- From: Al Gilman <asgilman@iamdigex.net>
- Date: Mon, 18 Sep 2000 10:57:30 -0400
- To: w3c-wai-ua@w3.org
At 10:12 PM 2000-09-15 -0400, you wrote: >Hello, > >Checkpoint 9.1 in the 1 September draft [1] reads: > > 9.1 Ensure that when the selection or content focus changes, > it is in a viewport after the change. [Priority 2] > >I propose that this checkpoint be for text or graphical renderings >only. I have trouble making sense of this requirement for audio >viewports. (What does "in" mean in that context?). > AG:: You are correct that the checkpoint as written does not apply to pure-audio interfaces. The proposed remedy misses the mark a bit, because there is a logical requirement that then goes un-covered. The problem lies in the statement of the parent guideline: Guideline 9. Notify the user of content and viewport changes. Alert users, in an output device independent fashion, of changes to content or viewports. The actual requirement is something more like: [Guideline 9] Orient the user after discontinuous changes in the dialog state. Changes to the dialog state include changes to the document and changes in the state of the user's relationship to the document. The latter include transient properties such as selection and focus. [Checkpoint 9.1] Display the selected or focused content. Audio display does not provide a persistent display. Tactile and visual devices do. If a persistent display is available, ensure that the the newly-selected or newly-focused content is actually displayed following a change in selection or focus. If the entire selection or focus content does not fit within the limits of the display region: a) if the region actually displayed prior to the change was within the selection or focus, do not move. b) if the region actually displayed prior to the change was not within the newly selected or focused content, move to display [at least the initial fragment of] such content. If a persistent display is not in use, place a mark at the point immediately before the newly-statused content. This is a mark from which the user may elect to, or the system may by default, depending on the current step or smooth reading mode, start reading. The root problem is in describing the interaction state as "changes in the viewport" where "viewport" is not a device-independent concept. The viewport presumes a display medium which is persistent as viewed by the user agent and random-access as viewed by the user. Audio does not provide such. In a pure audio environment, the whole persistent context is in the mind of the user. In a GUI there is a large shared buffer of dialog information in the display. In audio there is no such sensible patch of interaction that is maintained by the computer and accessed ad lib by the user. The audio display of content requires the elapse of time and time becomes a scarce resource and the flowing of content through the display has to be managed more carefully, which means that in accessing content which was edited at the source for GUI use, it generally has to be managed actively. This need to manage content flow in the user agent is graphically illustrated in the situation with repetitive page header contents. As Kynn Bartlett has explained lucidly in < http://lists.w3.org/Archives/Public/w3c-wai-ig/2000JulSep/0356.html>, the visual reading order of page contents is not the same as the order in the hypertext coding. In the hypertext encoding the content starts at the upper left. In visual reading, the user starts at the lead headline or central layout block. Note the term "discontinuous" changes in the state of the user's connection with the document. In at least one highly effective audio interface, when cruising through the text, the technique is to announce "link" before reading the content of a link. It is not appropriate to shift to "announce that this link has obtained focus" mode at this point because the "link" is a bad enough break in the continuity of the text reading. The author's concept of the link is most clearly perceived by embedding the link text in its context. The conventions of the interface are that link activations go to the most recently read link within some timeout window. Activation readiness goes with reading for the 'link' class of objects in this user interface. This is according to standard pattern, and does not constitute an innovation, a discontinuity or introduction of anything new beyond the fact that "what I am reading next is a hyperlink." Another way to restate Guideline 9 would be to say that the user must be briefed on jumps in context. Anything that invalidates what the user might otherwise be inclined to assume is still valid from their recent experience. Understanding the context in which the system will interpret user actions is particularly critical. And selection and focus are key factors in that context. So tracking selection and focus is required to keep the system response predictable, by allowing the user to have in memory the context which the system perceives for user actions. Al For people with visual disabilities or certain types of learning disabilities, it is important that the [431]point of regard remain as stable as possible. Unexpected changes may cause users to lose track of how many [432]viewports are open, which is the current viewport, etc. User agents should notify the user of content and viewport changes caused by scripts, or allow users to turn off scripts entirely (refer to [433]checkpoint 3.6). Refer to [434]checkpoint 5.5 for requirements about notification of user interface changes through an API. Checkpoints for user interface accessibility: 9.1 Ensure that when the [435]selection or [436]content focus changes, it is in a [437]viewport after the change. [Priority 2] Note: For example, if users navigating links move to a portion of the document outside the viewport, the viewport should scroll to include the new location of the focus. > - Ian > >[1] http://www.w3.org/WAI/UA/WD-UAAG10-20000901/ >-- >Ian Jacobs (jacobs@w3.org) http://www.w3.org/People/Jacobs >Tel: +1 831 457-2842 >Cell: +1 917 450-8783 >
Received on Monday, 18 September 2000 10:39:20 UTC