DIA trip report Author: Jacco van Ossenbruggen W3C Workshop on Web Device Independent Authoring (DIA) Tuesday/Wednesday, 3/4 October 2000 HP Labs, Bristol, OK http://www.w3.org/2000/10/DIAWorkshop/ While the workshop participants all had pretty diverse backgrounds, I think you could roughly distinguish: - W3C incrowd (Daniel Dardailler, Dave Raggett etc) - Accessibility (WAI WG, etc) - Mobile phone (Erikson, Nokia, phone.com, etc) - Miscellaneous (including interactive TV, UIML, multimedia, etc) Despite this diversity, there was consensus about the need for a W3C activity on DIA, and to a certain extent there also was consensus about the direction this should go. This included making guidelines about how to use the current specs in a way that optimizes device independence; similar to the WAI guidelines for accessibility. It also included cross-working group activity to coordinate the various standardization efforts, and some empirical study (a la the former web characterization activity) to investigate the current state of the art on DIA: what and how do people currently do it and what else is needed. The main purpose of my talk was to get rid of the assumption that device independence is only an issue for authors that are dealing with text-based content, and that if you author multimedia you give up device-independence by definition. See http://www.cwi.nl/~jrvosse/talks/bristol/ for slides and paper. Many of the other talks illustrated the current attempts by content providers to make device independent content for WAP and interactive TV. This often (always) fails because of: - lack of standardization (a Microsoft guy discussed more than 9 "international standards" for interactive TV) - underspecified standards (a phone.com guy explained how two WML-compliant phones interpreted the same WML-cards completely differently) - hugely diverse devices (a Nokia guy explained the many differences that exist among all the Nokia phones: landscape versus portrait screens, various screen sizes, resolution, color depth, voice interface/full keyboard/phone keyboard, etc) There was also consensus about the fact that there are different levels of device independence, each level might require a different approach. E.g. there is a big difference between making the same content available for two different WAP phones or making the same content available for a WAP-phone and as a fully fledged SMIL presentation. The problems in the first case are perhaps better solved on the client, the second case on the server. Also the problem of offering alternatives using a fine versus coarse grained split up (in SMIL terms: have one switch at the root of your document tree or many switches at the leaf nodes) was discussed, and the consensus here was that it was good to allow both options (as is the case in SMIL), but that in general DIA requires something at a higher level than this "#ifdef" style of authoring. Alternatives such as transcoding existing HTML into other formats is often used, with, as expected, very limited success. More viable seems the use of the same database back-end with different templates for HTML, WML etc. Issues here are the level of device independence of the database content and the control the content author has over the end result. The presentation about UIML (user interface markup language, see www.uiml.org for more info) was a typical "this is cute but does it scale" type of presentation. The idea of UIML is that you define your user interface at an abstract level in XML, and then provide mappings from this model that define the realizations using Java/MFC/Qt widgets, HTML forms, WML, etc. Another recurring trend was the wish to get from a "page-oriented" model to an "application-oriented" model. This came up with the WAI-compliance checkpoints, which assume a page by page assessment to award a zero (not accessible) to "AAAA" (very accessible) status. How can a site with an large/infinite number of database generated pages get a AAAA status? Same for WAP: you want a whole (part of a) site to be available on your phone, not a random set of pages within a site. Same for voice browsers, where the page is replaced by a dialogue metaphor. It also came up when authoring support was discussed: tools need to ensure a complete application has certain characteristics, instead of the current page-by-page metaphor. The application metaphor also implies that device independent user interaction, event handling, etc is at least as important as device independent presentation.