Re: TR Checkpoint 5.4 - User Interfaces

First, contratulations and thanks are due to Lee for his excellent
work in analysing and writing proposals for the comments on
checkpoints 5.3 and 5.4. This is very much what we need at present,
and will facilitate the working group's progress significantly.

Turning to the substance of the comments, I worry about the scope of
this checkpoint - under what circumstances exactly are we proposing to
invoke the jurisdiction of the User Agent Accessibility Guidelines?
Consider the following scenario, which already exists in practice (cf.
Mozilla and possibly other comparable technologies). An interactive
application is written in a combination of Ecmascript, XML and CSS; it
has a URI, and when it is interpreted by the user agent a user interface, as
interactive as you please, is constructed. The Ecmascript code calls
functions of the user agent (and possibly the operating system);
the XML describes the user interface in more or less device
independent terms according to its design, and the CSS component
regulates the style of the interface. Here we have an application that
is, under all reasonable definitions, Web content. In fact the user
interface of a user agent can be constructed in this manner, thereby
complicating the distinction between "content" on the one side, and
"user agent" on the other. The application could equally well be an
authoring tool, for that matter.

Under this scenario, the question is whether checkpoint 5.4 should
apply at all, and if so how it interacts with the remainder of the
guidelines. Surely an application of the kind described above should
be subject to WCAG 2.0. What additional requirements, then, should be
imported from UAAG 1.0 and how do the two sets of guidelines interact
in such a case? One could propose that only non-standard user
interface code used by the content (code which bypasses the
accessibility support of the user agent and/or operating system)
should be subject to checkpoint 5.4, but that kind of incompatibility
with alternate user interfaces and assistive technologies is part of
what checkpoint 5.3 is supposed to exclude, implying that checkpoint
5.4 is redundant under this interpretation. I think we need to go back
to the beginning and, possibly in cooperation with the User Agent
Accessibility Guidelines working group, examine relevant scenarios and
decide how the two sets of guidelines should interact in such
circumstances; and the same holds with respect to authoring tools
(obviously if the "content" is also an authoring tool, the developer
should apply ATAG).

Received on Monday, 27 January 2003 19:44:54 UTC