QCon Notes Part 2: The Future Application Platform

Unsurprisingly, there were several talks about the web and JavaScript at QCon. There is no doubt about the meteoric rise of JavaScript in the recent years, and it's hard to imagine that this will not continue. Web browsers have been powered by JavaScript for years and more and more desktop applications are moving to the web. Node.js proved that JavaScript is as good as any other language for building a framework for web applications. Mobile web seems to be the next frontier and where progress seems to be among the fastest.

JavaScript Today and Tomorrow: Evolving the Ambient Language of the Ambient Computing Era

Allen Wirfs-Brock (Mozilla)

Download Slides

Allen is a Mozilla Research Fellow and was the project editor for the ECMAScript 5/5.1 standard.

Allen started off his talk by illustrating the two major eras in computing, the corporate computing era and the personal computing era. A major shift in computing happened in the late 70s and early 80s, where the shift to personal computers radically changed the nature of computing. Currently, we are undergoing another significant shift to what could be called the "ambient computing" era. Ambient computing has the characteristics of being device based rather than computers and being ubiquitous.

Every computing era had a dominant application platform. The dominant platform emerged as the winner through a combination of market demand, good enough technical foundation and superior business execution. The dominant platform for the corporate computing era was IBM mainframes. In the personal computing era, the dominant platform was the combination of Microsoft Windows and Intel PC (much lovingly called Wintel). In the emerging ambient computing era, it is becoming clear that the new application platform will be the web.

Each computing era also had a canonical programming language - COBOL/Fortran for mainframes and C/C++ for personal computing. The canonical language for the web and thus the ambient computing era appears to be JavaScript. Allen brought up the interesting question about what could replace JavaScript and how that could happen. JavaScript, even with it's quirks is "good enough" and there doesn't seem to be any apparent way that it would be replaced by anything else. As such, his claim that "JavaScript will be the next canonical language for the next 20 years" seems spot on.

After the ECMAScript 4 fiasco, TC-39, the committee responsible for deciding the future of JavaScript, is moving a lot faster and is more driven and organized to improve the language. There are a lot of improvements to the JavaScript language coming with ECMAScript Harmony, which represents ECMAScript post version 5. Some might be considered controversial, such as the inclusion of classes, and are ongoing current discussion. Considering the slow browser adoption rate, even ES5 is not yet mainstream and will not be for a couple of years more. This unfortunately seems to be one of the biggest bottlenecks in moving the new ambient computing platform forward.

The Future of the Mobile Web Platform

Tobie Langel (Facebook)

Tobie is currently the chair of the Core Mobile Web Platform Community Group which is dedicated to accelerating the adoption of the Mobile Web as a platform for developing mobile applications. Tobie and his team at Facebook put a lot of effort into analysing the most popular native applications and finding out what capabilities were missing in web applications to make them on par with native applications in terms of user experience.

Facebook recently launched ringmark, a test suite aimed to accelerate the adoption of HTML5 across mobile devices and provide a common bar for implementations of the mobile web standards. Ringmark provides a series of concentric rings, where each ring is a suite of tests for testing mobile web app capabilities. There are currently three rings, however the intention is to continue the project by adding more rings as the capabilities of mobile devices increase.

Ring 0 is designed as the intersection of the current state of iOS and Android and 30% of the top 100 native mobile applications can be implemented using ring 0 capabilities.

Ring 1 includes features such as image capture, indexDB and AppCache. Browsers implementing ring 1 should be able to cater to 90% of the most popular native applications, most of which actually don't or need utilize advanced device capabilities such as 3D. Tobie highlighted that getting ubiquitous ring 1 support should be the short term goal for mobile browser vendors and developers to drive mobile web adoption.

Ring 2 will fill the gap with the final 10% of applications, with things like WebGL, Web Intents and permissions. Ring 2 is aimed to be a longer term goal.

Mobile Web should also be able to achieve beyond 100% of the native apps, with capabilities such as hyperlocal applications (e.g. an application tailored to a certain local event) and deep linking.

Lack of standards for mobile web applications when it comes discoverability or manifest files was also mentioned as one of the hurdles that mobile web needs to overcome. It will be exciting to see how fast we will be able to reach there.

The Future Is Integrated: A Coherent Vision For Web API Evolution

Alex Russell (Google)

Slides (Built with HTML5!)

Alex is a TC-39 representative for Google and is also a member of the Chrome team. One of Alex's missions has been to drive the web platform forward. He is as frustrated as the rest of us developers with the current state of fragmented support and slow progress.

WebIDL and JavaScript have a cognitive dissonance problem. DOM was specified as an API for the browser programmers rather than the actual consumers of the API who are the JavaScript/web developers. It was also devised at a time where there were expectations that other languages than JavaScript would be consuming it, and artifacts of such an ideal still persist in the API. Moreover, DOM does not conform to normal JavaScript rules. The DOM types cannot be extended or constructured. It is not possible to do a new HTMLElement() whereas it would be very useful for many scenarios.

As web applications have increased in complexity, the disconnect between application data and the browser model has grown making web development painful. The developers have been trying to solve this using frameworks such as Backbone.js, however they are not perfect. Alex outlined two proposals to W3C that seek to make web development easier.

Shadow DOM is a way to create web components by a browser provided API. Modern browsers include native controls, such as the standard HTML form components. These built in controls are isolated from the rest of the page and are only accessible through whatever API they expose. There is currently no API to create third party components with the same strong encapsulation enjoyed by the native components.

The other proposal is Model-driven Views which reminded me a lot of how Knockout.js works. MDV provides a way to build data driven, dynamic web pages through data binding and templating via native browser support.


Also interesting, but didn't get the chance to attend:

Mobile, HTML5 and the cross-platform promise

Maximiliano Firtman

Download Slides

Wrap Up

The various efforts around HTML 5, JavaScript and the mobile web all point to an improved developer experience. The question is how soon will this future will arrive? Combined with browser vendors pushing updates aggressively and consumers changing mobile phones every 1-2 years, it might not be as far as it seems. Listening to the talks also confirmed my opinion that native mobile apps are only a stopgap solution and the future lies in HTML 5+ and JavaScript as the platform that will power applications in the future.