RE: Access Keys for Hyperlinks on Web Pages for the Visually Impaired [www-html-editorReference: Role?id=7809]

Orion Adrian wrote:
> And this [ACCESS element - JF] still has the problem of specifying
> The UA needs to be
> responsible for assigning keys and the content needs to be
> responsible for specifying roles and the W3C needs to be responsbile
> for creating a standard set of roles that everybody can use.   

Yes and I screamed bloody hell when I read that authors would still be
able to declare specific keybindings.  

I have raised the issue officially, and it seems that the Draft authors
are determined to continue to allow authors to bind keystrokes, although
the justification is weak IMHO

I have also been told, unofficially, that the use case recommendation
would see a "cascade" type mechanism that would go something like:

	1)User defined keymappings over-ride all other settings
	2)User-agent keymappings over-ride author declared bindings
	3)Author declared bindings

We need to ensure that this becomes "codified" into the
spec/recommendation: that it moves from "unofficial" to "official". This
also, however, requires the software tools to deliver to spec.  Maybe
they will. Maybe they won't...  But at least *maybe* the users who would
be genuinely adversely affected by this may find some relief.  

***Therefore, if XHTML2 must contained this flawed functionality, then
clear conflict resolution MUST be part of the

The Draft Authors have stated: "...Author-defined key bindings are a
requirement of many members of our user community...".  Can the Authors
show us one direct communication with the Authors stating this specific
need?  I have searched high and low, I cannot see it.  Adaptive
technology clearly do not need it - they have long since developed their
own binding requirements and mechanisms that completely sidestep author
declared bindings - except of course when there is a conflict.  So who,
pray tell, are these members of the user community?

I simply do not understand why the XHTML draft authors continue to force
a BEHAVIOR unto the semantic logic layer - this type of functionality
should be reserved for the scripting layer.  I cannot find one other
instance where this is deemed "reasonable": where an element has, as an
available attribute, an author declared behavior within XHTML2.  It
flies in the face of reason and the declared goal of truly separating
semantics from behavior and style.  The justification is not good enough
ladies and gentlemen.

I do not and cannot see *WHY* content authors continue to require the
ability to apply specific key-bindings, especially at the semantic
level. The emergence of ACCESS and @role will allow authors to declare
intent (which is good!), leave the rest to the user agents.  The reason
given: "...A good example is a mobile application where links need to be
mapped to numeric keys..." sidesteps entirely the issue of numeric
keybindings previously assigned in non-mobile applications (see: for existing
reasons why numeric keybindings may cause issues).  If this type of
functionality were moved to the style or scripting layer however, it
would be a non-issue. That these layers currently cannot support this
type of functionality (due to DOM issues?) is not a very good reason -
breaking XHTML2 because Styles and Scripting is broken is dumb.

So, Draft Authors, what gives?  Will XHTML2 emerge pure, or continue as
a hybrid mongrel mixing semantics with behaviors?  And if the answer is
"B - Mongrel", then please, I repeat again, clear conflict resolution
MUST be part of the specification.

John Foliot
Web Accessibility Specialist / Co-founder of
Web Accessibility Testing and Services   
Phone: 1-613-482-7053  

Received on Monday, 14 November 2005 02:53:22 UTC