On the road to Bandol

Even tough Smart Mobility is very much focused on security, mobile application frameworks are at the heart of the debate. Many speakers, including myself, have more or less compared some features of various mobile application frameworks. Of course, our coverage was quite varied, but the main focus has definitely been on traditional platforms, including Symbian, Java ME, Windows Mobile, iPhone, Android, and a few more.

Naturally, many people (including myself) complained about the inconsistency of these frameworks, which of course leads to fragmentation. The fact that some of these platforms (notably, Java ME) have fragmentation issues, makes the problem even worse.

After hearing several presentations, I feel quite uncomfortable. We all look like we are trying to address a problem that looks untractable: at the level of these frameworks, fragmentation is here to stay. We may be able to fix this temporarily but eventually, we need to move to something else.

My (current) feeling is that the solution lies in a combination of browsers and widget engines. Browsers can be heavily standardized, and widget engines can be a very nice abstraction layer, providing a similar user experience on different devices. On that side, we have heard from OMTP’s BONDI, which is a Web/Widget application framework, to be integrated in mobile browsers. This is significant, because OMTP represents the interests of major mobile operators, who are powerful actors in the mobile world, able in many cases to impose their views. At the same time, Yahoo! has updated their Blueprint widget description language, and they have announced the availability of their widget platforms on many platforms, including Symbian, iPhone, and Java ME. This kind of frameworks may not be the solution for highly interactive FPS games, but they can be quite effective for many Web-based applications, including social applications.

Of course, the main advantage of these high-level Web-based frameworks is that they build an abstraction layer on top of the fragmented native frameworks. There are actually two approaches to do that:

• The first approach, used by BONDI, consists in extending the browser capabilities, in order to make them support an extended framework of widgets and Web applications. This approach also includes a flexible security model, and many other nice features (including a possible access to all features offered by the native OS). However, some people may claim that this approach simply is a new application framework, which adds even more fragmentation to the market.
• The second approach, used by Yahoo! and other widget engines, consists in building a widget engine that is simple enough to be implemented on top of all the other frameworks. The main advantage is here that the widgets written for such a framework, will (supposedly) work on any device, addressing the device fragmentation issues. A major drawback of this approach is that the widget engine is considered as an application by the underlying frameworks, and it of course requires elevated privileges. In terms of security, such widget engines simply are dangerous, as very few people have any information about the security features that they offer.

Both approaches look interesting, and we also need to consider the economic models associated to them. We will see how this evolves in the next months, but it is definitely worth following.