"Live Notes" from the Mobile Ajax Workshop

29 September 2007, Mountain View, California, USA

IRC information:
irc.w3.org
Port 6665 or 80
#mobile-ajax

Zakim bridge: +1 617 761 6200
Conference code is 2529

Ground rules
    No ratholes
    No religion
    No product pitches

Issues around one web:
    From the developer POV?
    From the user's POV?
    Alan Tai: small screens mean "one web browsers" (such as iPhone) still aren't the answer
       (DKA: Will never be the whole answer - content must adapt)
       (Andrew Sledd: designers also need to design service for use on mobile)
        Fragmentation of JavaScript support across existing mobile browsers
    NTT DoCoMo: new programming model needed (unpredictable off-line vs. DoJo )?
    Dave from Google: Ajax short-comings - pixel level control of UI (W3C SVG and canvas can be)
                Offline operations: google gears, html 5
                Access to low-level functions on the device (location, etc...)
                    Security model is an important issue for access to low-level
    PavingWays project - Rocco Georgi: working on a library for constrained browsers Frost
        Current implementations do not support sophisticated javascript
         Library can react when being off-line
        Server component (PHP) and client component -
One web means one set of technology?
Need for a baseline? Testing is di

Widgets

ACCESS:
* Ajax good for minimizing bandwidth
* Most Web 2.0 services today are designed for desktop
- mobile devices: imited memory, lower power CPU, narrow bandwidth, limited input
* Proposal:
- extra-DOM that is optimized for ajax apps
- explicit garbage colection
- alternate virtual key event
* Intend to submit to a standards orgaz

General issue: Code signing: do we need it?

Widget packaging and interoperability of widgets across frameworks

What is a widget?  Why is it different from the web application platform in general.

Widgets are gaining mindshare in developer and user worlds.
Widgets are a way to address fragmentation [but we have widget fragmentation now]
"Weblets" (!!) "Mashlets"  Midgets?
Two problems conflated?: what technologies vs. deployment of apps
Two more problems conflated: why the name widgets? - developing end-user view of a widget which is independent of tech vs. developer view

Can we make widget development easier for app developers?
Is there a difference for the user experience of a desktop widget vs. a mobile widget? (background vs. app)

What new standards?
    Local caching
    Better access to device capabilities - new tags rather than javascript APIs
    Mashups are an important innovation - security model [in browser] is not up to scratch
    Google gears "worker pool" proposal could be a starting place
    Key mapping
    Incremental approach vs. mobile subsetting
    Existing and in-development work like dojo-mobile and frost work NOW
    [OMA Browser 2.4 includes normative reference to XHR work in W3C (and represents convergence)]
    OpenAjax Hub as a way to allow cross-platform access to device capabilities?
    New javascript APIs
    DOM extensions
    CSS extensions
    Standardized security regime - cooperation under mutual suspicion
    Standard microformats - eg contact, calendar, etc... in sync with capabilities of the phone...
    [everything in Javascript runs in a global space]
    Proper module boundaries - gears solves through worker pool threads.
    [issue: gears is being contributed to the whatwg so does that mean it's all-or-nothing HTML5?]
    Smart caching into the browsers as an alternative to gears?
    Push / remote updates?
    Best practices - can we do them?

Problems / Challenges / Issues
    Regulatory / Privacy issues around location and other personal information
    Can you have a voice and a data / web app experience at the same time - shared experience?
    Discovery & Download of services is a challenge
    Standards activities aren't good for inventing and do better with existing practice
       (Market creation vs. Market consolidation)
    Need to identify individual topics that can be isolated and standardized - "component standardization"
    Be careful of linkages

Education and evangelism (e.g., Ajax best practices)


In desktop world in the past:

In mobile, only thing that is vendor-independent might be dev.mobi, but various vendor-specific projects
Need clear benefit and clear goal. Zen and WSP had this. Zen had clear educational goal.
Showed a subset of CSS that worked.
We are bemoaning fragmentation. If someone can show way through fragmentation, that might help.
Ajax is about navigating through desktop browser fragmentation.
MWI BP were a set of best practices, versus Zen Garden which had users contributing sample content
Maybe create a set of ATOM feeds or other services and allow developers to showcase their products
InteropFest? Good success with latest OpenAjax InteropFest. 20 companies, 20 toolkits.

Describe a set of "Ajax OK" badge for mobile browsers?
MobileOK from W3C is more towards content. Does content meet tests for MWI BP. For low-capability devices. Does it encourage LCD approach?
MWI talked about a higher baseline, but that was rejected for fear the world might have moved beyond it

Need device manufacturers to put full Web browser on mobile devices. Nokia leading the way. Running in heterogeneous operating systems. Huge cost to get browser available on all devices.
Now, putting full browser is an economic decision (mainly people resources) not technology reasons. This results in fragmentation because of economic reasons.
Do we want to say what criteria are for a Mobile Ajax device? Or was that OMA's latest specs (with XHR)?
Acid test for either OMA specs or full desktop browser?
OMA work in test-fest area from W3C. In form of a set of individual feature comformance tests. People can contribute and extend. Meant for self-testing, not for evaluating conformance of particular browser.
There is an opening for acid test for OMA browser 2.4, but some organizations need to worry about liability things.
in early days of web, copy/paste helped adopt. Doesn't work with Ajax (says Rotan).
Evangelize to book authors?
W3C Schools involvement?
What about Ajax content that don't have fallback. Degrade gracefully.
OpenAjax could evangelize to toolkit developers
For browser vendors, OpenAjax is starting a Runtime Task Force to make a list of features we want to see in browsers
IDEs - often a blocker. Different class of problems. Developing on a different device than you are delivering content. Usually need the device in your hand. Zillions of different devices.
Visual Studio has very good support for mobile app development, but not mobile web development yet. (Some support, like validate against schema.)
Aptana uses browser debugger like Firebug. Maybe evangelize to browser vendors that they need to support a feature like Firebug. Aptana got debugging on iPhone via local web server approach that no one knows the details.
NPAPI scriptable API, but not quite complete everywhere. Would allow a Firebug-like extension.

Wrapup

Maybe we can pull together info and publish info from the workshop in a digestible form would be useful to the community.
Industry is moving forward, such as the rise of Webkit. This is the future. Very encouraging.
This event should help us craft activities such as what happened with  Mobile Web workshop from 2007

People's various takeaways from today:
OpenAjax mobile tf are of wide interest (affirmation) and guerilla tactics might make good
Good that W3C and OpenAjax are talking and working on the issues
+1
Standards are part of the solution (and part of the problem)
(+1 on problem side)
Opportunity to define a standard platform
Scratching head about next steps
+1
Mashup comments were interesting
Moore's Law is going to be a big factor
APIs stick out, security, offline, but maybe not ready yet for a whole lot of standards, maybe open source better
+5
Useful, but too short to discuss all issues
Should focus on existing standards efforts rather than starting new ones and you should participate in them
+1
Common vision bolds well - use web as application platform for mobile
+1
iPhone more popular than expected
Security important
+2
Needs to be a community around "Web on Mobile"
Glad to see most of crowd is pragmatic, requires lots of coordination between many entities
Attempting to lobby vendors is hard, but finding a way to assemble browser vendors to get involved to make agreements, whether standards or not

Dan's summary
* Interesting, informative and helpful
* Need for community in this space, not necessarily in standards effort. e.g., Maybe some conferences or developer events
* Call for working in existing activities rather than start new things, bearing in mind that many problems will not be solved in standards bodies
* Need to turn this discussion into concrete work items. Going forward, things will be proposed and charters will be written. People need to get involved in the follow-up activities.