- From: <thatch@us.ibm.com>
- Date: Fri, 30 Jul 1999 20:28:16 -0500
- To: Jon Gunderson <jongund@staff.uiuc.edu>
- cc: w3c-wai-ua@w3.org, ij@w3.org
I apologize for commenting so infrequently. Your comments about browsing prompted this response. I have indicated your writing with JRG, mine with JWT. JRG: Guideline 8: Provide navigation mechanisms JRG: Proposed Sub Head Text JRG: Users need to be able to navigate the web content. Speech and Braille users especially need to have advanced navigation techniques, since speech and dynamic Braille displays offer only a temporal or very limited view of the document at one time. Users need to be able navigate both active and non-active elements using strategies that support the users current familiarity of the document and the task they are trying to accomplish. The following are four basic types of navigation strategies user agents need to support: JWT: I would call them refreshable braille displays. In both cases, speech and braille, the limitation is the serial nature of the access, the inability to "see" the broader picture and to be able to go where the author would really like you to focus. This last statement is reason that there is hope for authors to adopt a "skip to main content" idea. It is less the "size" of the window. Anyway, the issue is not the size of the view. If, when I entered microsoft.com, I could hit one command and be at the main content, that would be great, whether or not I was listening, or had a 20 character braille display. JRG: Sequential: Move sequentially between a set of element types based on the linear order of the element(s). This type of functionality is important for exploring the contents of a new document. The user knows they will view all elements that are part of the sequential navigation. JWT: This is too abstract for me. I think the sequential navigation should require stepping through the document item at a time, with 'item' not defined! JRG: Direct: Move directly between an element or a set of elements based on the element content or numerical position of the element. This type of functionality is important for faster access to web content when they know the location of the information of interest. This often happens when the user is using a frequently visited document. JWT: I don't know understand what this means. And I worry, if all the navigation methods were provided, a user could never find their way. It is easy to say to navigate by many different methods, but it may be very difficult to present all these hypothetical/academic methods to a user. For example if I take this literally I have to have some command structure that moves by element or set of elements. T there are dozens of elements and two to the dozens sets of elements. Do you mean, like HPR does, previous, current and next table, or header? Are you suggesting such for all elements? JRG: Searching: Search for an element based on element content or attributes. All elements or only a sub set of elements maybe part of the search. This function is important to users that are looking for particular key words or other type of web content to efficiently identify if the information is in the document and move to the element containing the information. JWT: Searching is one of the best strategies for navigation. But you make it close to impossible. You require searching content (text, like crtl+F in a browser but include attributes, and arbitrary restriction of search. This is left field. I know of no useful restriction of search. The only attribute I think is useful is alt-text. And it is important to move to forms, which, with hpr, you do with a search for 'form.' JRG: Hierarchical: Based on the document model tree move between the logical relationships between elements. This type of navigation allows the user to efficiently move between logical units of the document. This can be very useful for strongly structured documents like books or instructional materials. JWT: I wish someone would tell me what logical elements of CNN, NYTIMES or IBM are! Again, I think this is left field for a realistic browsing strategy. JWT: My bottom line is that your should encourage browser developers to provide navigation strategies. Let the market determine the best strategies. I have seen a couple of navigation strategies used by screen readers that we will use in the next HPR. This is the way browsers should/will develop. Please do not try to prescribe navigation. No one will listen, and such prescription will lessen what you say in other areas. Jim Thatcher
Received on Friday, 30 July 1999 22:10:01 UTC