WAI / GL Session - Sunday, March 22, 1998 Atlanta Room - LAX Marriott Hotel Notes collected by Chuck Letourneau (Note - Issues came up and were recorded under what I heard or decided were the correct topic headings. This collection of comments is not necessarily chronological (i.e. each section was not necessarily a coherent discussion following from the previous). However within each section the order in which comments were made was not rearranged. ) cpl. Collecting outstanding issues. BROWSER SNIFFING: Phill Jenkins - different browsers can present a different page from server. Http allows it. Browser sniffing. Site managers need to be informed to clean up their act. Gregory - is an issue if you are shunted off to a less well kept site. Also issue of frame/noframe telling you to get into the browser. Al Gilman asked it to be recorded Re: Browser sniffing. He had already put up a discussion. and will resend the URL to the GL list. Daniel: Defeats caching ability of Web. This is really a server issue. Should it be for Jutta? It should be for User Agent (Jon Gunderson) until there is a server. DEFINITION OF ACCESSIBILITY: Dave: Trouble defining accessibility as it relates to the Web. Current definitions cause undue hardship for the developers. We need to identify the least common denominator, e.g. Netscape 2.0 and try to find some kind of middle ground. Gregg V: are you suggesting we only recommend using features supported in older browsers. Chuck Opperman: hard to define what is accessible - instead we must support best practices, rather than set a baseline: say if you follow these guidelines rigorously, you will make an accessible page more often than not. Phill: separate documents for HTML 2.0, 3.2 and 4.0. Judy - a separate set for CSS2, CSS1, SMIL, XML XSL? Not a realistic idea. Gregory - agrees with Judy. Hard enough to get people to use one document. Gregg - an author is never in an HTML 2.0 environment. Len - re intranet there may be a situation where some employees have disabilities. Make sure all employees have latest and best, so have simplified guidelines that only take into account the latest stuff (i.e separate out legacy stuff). Judy - note, but don't separate. Len volunteers to walk though guidelines and separate out stuff for most current setups. Chuck: guidelines code to browsers differences. Cindy: isn't the use of using interim doing the level of back-level legacy support. Jap - looking at guidelines, we have the methodology. Bill - no matter how arrived at, the end result is an HTML document - and we must continue to focus on only those things that only the author can fix. SITE DEVELOPMENT TOOLS: (as opposed to page authoring tools) Cindy. WAI has an authoring tools group. We will capture this idea for Jutta's WG meeting in Brisbane. Gregg - usability is the key to accessibility for PwD. Instead of just saying NEW, we have a baselines set of assumptions about browsers, beyond which things refer to NEW (NEW means the construct is in HTML4 or CSS2 but not implemented widely). Gregory - what is the future of the term "interim"? Gregg - have we deprecated some things too early? Daniel: there are too many style areas that are "required" that should be recommended. Daniel will lead the discussions on this. Al - we need to increase or knowledge of the demographics of Web and tools users. Guido - interim category is vast, and some have marginal value. Why not reintroduce an obsolete (or deprecated). Bill Loughborough will look at interim things. Kevin - demographics are expensive to do properly. Madeline: many sites collect statistics. Kitch - Georgia Tech does a twice-yearly survey and asks some disability questions. Bill: is there anyway to get adaptive configurations. FORMS: Jap: why not add guidelines on how to write a form to be useable. Gregory Rosmanita, William Loughborough will work with Jap. Gregg - what we don't want is to fit an entire book between items 3.2 and 3.3. People really want examples of good forms for the reference area. We should grab and put forms. Judy - This should be in the EO group. SCRIPTING: Jap: there are no text based browsers that support script languages. (e.g. entry validation by script) We should recommend to the browser group to include script awareness in text browsers. Daniel indicates that you can do everything in HTML, but scripting is often more efficient. RELEVANCE OF HTML GUIDELINES: Chuck O - in a world where third party applications are increasingly making sites why are we still promoting author guidelines. Gregg - these guidelines are being used as the feeders to UA and UI guidelines. Jap - change the focus or make mention that the focus of the page author guide is more than just to reach individual authors. WHAT TO DO UNTIL BROWSERS CATCH UP? (Current browsers don't support 4.0 or CSS1 / legacy browsers (graphical or text) don't support anything) (What about server-based browsers - owners won't upgrade) SIMPLIFY THE GUIDELINES: Bill Loughborough: why not separate out the things that only the author can do, and forget (in this group) the things that browsers should do. Gregg V. The Interim Tag is meant to tag things that are not always meant to be an author responsibility - things that will go away. Bill: just get rid of "Welcome to [IMAGE]" - everything else can be conceivably done by proxy or user agents. TOOLS TO GRAB HTML FILES: Harvey tried to download guides and associated files and encountered 8.3 restrictions. Gregg: we can't do anything about that, so you can get shareware tools to fix. If anyone knows of useful tools, please send info to chisholm@trace.edu. Al - asked Len if RC should be responsible for collecting that type of information. A universal accessibility problem. W3C is the place to look at it. LYNX DEVELOPMENT: Gregg would like to see if the W3C/WAI can pump money into the LYNX development to encourage some further work. Is there a lynx registry to list all the versions? Is a person really linked to a particular version of LYNK if you can telnet to any site that has the latest LYNX. Gregory: has seen that there are decreasingly available public sites, and even there are increasing restrictions. User has little control, and webmasters care little about shell access. However , we do need an ongoing dialog with the developers and distribution points. The LYNX development consortium is doing a good job, but the W3C might be good to endorse and recommend a certain level and get the word out. George: it is an extra onus on people with little enough skills and resources anyway. Could Gregory lead a group to look at what WAI could do to facilitate the technology level of the latest version, the availability of LYNX on public sites and the adoption of the latest version of LYNX by service providers? Maybe encourage some industry to create a network of accessible LYNX points. Daniel: the proxy server running LYNX - wants opinion of users using an advanced proxy through their existing LYNX server.. Chuck O: not all ISPs run Unix or Linux so don't be too hard on them. Perhaps this is an area for EO. Also explore desktop implementation of LYNX versions. Bill - explore use of LYNX as a text bed, like Amaya. ALT TEXT: Propose a CSS attribute to display alt-text or not. What is role of ALT-text? Role changes between a link and non-link application. There are groups who feel that purely decorative images need no description, and others who think all should be described. Then TITLE can add further information to link text, then LONGDESC can add further information to descriptive. Then machines may automatically generate some alt-text. The machine could put into alt="chart: description goes here" Graphics used to position things will be handled differently. Al: To come to a resolution on ALT we will have to marry authors to browsers to Kevin - it is up to the author whether an image is decorative or integral to the meaning of a site. Len - there are more or less important things about any description - use a p top-down progressive. William - there is no such thing as a decorative image - everything is important. And there is no machine way to extract important information. Gregg - No, auto-naming is a real issue now. Steve - give some kind of concrete feedback to companies who are asking and suggests Len's progressive information string is the most concrete guideline. David - add another tag to HTML - first time we really get into the content of the page. Need rational explaining why full description of creative content is needed and make case to the authors. Guido - won't be able to infer what a graphics is : why not add a class attribute list like decorative, background, wallpaper, logo, etc from which authors could choose and let the end user. Kynn - agrees with Guido - use CLASS and Styles to let author specify what he wants rather than read the mind. Tom - thinks we are putting too much faith in authors to do Chuck O: thinks it is bad idea to add more attributes when people aren't using the ones that exist now. Thinks that it is too complex with ALT TITLE and LONGDESC. IE uses tool-tip on ALT to raise the awareness of users of the availability of ALT. Display title in all cases. Steve: CLASS - amalgamate Guido's with Len's Geoff Freed - agrees with Chuck O - and get rid of LONGDESC. Presentation by Sarah Morley on Dr. Helen Petrie and researcher Chetz Collwell's project to evaluate the guidelines and by students for blind persons designing and using the results. Using a range of browsers, with a range of screen readers and Braille readers with range of user experiences. --------------------------------------