- From: James Shattuck <progman@ecst.csuchico.edu>
- Date: Thu, 17 Apr 1997 18:55:01 -0700
- To: HTML List <www-html@w3.org>
Introduction I am new to this list, and infact, I am only here by accident. I had attempted to subscribe to the www-amaya list in order to be informed of its port to PC/Win. I stayed because I liked the discussions. I have noticed that the topics don't stay completely on HTML, they tend to wander a little if the discussion warrents it. I am also on another list called dev-access, which deals with accessiblility in general and access to developer envirements in particular. One of the posters has a question he would like posed to those that have a say in how documents on the www are treated by User Agents. If anyone knows of a different forum this could be broached on, please let me know. This is his premise: For the last few years I have been active assisting users of the Lynx World-Wide Web browser. Some of these users are blind and use speech output in lieu of video display screens in their dialog with computers and the Web. Since Lynx is optimized for text-only browsing of the Web, it has developed a following of devoted users among the speech-interface user community. Based on this contact I have developed an interest in access adaptability for the World Wide Web. This is a natural extension of long-term interests that you might summarize as modeling classes of knowledge. Proposal In two companion [messages], I summarize two change notions for the basic operation of the World Wide Web. These are inspired by the difficulties that I observe Lynx users [are] having with the way the Web works today. Perhaps better solutions can be found. But based on my experience to date, these look like they would be a win for the blind Web browsing community, and very lightweight in their cost to the rest of the Web world. One of these dodges I have wanted from long ago and just recently realized that the blind need it more than anyone. That is a "where-it-says" clause in an URL that triggers text-searching in the client that exercises the URL. In other words citations would not be limited to anchors named by the author, but an URL could go precisely to a location defined in the cited document by a third party. This is a way of formalizing in an URL what the folks on lynx-learners use as their standard form of citation. The basic idea is for a citation creator to be able to embed a text string in an URL. The URL with this string in it is intended as a localized reference to the first occurrence of that string in the document which can be retrieved using the remainder of the URL. Searching for that string localizes the reference within the document retrieved. FTP and HTTP servers would ignore this string, but client programs would search for the string and locate their presentation around the location-in-document so found before presenting the document to the user. This requires a standard mechanism because the citation creator needs to be able to communicate the reference to other people in a way that is independent of their choice of client software. One rough idea of a possible implementation is as a search-part parameter name "where-it-says" such that an example URL would be: scheme://path.to.server/path/to/file.typ?where-it-says=I%20told%20you%20so Current usage is that search-part parameter names are not standardized, but defined by servers. In this case that would not be good, as the implementation of the searching is distributed over a varigated lot of client programs. Before this implementation is selected, one would have to examine possible interactions with other uses of this syntax. Rationale Plain-text resources: Many of the currently Web-fetchable resources that are the most use in a speech-output environment are plain text files. People writing for paper write with a prosody closer to oral prosody than those who currently write documents specifically for the World Wide Web. The latter user more the prosody of a highway billboard. However, these documents can get large and speech access is slow. For these reasons, it is particularly valuable to be able to forward a reader to a specific point in the document. Hence the frequent use of "Search with '/' for..." in discussions on lynx-learners. Named anchors -- mostly missing or misplaced: For documents written as HTML, one has the capability to localize a reference by placing named anchors in the document. However, the practice in this regard is very uneven. Particlarly for the speech user, the start-points created by these anchors are mis-located relative to natural starting places for aural browsing. Empowering readers: This may be a subtle point, but permitting third parties to define reference points in Web pages has the effect of flattening the texture of the Web, making it more democratic and a better medium for collaborative problem-solving. The first writer of a page is not the only one that can define focal points. A community of the readers of a page can converge on their own definition of critical points within the document without having to edit what the originating author wrote. In the particular case of speech users, it empowers the speech user community to solve their own peculiar problems by developing a light, adaptation layer over speech-insensitive resources that makes the resources effectively more speech-friendly. To make this serve the speech community as efficiently as the rest of the Web infrastructure supports the background population, it must be possible to institutionalize the speech-adaptive reference points in Web documents. In this line of thought, I am very much influenced by the good work of Gregory Rosmaita in providing a speech-adapted interfaces to search engines by means of a layer of HTML pages. ****** This is the end of his first proposition, I request discussion on the feasibility of implimenting this in the developement of HTML and UA's, and welcome questions for clarification. I will present the second proposition after the first one has run its course. Regards, James -- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ James Alan Shattuck mailto:progman@ecst.csuchico.edu Shattuck Consulting, Chico, Ca. http://www.ecst.csuchico.edu/~progman System Design/Installation Repairs/Upgrades Web Design Business/Personal Training Disabled Access ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Born: 9/3/63, Adopted: 7/71 as Anthony Douglas, San Francisco, Ca Searching for Birth Family Adoption Page: http://www.ecst.csuchico.edu/~progman/adoption/ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Received on Thursday, 17 April 1997 21:55:22 UTC