W3C home > Mailing lists > Public > www-html@w3.org > June 2003

Re: Accesskey in XHTML 2

From: Tim Bagot <tsb-w3-html-000B@earth.li>
Date: Wed, 18 Jun 2003 19:06:36 +0000 (UTC)
To: <www-html@w3.org>
Cc: Arve Bersvendsen <arve@virtuelvis.com>
Message-ID: <Pine.LNX.4.33.0306181819480.1021-100000@213-152-52-166.dsl.eclipse.net.uk>

At 2003-06-18T08:30+0200, Arve Bersvendsen wrote:-

> Current accesskey implementations don't seem to be very well thought out:
> In for instance Mozilla and MSIE on Win, the accesskeys are invoked by
> pressing Alt+[accesskey], which is the same normally mechanism used for
> accessing the main menus of said programs.

The first line hits the nail on the head - this is, to be blunt, a stupid,
fundamentally flawed implementation. It is clearly impossible for authors
to avoid all keyboard shortcuts used by all browsers in all locales. It is
up to each browser's authors to employ a (preferably configurable)
modifier/prefix combination that will not conflict with its own shortcuts.
This is one good reason why the accesskey attribute does not attempt to
specify the exact sequence of keystrokes.

> Which brings up the question: perhaps the accesskey mechanism in XHTML 2
> should be replaced with an accessibility mechanism where the accesskey is
> bound to a function, rather than mapped to a specific key?

Not really. Mostly it suggests that certain browsers need to be improved.

> Something along the lines of <input type="text" access="username" />.
>
> I know this would take away some freedom from authors in creating accesskey
> schemes, since there would have to be a definitive list of accesskeytypes,

Indeed - anything not in the list could no longer be given an accesskey. I
do not think this would be acceptable. Worse, the list would necessarily
be quite short, if we assume the goal would be mostly reasonably good
mnemonics in most languages.

> but this limitation would also enable users to learn the application once
> instead of learning and relearning accesskey schemes for each and every
> site he/she visits.

Perhaps a parallel mechanism would be of some use. It is already possible
for browsers to provide shortcuts (separate from accesskeys) for link (and
even a) elements based on well-known link types. Perhaps a handful more
could be added, and maybe the mechanism extended to form fields (probably
with a different attribute name), but only the most basic could be handled
in this way, and it is likely that authors would naturally tend (in so far
as they bother with accessibility at all) to choose roughly the same keys
(in any given language).

Specialised accesskeys seem to me to be perfectly sensible - almost every
piece of interactive software does the same. The key is to indicate them
both clearly and non-intrusively; this can be difficult, but I feel it is
mostly the realm of styling and design.


Tim
Received on Wednesday, 18 June 2003 15:36:03 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 27 March 2012 18:15:55 GMT