Dennis Bournique of WapReview wrote a nice piece on Mobile Web applications, where he ponders on the topic after attending both Mobile Web Megatrends and the CTIA Mobile Jam Session.

I agree with most of Dennis’ points. But in the article Dennis mentions two points, from Michael Mace presentation, that are pointed out as necessary for mobile web applications to take off: 1) offline functionality, and 2) access to core phone features like the phone book, calendar, camera and location.

But it is all relative…

There are classes of applications. Many applications are just fine as mobile web always-connected/no need for offline functionality and/or other handset capabilities.

Don’t take me wrong, the above two points are very important and defines the next step on mobile applications (which I call Lightweight mobile apps).

It is clear that less platforms and less fragmentation is a good thing. And that the promise of mobile web as a consistent and centralized platform is great. But for the time being and for a while (OK long time), fragmentation is here to stay: web, local, widgets. So work with the tools available to you today.

If your application is sufficient as a web app, go web. Need more functionality or richness *today*? Go (or add) local. Is the application pure query-response? Go SMS. Is there an exciting web runtime that you care about? Make a widget out of it. But there will always be web vs local apps, and it all depends on what the application needs to deliver, what your customer wants, the use cases, the target users, security, and so on… while trying to minimize development costs.

A point I want to make is the Widget vs. non-widget applications. A widget is really a design pattern. Regardless of authoring (i.e. local, web, etc) a widget by definition is supposed to be a very focused, typically single-tasked application-behavior-and-User Interface; examples are clocks, alarms, weather, search that are easily accessible and visible. Because of today’s approaches to web runtimes (i.e. differences in authoring and packaging, APIs, security and so on) we have many widget runtimes and widgets are downloadable, similar to local applications. It is a phase. But once all this is overcome, there won’t be a difference between a widget and a web application, except for the design pattern mentioned above. Efforts like Google Gears or BONDI will help with this.

My main concern with mobile web applications and widgets and the new web runtime models that will provide access to local functionality is the security and policy models, which I am afraid will be a repeat of local applications models (and related nightmares).

At the MobileWidget Camp Austin the folks from BONDI presented their initiative, and I will say that it looks promising, assuming the network-provider-centric/natural bias of the OMTP won’t become a roadblock; but the guys seem very open-minded, and I didn’t get any indication that will be the case…

ceo