the community release back to the official release...

Anyway, enjoy this great new resource.

Mark

New Stuff for Symbian C++ Developers

2008-09-02T12:05:01+03:00

I thought I'd write a quick post to highlight some of the things I've been up to lately.

First, there's Multimedia on Symbian OS: Inside the Convergence Device, which I co-authored for Symbian Press. It should be available next month. Lots of great stuff for multimedia developers, or those wanting to add multimedia to their applications. It also does a really good job (well I would say that wouldn't I) of explaining the multimedia architecture, so you can understand how it all works underneath.

Then there's a pair of paper's I've written for the Symbian Developer Network. One explains how to port software from Linux to Symbian OS and the other explains how I ported an open source guitar tuner application (with example code).

I may well be talking about this and other porting work I've been doing at the Mobile Developer Days in Berlin next week. There will be lots of interesting things to learn about there, including plenty of presentations and tutorials from Trolltech about Qt on mobile devices. Forum Nokia are hosting tutorials on Flash Lite and Web Runtime. You can also learn about Python, sensors and much, much more. Don't miss it!

Mark

S60 Under-engineered?

2008-08-07T11:45:16+03:00

Having quoted a Nokia employee in my recent blog post, saying that open sourcing S60 was like spreading manure out on a field, I thought I'd share my latest insight with you. I'm reading "Refactoring to Patterns" by Joshua Kerievsky; so far, it's excellent. While I was reading his description of an under-engineered system, S60 immediately sprang to mind - see if you agree:

"...While systems you've worked on may not be so gruesome, it's likely you've done some under-engineering. I know I have. There's simply an overwhelming urge to get code working quickly, and it's often coupled with powerful forces that impede our ability to improve the design of our existing code. In some cases, we consciously don't improve code because we know (or think we know) it won't have a long shelf life. Other times, we're compelled to not improve our code because well-meaning managers explain that our organization will be more competitive and successful if we "don't fix what ain't broke."

Continuous under-engineering leads to the 'fast, slow, slower' rhythm of software development, which goes something like this:

1. You quickly deliver release 1.0 of a system with junky code.

2. You deliver release 2.0 of the system, and the junky code slows you down.

3. As you attempt to deliver future releases, you go slower and slower as the junky code multiplies, until people lose faith in the system, the programmers, and even the process that got everyone into this position.

4. Somewhere during or after release 4.0, you realize you can't win. You begin exploring the option of a total rewrite."

So, first S60 phone released mid-2003 following fast and successful development project. S60 2.0 (2nd Edition) ships in the 4th quarter of 2004, a bit less than 18 months later. S60 3.0 (3rd Edition) ships at the end of the first quarter in 2006, more than 2 years later following a major struggle to add platform security and move to a new Symbian kernel. Will we get the next major release (5th Edition as 4th is skipped, supposedly because 4 is an unlucky number in some cultures) in our hands before 2009? And during this development Nokia are planning to completely switch the development model to open source and are clearly considering starting again, although they don't have time to write a new framework from scratch so they bought Qt!

Is this a fair representation of what's happened? Comments welcome!

Everyone is making a big fuss of the iPhone but it's at step 1 (the 3G version really doesn't add enough to be considered step 2 in my opinion), while S60 is at step 4. Are Apple going to do better, only time will tell.

Of course Joshua Kerievsky suggests a solution to this under-engineering problem - Test-Driven Development (TDD) and continuous refactoring. There are plenty of teams in Nokia and Symbian who are already using these practices. However, there's another issue that prevents major re-design that is specific to open systems - binary and source compatibility guarantees. The developers of the code can't refactor everything they'd like to because they don't actually know who else is using the interfaces and how; they've just promised not to change it. What's the solution to this? It seems the one we're most likely to get is parallel interfaces. Leave the old ones as they are and add new ones alongside them. Only add new features to the new interfaces so that developers eventually have to migrate anyway. This means we end up with an increasingly bloated code-base, carrying the remnants of old releases around almost forever (a bit like Windows really). Is there a better way? Or are all successful systems doomed to follow this course?

I'd love to know your thoughts on this.

Mark

Symbian Signed - a proposal

2008-05-29T22:20:19+03:00

OK, so it's easy to criticise Symbian Signed but they have been improving things. I still think there's further to go so here's my simple proposal for the next step.

I'll call it Free Signed.

Free Signed is just like Express Signed except that it's free and there're no test criteria.

Here are the compromises I'd suggest:

Like Express Signed, you can't access the most sensitive capabilities with it - there really are good reasons why the developers need to be trusted for those.
You still need a publisher ID (or someone with one) to sign your application - without some kind of chain of identity verification there can be no trust, without trust there is no security model.
All applications that go through Free Signed have to have a warning box on installation that says they aren't Symbian Certified, you are installing the application at your own risk and if you have any problems with your device after installation then you should remove the application and/or contact the supplier before contacting your device vendor or network operator. This could be just a text file in the SIS initially but later enforced by the software installer in new device firmware. For the text file option this could be spot checked and anyone omitting it could have their publisher ID blocked.
(Implied by 3) Free Signed is identifiable by the certificate such that software distributors can set a policy on the signing methods that they will allow.

I think this method could replace a lot of the current usage of Open Signed (although I see no reason to remove that option) and could be very useful for freeware and open source, friendly user and public trials for commercial software and also in future, internal projects in large organisations (for whom mobile will become a growing part of their IT strategy).

To really make this work for freeware and open source though we'd need a network of publisher certifiers. I'd suggest one ideal source of those are people who run popular blogs and websites about mobile applications - being able to supply installable copies of free applications could drive more traffic to their sites. They get sent a copy of the application and test it anyway, if they have no major problems with it they can sign it for wider distribution. There may also be other Forum Nokia Champions who are willing to do this and probably some staff at various companies in the Symbian ecosystem who also have a personal interest in the technology.

The key issues would be zero or extremely limited cost and liability for the publisher certifiers. They could sign something to say they would provide their best effort to help track down the originator of any malware or cracked application that they inadvertently sign.

There - a dull post with no links, pictures or videos! Just an idea.

What do you think?

Mark

Application Quality & Numpty Physics

2008-05-24T16:28:42+03:00

In my last post I talked about a smart new S60 application and its implications for Symbian Signing and platform security. I said I'd talk about the other part of Symbian Signing, application quality, in my next post... so here it is already.

My personal opinion is that what is currently Symbian Signed should have absolutely NOTHING to do with application quality. It should be entirely about identity and trust. In addition to this I do believe there should be a signing program that is ALL about application quality. Here's why, the current system basically assumes that there are developers and there are dumb users and nothing in between. There is a quality gate that can actually prevent you from properly distributing many types of application. In the real world there is a sliding scale of user knowledge from the very advanced power-user to those who've never used a computer or smartphone beforeand have no idea what software really is or what it can do. Where those different groups of users discover content is likely to be very different. Software published in some places is unlikely to be found by anyone but the enthusiast - in this way the audience is somewhat self-selecting. On the other hand, the Download client built into Nokia devices and operator portals should only carry content of an appropriate quality - otherwise there are likely to be serious customer service issues.

For open source and freeware developers, very often it is the advanced users that are in fact the alpha and beta testers. You can't really rely on finding them all before you start testing - you just publish your first vaguely usable version and see who wants to play with it and what feedback they give. That's one of the wonders of the internet - it seems there's almost always someone out there who's interested. Release early and often is one of the major practices in the open source world. The current Symbian Signed program and criteria are incompatible with this practice.

Now, to illustrate my point I want to use the example of my new favourite application on the N800 Internet tablet - Numpty Physics. It's based on the concept for the amazing Crayon Physics, here's a video showing the gameplay for the upcoming commercial version of that, Crayon Physics Deluxe:

Numpty Physics is listed as "The pearl" on Maemo.org as I write this (which basically means it's a lot of other people's favourite too) and it's publicly stated that it's only beta quality. In fact I've had it crash so badly I had to take the battery off and also the only way to quit is via the 'q' key which is only available on an N810 - oops. Do I care about these issues - no. I love that I can play it now - I'll like it even more when it's finished. If I had the time and inclination I could get involved and help improve it. If I had to download an unsigned package and submit it to a signing portal would I have installed it and then got excited enough to think about getting involved in the project? Almost certainly not.

And that is really my point. What open source and free software projects need are motivated users and developers to easily access them and give them a quick try.

Signing an application just to enable it to be installed on any device should be both free and without restriction (except some kind of identity/trust chain). I'd be happy to sign other people's open source projects after fairly limited contact with them (because I can see the code after all - I'm not at all so sure about closed source freeware, I'd want to get to know the developer quite well first) if I didn't then have a legal responsibility for any harm they might do. That responsibility has to rest with the end users that choose to install them.

Application quality should be policed at the point of distribution. That can certainly be via a central signing program, it just shouldn't be the same signing program that gates whether on not an application can be installed on a device at all.

What about security for the end users? Well, I think my last post showed that Symbian Signed doesn't really do much about that anyway.

Any thoughts or suggestions?

Mark