[whatwg] On accessibility

[Sander Tekelenburg]

At 10:29 +0700 UTC, on 2006-06-15, Alexey Feldgendler wrote:

[...]

> Here is what I think should be standardized: a user agent which supports
>accesskeys MUST provide an uniform method of invoking any accesskey which is
>a letter or a digit. This method should be designed so that the UA's own key
>bindings never conflict with the accesskeys.

FWIW, IMO this is not the terrain of a HTML spec. That aside, it seems to me
that a browser that allows such keyboard-shortcut conflicts to occur is so
obviously broken that I don't even understand why it needs to be discussed at
all, let alone /here/.

*If* you want to go there, the effort should be much broader: you should then
also state that browsers must by default indicate the existence of a TITLE
attribute; that browsers must indicate when their ESP engine has kicked in so
the user knows that what is rendered is at best an educated guess; etc.

It might very well be useful to have such a spec -- something along the lines
of the GNKSA[*]. It might certainly be helpful to users to have a comparison
chart available, maybe with a scoring per browser or even some sort of
certification. But I don't see why such requirements should be spelled out in
a HTML spec.


[*] <http://www.newsreaders.com/gnksa/>


-- 
Sander Tekelenburg, <http://www.euronet.nl/~tekelenb/>