About the Workshop

The Workshop on Mobile Ajax, co-sponsored by the World Wide Web Consortium (W3C) and OpenAjax Alliance, was held on 28 September 2007 in Mountain View, California. The chairs were Daniel Appelquist of Vodafone, co-chair of the W3C Mobile Web Initiative (MWI) Best Practices working group, and Jon Ferraiolo of IBM, representing OpenAjax Alliance.

The workshop was designed to provide a forum for participants to exchange information about the present state of Mobile Ajax, share visions for its possible future, and identify opportunities for industry collaboration in order to promote end-user and industry success with Mobile Ajax. Representatives from thirty-six separate organizations attended the workshop.

How the Workshop was Organized

The workshop was announced to the general public on the W3C site and invitations were sent to all members of the W3C and the OpenAjax Alliance.

In order to attend the Workshop, participants were required to submit a position paper (see complete list of position papers) no longer than 5 pages that expressed their perspectives on Mobile Ajax and identified the topics that they wanted to discuss at the workshop. The co-chairs and the Program Committee reviewed the position papers, extracted a list of common areas of interest, and selected a subset of papers to be presented at the workshop in the form of lightning presentations.

The workshop itself was divided into topic-focused sessions, where each session lasted between 45 and 90 minutes. The sessions were:

The day ended with wrapup session in which the attendees expressed the most important takeaways from the workshop and discussed what follow-on actions needed to be taken. That was followed by a workshop summary from Dan Appelquist.

Minutes were taken over IRC and published on the W3C website. The chairs captured key discussion items in a scratchpad document.

Sessions

The sections below summarize the workshop's sessions:

Welcome and Introduction

After welcoming comments and participant introductions, Dan characterized Vodafone's opinion that within 5 years mobile access to the Web will be dominant. He characterized key differences with mobile, where there are restricted resources (e.g., small screen) and additional capabilities (e.g., camera). Dan showed demos of mobile applications already available today running on a mobile handset. One interactive example -- related to German football -- used SVG. Dan characterized a "One Web" vision where content at a given URL can adapt to the needs of any any device (a user-centric definition of "One Web" - the key use case being characterized by content from any given URL being "thematically consistent" a-la the Mobile Web Best Practices ). Jon described a "One Web" vision for long-tail applications where one version of an application runs anywhere (a developer-centric definition of "One Web"). When asked how these two visions relate, the answer was that they largely overlap. Mike Smith summarized current W3C activities around related to Ajax and mobile browsing, including work being conducted in the W3C MWI Best Practices working group, as well as in the W3C HTML, Web API, Web App Formats (WAF), Ubiquitous Web Applications (UWA), and SVG working groups.

"One Web"

Multiple position papers expressed a vision for the future where the Desktop Web and Mobile Web are one in the same. As this is a fundamental issue, it was the first topic of discussion. The session began with lightning presentations from AOL, Google, Paving Ways and NTT DoCoMo (see complete list of position papers). The identified discussion topics included:
AOL showcased their successes around improved user interfaces for mobile search on higher-end mobile phones. NTT Docomo described the primary characteristics of the Japanese market and how that differs from US and Europe and highlighted special considerations of mobile devices. Google expressed strong interest in the Web as the platform for Mobile Applications, saying other technologies splinter the application-development landscape and require specialist skills. Mobile Ajax is either there or coming, but the critical needs are pixel level control (i.e., SVG or canvas), access to device capabilities, and security. Paving Ways described their work on a lightweight JavaScript library for cross-platform mobile development. Jon described a model for the future where the long tail of the existing Web moves to Mobile Ajax in two waves. The first wave is just getting current desktop content working at all, even if suboptimal user interface, and the burden here is on device manufacturers to include a full-feature browser on the device. The second wave would have many content developers adjusting their Web site for better user interface on mobile once they realize how large of a percentage of users access their site from (small screen) mobile devices.

Access to Device Capabilities

One common theme within the position papers was that Mobile Ajax requires the availability of scripting APIs (e.g., using ECMAScript/JavaScript) to device services, such as GPS (location), PIM, messaging, and camera (to name a few). Sun and Aplix gave lightning presentations and SAP's position paper was presented in summary form as an example of the application developer world having a requirement for device services. The identified discussion topics included:
The need for access to device capabilities was identified as a critical requirement by a large number of participants. One possibility is to offer a simple bridge from JavaScript to JSME, which already has access to device capabilities. JSR 226, 287 and 290 provide some of this bridging. Aplix has other technology (called JSX at one point) which provides a different bridging approach. Some participants discounted this option as it requires that the browser engine and JVM execute at the same time. Jon Ferraiolo reported on recent discussion at OpenAjax Alliance about pursuing incremental microformats for access to device capabilities, with the payloads passed over the OpenAjax Hub via messaging. Concerns were expressed about clumsiness of an asynchronous API approach and how a markup approach might be better.

Microsoft has its own APIs for access to device capabilities, as does Opera. Vodafone offers VodaScript.

Security is a key issue. There was discussion about how Java offers a security manager, but others spoke up on the topic of artful balancing of user interface versus robust security mechanisms.

Widgets, offline, and device-resident

The widget phenomenon represents another use case for Mobile Ajax, where Ajax mini-application are launched from the mobile "desktop" rather than from within the mobile browser. This session had lightning presentations from Nokia, Google, ACCESS and Sprint. Discussion topics included:
There was discussion about terminology and how the term "widget" means multiple things to multiple people. Many attendees expressed the opinion that installable widgets are an important recent phenomenon with considerable traction. Multiple mobile platform vendors indicated that they already support installable mobile widgets or plan to do so in future products. Mobile widgets represent a second major workflow (in addition to the browser) for Mobile Ajax technologies. Some attendees recommended that widget packaging technologies should be developed independent of widget rendering technologies, although other attendees pointed out an industry-wide efficiency benefit if the industry would gravitate towards a single packaging and rendering stack for both browsing and widgets via Ajax, along with packaging for mashup components, installable desktop widgets, and installable mobile widgets.

Role of server-side adaptation

Presenters included ICEsoft, Novarra, MobileAware and Volantis. ICEsoft showed two demos of Mobile Ajax in mission-critical applications, including one application in deployment today where a Canadian shipping company deploys an Opera-equipped smart phone to its truck drivers for dispatching delivery assignments to particular trucks. Novarra, MobileAware and Volantis presented slide decks that explained the advantages of server-side adaptation and how it works. Discussion topics included:
During the discussion, the presenters summarized various W3C activities in support of server-side adaptation. One member of the audience suggested that mobile fragmentation will ensure a long life for server-side adaptation products.

What new standards efforts do we need?

Some of the position papers called for new standards efforts across a spectrum of topics. Implicit in these proposals is the general notion that if the industry can agree on key standards, then the industry would be unleashed to pursue major innovations due to cost efficiencies achieved due to unification of the delivery platform. However, standards take years to develop and the marketplace is moving quickly. There were lightning presentations from Dojo, Wake3, Yahoo and Ikivo. Discussion topics included:
Alex Russell (Dojo) said he supports any effort that gives better access to device capabilities. Doug Crockford (Yahoo!) said the biggest problems are with mashups (which he characterized as a groundbreaking change within the industry) and security. Doug expressed hope that mobile devices are replaced faster than desktop devices and this might allow faster incorporation of key new features and standards. Andrew Sledd (Ikivo) said fragmentation is a big problem now and will get worse before it gets better and suggested that we need to refactor and refine problem statements. Dan Zucker (Wake3) says we need mobile subsetting (it's a question of device physics) and proposed a range of well-defined profiles from "Tiny" to full versions of the Mobile Ajax platform.

The attendees listed many candidates for standards activities: local caching, smart caching, APIs to device capabilities, key mapping, security, mashups, microformats for PIM information, server push, DOM extensions for mobile, CSS extensions for mobile, and best practices.

Bennett Marks (Nokia) announced that OMA browser 2.4 include XHR and will be the last browser standard they will develop. OMA has concluded that full browsers will be sufficiently ubiquitous soon enough, so it does not make sense to do further updates of the OMA mobile browser subset standards.

There was general agreement that standards activities are not good places for invention and sometimes better when attempting standards to achieve market consolidation.

Education and evangelism

Technical standards are one tool to unify the industry around a common vision. Other tools are education and evangelism, which sometimes are as important as standards. Both W3C and OpenAjax Alliance have activities in the areas of education and evangelism, so this session was led by the workshop chairs, Dan Appelquist and Jon Ferraiolo. Discussion topics included:
The attendees discussed what could be learned from various community-driven initiatives that helped move the desktop HTML & Ajax industry forward, such as CSS Zen Garden and Web Standards Project acid tests. The attendees characterized CSS Zen Garden (and AJAX techniques themselves) as helping to show the way through a fragmented desktop browser world. We should try to energize similar community activities that show how to navigate a successful path through the Mobile Ajax world. The acid tests helped identify shortcomings of existing desktop browsers and therefore put community pressure on the browser vendors towards increased interoperability for key features. The attendees discussed the possibility of a conformance brand for mobile browsers.

The attendees brainstormed about what sorts of similar activities the community should pursue for Mobile Ajax. It was pointed out that the Zen Garden allowed for community-contributed style sheet examples. One proposal was to kickstart a similar community effort by developing 10 to 20 demos but allow the community to add additional demos. There was discussion about "test fests". Jon mentioned OpenAjax Alliance recent InteropFests, which aligns somewhat with both the community-driven demo idea and the test fest idea, so either approach might be feasible.

We identified a need to evangelize good support of Mobile Ajax to Ajax toolkit developers, which is likely a good task for OpenAjax since many toolkit vendors participate there. For improved IDE support of Mobile Ajax, we could evangelize the requirement of a Firebug-like feature across all browsers, desktop and mobile.

Wrapup

During the wrapup session, the chairs summarized what they felt were the key takeaways and appropriate next steps. Then, the microphone circled the audience and each attendee identified what was the most important takeaway for him. The workshop (in the form of the position papers, the discussion, and wrapup comments) reflected a generally shared vision among the participants:

Summary

Dan ended the day with this summary:

Next steps