RE: 2.1 thoughts

Though I haven't arrived at a satisfactory solution, I do have some
ideas to share.

1. The requirement under checkpoint 2.1 should not forbid the use of
   certain kinds of device-specific event handling. Rather, it should
   insist that the web content ought not to rely on them in order to
   be operable.

2. Text input is, I think, a reasonable expectation. It isn't possible
   to design, for example, a large data base with a search facility
   that doesn't rely on text entry. We could however add a suggestion
   that alternatives to text entry be provided; actually I think this
   is already mentioned under the "handling of input errors"
   checkpoint, or at least used to be.

3. The problem is: how to define the kinds of input processing that we
   want to allow, without confining ourselves to
   implementation-specific scenarios which have undesired
   implications, or unduly restricting the range of implementation
   choices. The fundamental idea, I think, is that preference should
   be given to methods of input handling which can be characterized as
   logical actions within an application context, rather than as
   events which reflect the physical state of the input device. A
   clear example of the latter is an event which takes the coordinates
   of a pointing device as parameter.

4. In general terms, then, the success criteria are as follows:

a. Where the implementation technology provides logically defined
events, ensure that the web application can be operated entirely using
these events.

b. Where the implementation technology provides only device-dependent
events (that is, events which are described in terms of the state of
specific input devices) provide parallel, redundant implementations of
the input mechanism so that it can be operated entirely from each type
of input device for which events are defined. Exception: support to
enable pointing devices to perform character input is typically
provided by assistive technologies and need not be supplied by the
application itself.

This is a very rough initial proposal. Improvements are welcome, as is
a draft of what would become the actual checkpoint. The exception at
the end of the previous paragraph is somewhat anomalous and could be
refined further.

Received on Monday, 6 May 2002 21:10:02 UTC