- From: Jason White <jasonw@ariel.ucs.unimelb.edu.au>
- Date: Fri, 6 Oct 2000 10:50:39 +1100 (EST)
- To: Web Content Accessibility Guidelines <w3c-wai-gl@w3.org>
Yesterday afternoon at the face to face meeting, there was discussion of content negotiation and the provision of multiple versions of "the same" material. In this message, I shall briefly delineate various circumstances in which this issue arises and point out how they are addressed in the current draft of the guidelines. Are there other cases that ought to be considered? Are the solutions advocated in the current draft sufficient? If not, what proposals would you suggest? Case 1: The provision of "alternative versions" of content for purposes of compatibility with the capabilities of user agents, including assistive technologies. Examples a. The transformation of content into different versions of HTML according to the level of support provided by the client (this practice is actually encouraged by XHTML 1.1 modularization but it could also be applied to earlier versions of HTML). b. Consider two markup languages, A and B. Suppose that the content has been created in A, but that the software which needs to process it (for example a user agent) only supports markup language B. A transformation from A into B is thus required prior to delivery of the content to the processing application. Case 1 (a) above is a special instance of case 1 (b). It should be noted that in either circumstance, the transformation may occur anywhere (that is, on a server, in a proxy server, in a pre-processing application run by the client--E.G. Tablin, or within the user agent itself, for instance in the application of an XSLT style sheet). The guidelines recognize case 1 in the note which follows checkpoint 6.1. Checkpoint 6.4 is also relevant. Case 2: Distributed processing and rendering. The content is stored in a structurally and semantically rich markup language and is transformed (e.g. via style sheets) into a format suitable for rendering, with or without further computation (for example text flow and positioning). Examples a. SVG with presentational attributes, representing the rendered content, ready for display on an output device. b. XSL formatting objects. In case 2, the transformation from the source document to the "presentational" version can occur anywhere (on a server, a proxy server, a pre-processing application, a user agent, within an output device--e.g. hypothetical printer that supports SVG directly). Case 2 is noted in the text accompanying checkpoint 2.3, where the underlying requirement is stated as being that a version of the content must be available in a semantically and structurally rich markup language for processing by user agents so that alternative presentations can be created by software operating under the user's control. Case 3: The document contains a number of "text equivalents" (checkpoint 1.1) and the user agent renders them in place of the (auditory or graphical) material to which they provide alternatives. In this situation, as Ian noted, the user agent can satisfy a "notification requirement"--indicating to the user that a "text equivalent" is being presented in place of an auditory or graphical presentation. Here, it is still possible to think of "multiple versions" of the content, though they are being generated entirely within the user agent through the selective rendering of the incoming data (and of course certain parts of it, such as the audio and image files, may not be retrieved by the user agent at all). Case 3 is covered in checkpoint 1.1. Case 4: The same as case 3 (above), except that the substitution of "equivalents" for the auditory/graphical material takes place upstream, on a server, before it reaches the user agent. Under these circumstances there may or may not be any awareness, on the part of the user, that the substitutions have taken place. Case 5: The author writes different versions of "the same" content, for different audiences (e.g. an exposition which makes extensive use of charts and diagrams, and another which is written as a textual commentary or explanation). In the case of the former, textual equivalents can be provided (as in checkpoint 1.1), but the user may opt for the latter version instead. This is a scenario that was discussed at the meeting. It isn't considered in the guidelines. Query whether one can consider the two versions to be, in the relevant sense, "the same" content. How should the requirement for "accessibility" be defined under these circumstances? As it stands, we would require the author to provide a text equivalent (descriptions and explanations) to accompany the graphical version, and of course there would be no restriction on providing the second version (or other versions aimed at different audiences). Is there an issue here? Of course, combinations involving two or more of the above scenarios can easily arise, for example a transformation intended to offer both backward compatibility with older user agents and upstream inclusion of "text equivalents". Such combinations would appear to fit within the guidelines (with each aspect being controlled by the relevant checkpoint as noted in the cases enumerated above).
Received on Thursday, 5 October 2000 19:50:46 UTC