W3C home > Mailing lists > Public > w3c-wai-ig@w3.org > April to June 2003

Re: Limited Number of possible accesskey values

From: Patrick Burke <burke@ucla.edu>
Date: Mon, 14 Apr 2003 18:47:18 -0700
Message-Id: <5.1.0.14.2.20030414174821.00c08470@collie.ats.ucla.edu>
To: "Hoffman, Geoffrey" <ghoffman@aztrib.com>, w3c-wai-ig@w3.org

At 04:55 PM 4/14/2003, Hoffman, Geoffrey wrote:

>I've searched the net far and wide, and the W3 site as well, and can't seem
>to find a complete reference for the available accesskey attribute values.
>
>Now that web developers have AOL, IE4, IE5, IE5.5, NS4x, NS6x, NS7x, Opera,
>Safari, and even more browsers, it seems there needs to be a list of the
>available accesskey values that remain, after subtracting all of the
>built-in keyboard commands already taken by the various browsers.
There are some old threads on this. I think the result was that the number 
of unused keys approaches zero as the number of browsers under 
consideration increases. ...


>The problem is best stated in this way:
>
>The total maximum number of possible accesskey attribute values (47?) is, in
>practice, a smaller subset of
>{abcdefghijklmnopqrstuvwxyz0123456789-=`[]\;',./} based on the following:
>
>- Internet Explorer has reserved:
>         - Alt-f : File Menu
>         - Alt-e : Edit Menu
>         - Alt-v : View Menu
>         - Alt-a : Favorites Menu
>         - Alt-t : Tools Menu
>         - Alt-h : Help Menu
>         - Alt-d : Address Field
>[others snipped]
Note that these don't all work the same: Most of them jump to selections 
from the menu bar. This means that even if you define an AccessKey for one 
of these, you can still get to the menu item by, for example, pressing ALT 
and then F for the File Menu. ALT-D is different, though: hitting ALT & D 
in sequence does nothing interesting (in fact it just exits the menu bar). 
The only way to use the Address Bar command is to hit the keys 
simultaneously. So defining an Access Key for ALT-D blocks the user from 
getting to the Address Bar via keyboard.


>This also brings up the question, how do you encode a page properly to
>support WAI standard(s), which has more than 35 links on a page? In
>otherwords, are accessible pages limited to 35 links per page?
[Sorry for bad editing, those would be:

"{cijklopqrsuxyz0123456789`-=[]\;',./}

35 available access key attribute values."


That seems like a very extreme conclusion.  On the one hand, I think access 
keys are much more useful for sections of a page, important form fields, 
etc., rather than for each individual link. In practical terms, a page with 
35 important sections should probably be broken up into smaller pages.

Another practical matter is the memorization needed to use the access keys, 
which limits you to much fewer than 35 for most people. (Most pages that 
use them wisely have 5 or 6. In fact, as a blind user, there isn't a site 
that I use frequently enough to memorize the access keys, so I have to look 
them up, which slows down the navigation process. More trouble than it's 
worth, in other words.)

It also seems very difficult to come up with a scenario where Alt-` or 
ALT-' would have a mnemonic connection to an item on a page, making these 
and other punctuation keys much less desirable for access keys.

If I don't do it, I'm sure others would point out that Access Keys cross 
over into user navigation tools that properly should be handled by the user 
agent. I guess I can't argue with that, but it can still be useful to have 
a few Access Keys if the page or site structure is very complex. If you use 
other structural navigation approaches well (proper heading markup, 
same-page links , clear & consistent layout, among others), then users 
won't be forced to rely on Access Keys.

So, to sum up, Access Keys can be good if used sparingly (far fewer than 
the maximum possible). However, they are neither necessary (other good 
techniques exist) nor sufficient (always likely to be a key conflict 
somewhere) to guarantee accessibility on their own.

Just a few thoughts. Thanks for raising the interesting topic,

Patrick
Received on Monday, 14 April 2003 21:47:24 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 19 July 2011 18:14:09 GMT