Summary of API requirements suggested by AT developers

This document:: http://www.w3.org/WAI/UA/2002/04/api-reqs
Editor:: Ian Jacobs, W3C

Status of this Document

This document has been produced for the User Agent Guidelines Working Group (UAWG) to summarize responses from assistive technology developers to a request made a the 28 Mar 2002 teleconference. This summary is likely to change at any time, and represents no consensus. This summary is expected to help the UAWG refine the requirement in UAAG 1.0 pertaining to communication between a conforming user agent and other software such as assistive technologies.

1. Content/ui/event/api requirements

The following table organizes content requirements expressed by AT developers. The table indicates by @@ where these requirements are not met by the W3C XML Infoset Recommendation (see email on establishing minimal set of API requirements in terms of the Infoset, for both XML and HTML).

Note: A linear version of this table is also available.

	Jill Thomas, ION	Jost Eckhardt, AISquared	Cathy Laws, IBM
Content	All text All attribute info items All parent element info items	All text All enabled elements Attribute info items for some elements (e.g., frames) (Navigation) ordered list of links, frames, form controls Functional blocks of content (in UAAG 1.0 terms, "important elements") Layout v. Data table	Retrieve and change all element content Retrieve and change all attribute values
Rendering structure	None	Coordinates of rendered elements Coordinates of rendered words Coordinates of rendered characters Mapping from screen coordinates to source element	For each rendered character, pixel location Mapping from character pixel location to DOM text node and character offset
Style	None	Text foreground and background color, font family name, text size, font characteristics	Text display attributes
UI/Session variables	None	Content focus (fast access) State of form controls	Selection (including selected non-text content). Get viewports for embedded objets or applets. Note: This is generally done through the operating system, not the API used to access content.
Events/Change notification	None	Content focus Selection Form controls: radio buttons, check boxes, choices, text area Download complete	Receive event notifications for mouse events, key events Changes to elements/attributes
Programmatic control	None	Move rendered content into viewport Move focus to enabled elements Activate enabled elements	For both rendered elements and characters, simulate an event even if not for an enabled element (left or right mouse click) Move focus to viewports with embedded objects or applets. Select content (text and images). Set focus
API requirements	Preserve document order Provide alert when document object construction is complete.	Preserve document order	Content information and events must represent rendered content (after error correction).

2. Out of scope for UAAG 1.0

From Jost:

Meaningful description of an image. This is an authoring issue (and is recognized by WCAG 1.0). If provided, then presumably this information would be available through any API we would require.

From Catherine:

Be able to determine an element's structural "containing element" (e.g., form, table, frame, page, etc.). This semantic information is determined by specification, not by the API.

Last modified: $Date: 2002/05/10 19:54:35 $ by $Author: ijacobs $.